Introduction To Particle Physics

Introduction to Particle Physics
Lecture in WS 2013/14 at the FSU Jena
Axel Maas
2
Contents
1 Particle Physics 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Natural units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Elementary particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Fermions and bosons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Particles and anti-particles . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.6 Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Scattering experiments 8
2.1 Non-relativistic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.1 Elastic scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.2 Cross section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.3 Luminosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Relativistic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Repetition: Relativistic notation . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Inelastic scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Cross-sections and particles . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5 Feynman diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3 A role model: Quantum electrodynamics 19

3.1 Electrons and photons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.2 Formulation as a field theory . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3 Indistinguishable particles . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.4 Gauge symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.5 Bound states and the positronium . . . . . . . . . . . . . . . . . . . . . . . 25
3.6 Muons and Taus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.7 Radiative corrections and renormalization . . . . . . . . . . . . . . . . . . 26
i
ii Contents
3.8 Fixing the inputs and running quantities . . . . . . . . . . . . . . . . . . . 27

3.9 Landau poles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4 Invariances and quantum numbers 31

4.1 Symmetries, groups, and algebras . . . . . . . . . . . . . . . . . . . . . . . 31
4.2 Noethers theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.3 Electric charge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.4 Implications of space-time symmetries . . . . . . . . . . . . . . . . . . . . 36
4.5 Parity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.6 Time reversal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.7 Charge parity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.8 CPT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.9 Symmetry breaking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.10 Fermion number . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.11 Flavor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.12 Chiral symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.13 Generators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5 Strong interactions 44
5.1 Nuclei and the nuclear force . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.2 Mesons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.3 Nucleons, isospin, and baryons . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.4 The quark model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.5 Color and gluons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.6 Chiral symmetry breaking . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.7 The Goldstone theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.8 Confinement and asymptotic freedom . . . . . . . . . . . . . . . . . . . . . 57
5.9 Hadronic resonances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.10 Glueballs, hybrids, tetraquarks, and pentaquarks . . . . . . . . . . . . . . . 59
5.11 Flavor symmetry and strangeness . . . . . . . . . . . . . . . . . . . . . . . 61
5.12 Charm, bottom and quarkonia . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.13 Top . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.14 Partons and scattering experiments . . . . . . . . . . . . . . . . . . . . . . 67
5.14.1 Fragmentation functions and jets . . . . . . . . . . . . . . . . . . . 67
5.14.2 Parton distribution functions . . . . . . . . . . . . . . . . . . . . . 68
5.14.3 Monte-Carlo generators . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.15 Some algebra and group theory . . . . . . . . . . . . . . . . . . . . . . . . 71
Contents iii
5.16 Yang-Mills theory and QCD . . . . . . . . . . . . . . . . . . . . . . . . . . 74

5.17 Cosmic radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.18 The QCD phase diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6 Weak interactions 83
6.1 decay and parity violation . . . . . . . . . . . . . . . . . . . . . . . . . . 84
6.2 Neutrinos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6.3 Flavor-changing currents . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.4 W and Z bosons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.5 The Higgs effect and the Higgs boson . . . . . . . . . . . . . . . . . . . . . 91
6.6 Parity violation and fermion masses . . . . . . . . . . . . . . . . . . . . . . 95
6.7 Weak isospin, hypercharge, and electroweak unification . . . . . . . . . . . 96
6.7.1 Constructing the gauge group . . . . . . . . . . . . . . . . . . . . . 97
6.7.2 Hiding the electroweak symmetry . . . . . . . . . . . . . . . . . . . 99
6.8 CP violation and the CKM matrix . . . . . . . . . . . . . . . . . . . . . . 104
6.9 Neutrino oscillations and the PMNS matrix . . . . . . . . . . . . . . . . . 106
6.10 The weak interactions as a gauge theory . . . . . . . . . . . . . . . . . . . 109
6.11 Anomalies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
6.12 Landau-poles, triviality, and naturalness . . . . . . . . . . . . . . . . . . . 115
6.13 Baryon-number violation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
6.14 The early universe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6.14.1 Global symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
6.14.2 Abelian case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
6.14.3 The electroweak case . . . . . . . . . . . . . . . . . . . . . . . . . . 125
7 Beyond the standard model 128

7.1 Inconsistencies of the standard model . . . . . . . . . . . . . . . . . . . . . 129
7.2 Gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
7.2.1 Problems with quantization . . . . . . . . . . . . . . . . . . . . . . 131
7.2.2 Asymptotic safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
7.3 Observations from particle physics experiments . . . . . . . . . . . . . . . 133
7.4 Astronomical observations . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
7.4.1 Direct observations . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
7.4.2 Dark matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
7.4.3 Inflation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
7.4.4 Curvature, cosmic expansion, and dark energy . . . . . . . . . . . . 136
7.4.5 Matter-antimatter asymmetry . . . . . . . . . . . . . . . . . . . . . 137
Chapter 1
Particle Physics
1.1 Introduction
Particle physics is the science of the smallest constituents of matter, and how they interact.
In a sense, it evolved out of physical chemistry, where it was first resolved that all chemical
substances, like molecules, are made out of set of chemical elements, of which we know
currently roughly 120. At the turn of the 19th to the 20th century, it was then found that
the atoms themselves were not elementary, but rather had constituents - a single nuclei,
which was characteristic for the element, and a number of indistinguishable electrons. The
latter number was in turn uniquely fixed by the nucleus.
In the early 20th century it was then found that the nuclei themselves are not elemen-
tary, but were made up out of just two types of constituents, the protons and neutrons,
commonly denoted as nucleons. Again, and as will be described later in detail, these nu-
cleons are not elementary, but are made up out of the quarks. At the current time, we
do not know, whether either quarks or electrons do have a substructure, but substantial
effort is invested to find out.
As is visible from this short historical remark, the study of elementary particles has
changed subject many times over the course of time. Today, elementary particle physics is
considered to be the study of the most elementary particles known to date, their interac-
tions, and whether there could be even more elementary constituents. Today, such studies
are inseparable inked to quantum mechanics, as quantum effects dominate the world of
the elementary particles.
It is this field which will be presented in this lecture. After some more general remarks,
the three known fundamental interactions, as well as the known particles will be presented.
Their theoretical description together constitutes the standard model of particle physics.
This theory is well established, and has been experimentally verified in the accessible range
1
2 1.1. Introduction
of parameters. But there are several reasons which prove that it can not be the final theory
of particle physics. These will be briefly summarized at the end of the lecture. Thus, this
lecture can only be taken as a snapshot of our current understanding of particle physics.
Suspiciously absent from these forces is gravity. The reason for this is twofold. On the
one hand, so far nobody has found a reliable way how to quantize gravity, though this
is a very active area of research. This quantum gravity is usually also considered to be
part of particle physics. However, because of the complications involved in formulating it
theoretically, there is not (yet) a standard model of quantum gravity, and just to describe
the more promising candidates in a fair way is a lecture of its own, and rather specula-
tive. The other is that also experimentally no particle has been yet observed which can
be considered as an elementary particle of gravitation, the hypothesized graviton. Thus,
quantum gravity is not yet part of the standard model of particle physics. This is proba-
bly the most stringent evidence for the incompleteness of the standard model of particle
physics.
As the standard model of particle physics in its current form has been theoretically
established at the turn of the 60ties and 70ties of the 20th century, with the final major
theoretical touches added in the early 80ies, there are many excellent textbooks on its
concepts. Especially, there are many theoretical textbooks on its foundations. The present
lecture, however, is a more phenomenological introduction. For the concepts, this lecture is
based on a number of older textbooks, but also some books on quantum-field theory have
been used. However, there are numerous textbooks on the topic, with very different styles.
Giving therefore a reasonable recommendation, especially in a field evolving as quickly as
particle physics, is of little value. At the same time, identifying a suitable textbook to
learn new concepts is a very helpful exercise for latter literature studies. Therefore, no
general recommendations for this lecture will be given.
However, the latest experimental results for the standard model are from the year 2012,
and two quantities still remain to be determined, as will be discussed below. Furthermore,
the formulation of a theory is one problem. Determining all its consequences theoretically
is a much harder problem, which is in many respects still unsolved today. We are far
from the point where we can compute any given quantity in the standard model to any
given accuracy. Furthermore, experimental results are never exact, and have errors. So, in
many cases we know for deviations from the standard model only upper limits, and cannot
exclude something is different. Thus, our knowledge of the standard model is in continuous
motion. The best repository of the current knowledge is from the particle data group,
which is accessible for free at pdg.lbl.gov. The most important discoveries (and information
for this lecture) can also be obtained from my twitter feed, twitter.com/axelmaas.
Chapter 1. Particle Physics 3
Note that a lecture on particle physics without full-fledged quantum-field theory can
necessarily often only give hints to the origin of many phenomena. The main aim is to
make you acquainted with the general structure, with concepts, and with ideas. A full
understanding of the details is necessarily relegated to quantum-field theory, and, in part,
advanced quantum field theory courses. Wherever possible, I will try to give as much
insights from these, without going into the technical details, skipping calculations for
hand-waving arguments. As a result, with the knowledge of this course, you will be able
to follow an experimental or theoretical overview talk at a mixed conference, but it will
not be sufficient to follow a specialist talk in a parallel session, or likely an overview talk
at a purely theoretical conference. Nonetheless, even in the latter cases, many of the ideas
and underlying physics should sound then at least familiar to you. Eventually, the aim of
this lecture is to prepare you for a full quantum-field theory course on particle physics.
1.2 Natural units

The usual units of the SI system, like meters or grams, are particularly unsuited for particle
physics. The elementary particles live on very short time scales, move at high speeds, and
tiny masses. E. g. typically energy and distance scales are
1 eV = 1.60 1019 J (1.1)

1 fm = 1015 fm, (1.2)
which therefore have earned their own name, electron volts (eV), and Fermi (being identical
to a femtometer). Typical energy scales are actually rather 109 eV, a GeV.
Furthermore, in particle physics the system of unit is further modified. Since units
are man-made, they are not necessary, and the only thing really useful is a single scale to
characterize everything. This is achieved in the system of natural units, which is obtained
by setting
c = ~ = kB = 1
leaving only one unit. This is usually selected to be GeV, and thus everything is mea-
sured in GeV. E. g., times and lengths are then both measured in GeV1 . In certain
circumstances, GeV are exchanged for fm, which an be easily converted with the rule
~c = 1 0.1973 GeV fm
and thus energies are then measured in inverse fm.

These units are the standard units of particle physics. No relevant publication, text-
book or scientist uses any different units, except for occasionally using a different order,
4 1.3. Elementary particles
like MeV, being 103 GeV, or TeV, being 103 GeV. Also other dimensionful quantities are
then expressed in these units. E. g. Newtons constant is roughly 1.221019 GeV.
1.3 Elementary particles

From the point of view of elementary particle physics an elementary particle is any particle
which, to the best of our knowledge, has no internal structure and is point-like. In quantum
mechanics the Compton wave-length is often used to characterize the size of a quantum-
mechanical object. Especially, this implies that an elementary particle is much smaller
than its Compton wave-length. E. g., for an electron it is roughly of the order of its inverse
rest mass, i. e. (511 keV)1 400 fm. The experimental upper bounds for its size is 107
fm. Today, we know of 38 particles, which are elementary in this sense, and these will be
introduced during this lecture.
On the other hand, there are composite particles, i. e. particles having an inner struc-
ture, which consists out of two or more of the elementary particles. Nucleons are an
example for these. Of course, composite objects can consists out of composite objects,
like nuclei, atoms, or molecules. Such composite objects are called bound states. In the
course of history, many instances are known that particles previously recognized to be
elementary have actually been composite. Today, nobody would be too surprised if any
of the elementary particles known would, in fact, be composite. This has happened too
often.
1.4 Fermions and bosons

Since there exist elementary particles which are massless, like the photon, they move nec-
essarily with the speed of light. Thus, any adequate description of particle physics requires
the use of special relativity, and only under very special circumstances is it possible to
obtain reasonably accurate approximate results by using only non-relativistic physics. Be-
sides this necessity, the union of quantum mechanics and special relativity, called quantum
field theory, implies many remarkable and very general results.
One such result is the fundamental distinction of particle types into bosons and fermions.
Bosons obey Bose-Einstein statistics; this implies that there can be arbitrary many of them
in any given state of fixed quantum numbers. On the contrary, fermions obey Fermi-Dirac
statistics and have to adhere to the Pauli principle: In any given state with fixed quantum
numbers can only be a single fermion.
The spin-statistics theorem, which is a very basic property of all experimentally verified
field theories, and of most hypothesized ones, implies a deep relation between the statistical
properties of particles, the Lorentz group, and the spin of a particle. It essentially states
that for any Lorentz-symmetric, i. e. special relativistic, quantum theory in more than
two (one space and one time) dimensions every particle obeying Bose-Einstein statistics
has integer spin, and any particle obeying Fermi-Dirac statistics has half-integer spin.
There are no further possibilities for the spin values. Thus, bosons have integer spin, and
fermions have half-integer spin. The electron, e. g., is thus a fermion, with its spin of 1/2.
The photon, with its spin of 1 is therefore a boson.
Note that spin is an intrinsic property, like rest mass, of an elementary particle. It is
not something which can be constructed from any properties, and has to be determined
in experiment, at least inside the standard model. For a bound state, however, the spin
can be computed as a function of the spins of its constituents.
A further important consequence is that the wave-function of a system with n particles
has to be antisymmetric under the exchange of two fermions, while it is symmetric under
the exchange of two bosons. This already illustrates the Pauli-principle. If there would be
two particles of the same quantum numbers (and thus of the same type, since type is a
quantum number) at the the same place, their wave-function must change sign under an
exchange. At the same time, they are identical, and thus numerically it cannot change.
Since the only quantity which at the same time changes and does not change sign is zero,
any such state has to be zero when the particles are at the same place, and thus they
cannot be - this is the Pauli principle. On the contrary, for bosons the wave-function
remains unchanged under an exchange of two identical bosons.
1.5 Particles and anti-particles

Another fundamental consequence of quantum field theory is the existence of anti-particles,
first noted, and experimentally verified, in the 20ies and 30ies of the 20th century. The
basic statement is that for any particle there exists an anti-particle of the same mass and
spin, but otherwise opposite (e. g. electric) charges. For the electron, there is a fermion
with also spin 1/2, positive electric charge, and the same mass: The positron. Only if
a particle has no properties except for mass and spin (like the photon) there is no anti-
particle, as the anti-particle would be again the particle itself.
As a further consequence, if a particle and its corresponding anti-particle meet, they
can annihilate each other. If, e. g., an electron and a positron meet, they can annihilate
each other, which will result in two or more photons. This does not need to happen
instantaneously. Electron and positron can also first form a bound state, positronium,
6 1.6. Interactions
which has a finite life-time. It decays because of the annihilation of the two particles,
decaying into two or three photons, depending on how the spin of the electron and anti-
electron had been aligned with respect to each other.
Since this is a very fundamental prediction of our understanding of particle physics,
great experimental effort is invested to check whether particles and anti-particles really
have the same properties, e. g. the same mass. Particular attention has been devoted to
produce anti-atoms, and measure their spectrum, as this is a very sensitive test of this
theoretical prediction. So far, no deviation has been found. Still, although the production
of anti-matter is tedious, and yields are measured in individual atoms, rather than grams,
these tests remain the most sensitive ones of the foundations of particle physics.
1.6 Interactions
The annihilation of electrons and positrons into photons is an example for an interaction,
in this case an electromagnetic interaction. Writing on a black board or not falling through
the floor are also examples of electromagnetic interactions. The earth orbiting the sun is
an instance of a gravitational interaction. Generically, anything, except the Pauli prin-
ciple, what makes a particle aware of the presence of another particle is classified as an
interaction. This is called a dynamical effect. The propagation of particles, on the other
hand, is just a kinematical effect.
Besides the knowledge of all the elementary particles, it is also necessary to know
the interaction of them with each other to write down a theory describing them. Such
an interaction is not necessarily restricted to connect only two particles. The maximum
number known so far is a four-particle interaction, and any interactions involving more
particles than four can be broken down to those with less particles. This does not mean
that it is not possible to have more; theoretical, this is possible, but there is no experimental
evidence that it is needed so far. Also, there can be interactions which appear at a coarse
scale to be an interaction involving more than four particles, but on a fine scale this is not
the case.
There are two particular properties of interactions, which deserve a special mentioning.
One is that interactions are not totally arbitrary. Rather they only occur between specific
particles. E. g. the aforementioned annihilations only proceeds with one electron, one
positron, and two or more photons. It will not occur with two electrons and two photons,
or one electron, a proton, and two photons. The reason are conservation laws, in this
case the one of electromagnetic charge. Throughout many more conservation laws will be
encountered, and any interaction will strictly respect any exactly conserved quantity.
However, mass is not a conserved quantity. As a consequence of the famous Einstein

equation E = m (note the natural units!), energy and mass can be freely converted into
each other. Thus in the positron-electron-annihilation process, the total rest mass of both
particles, roughly 1 MeV, is converted into the kinetic energy of the photons, which are
themselves massless. On the other hand, colliding two particles with a large amount of
kinetic energy will permit to create new particles, which have at most a total mass equal
to the energy (and rest mass) of the original particles. This is of central importance to
scattering experiments to be discussed next.
Before moving on to them, there is one more noteworthy kind of interaction. Quantum
mechanics forbids to measure energy and time at the same time arbitrarily precisely1 ,
Et 1. Thus, it is possible for a brief amount of time to have more energy in the
system. For this duration, it is possible to create also particles with this additional energy,
though they have to be destroyed at the end of this period. Since these live only for this
short amount of time, they are not real, but are called virtual particles. Such particles are
often denoted by an asterisk, *.
Such virtual pairs can either appear out of the vacuum, or can be emitted and reab-
sorbed by a moving particle. In both cases there have been many experimental confir-
mations of these processes, e. g. the Casimir effect for the first and the Lamb shift for
the latter. Such virtual particle fluctuations are today used in precision experiments to
access particles with masses larger than those which can be directly created using available
energies. This will also be discussed later.
1
Actually, this is not an elementary relation like the Heisenberg uncertainty principle, but follows from
identifying particles with wave packets.
Chapter 2
Scattering experiments
The primary tool to understand particle physics today experimentally are scattering ex-
periments, i. e. letting particles collide with each other. The reason has been allude to
before: When two particles collide they interact, and their kinetic energy can be used to
create new particles. The higher the energy, the more massive particles can be created.
Furthermore, because of the DeBroglie relation that wave-length and momentum are re-
lated by = 1/p, only high energies permit to resolve very small structures, and therefore
permit to investigate whether a given particle has a substructure.
Of course, due to the presence of virtual particles, it would in principle be possible
to also investigate everything using low-energy collisions or even measurements of static
properties of a particle. However, as will be seen later, the required precision to resolve new
particles in most cases drops with increasing energy. Therefore, to access new particles in
low-energy collisions requires very precise experiments. For some cases, particular effects
reverse the situation. As a consequence, today both low-energy precision experiments and
high-energy collisions work hand in hand together, both having their own importance.
The formulation of scattering experiments is thus central to particle physics. Almost
all discoveries in particle physics have been made with such experiments. As a consequence
also most modern theoretical tools have been developed and optimized to describe scat-
tering experiments. Thus, at least an elementary understanding of scattering processes is
a necessity in particle physics.
2.1 Non-relativistic
Many of the necessary concepts can already be understood non-relativistically, which offers
a simpler formalism.
8
Chapter 2. Scattering experiments 9
2.1.1 Elastic scattering

The simplest case is that of two incoming particles colliding with each other, and leaving as
the outgoing particles unchanged. This case is known as elastic scattering. A very suitable
coordinate system to describe this is the center-of-mass coordinate system, in which the
total momentum of the two incoming particles is zero, ~p1 = ~p2 with equal length p. Other
coordinate systems, like the laboratory frame, with non-zero center-of-mass momentum
can be obtained using a Galileo transformation (or later a Lorentz transformation), and
therefore will not be discussed here further. Furthermore, if particles have a spin, things
are also more complicated, but this will be avoided for now. Especially, in the following
always bosons will be assumed, and the Pauli principle will not play a role.
The initial state is thus totally described by the type of particles and their properties,
say two particles with masses m1 and m2 , and their two momenta. Non-relativistically,
energy and momentum conservation hold separately, and the initial state is completely
characterized by the quantities
p21 p2
E = + 2
2m1 2m2
0 = ~p1 + p~2 .
Important here is the absence of any potential energy. In scattering experiments, it is

assumed than in the initial and final state the particles are so far separated that they
do not influence each other. This is called asymptotic states. For electromagnetic and
gravitational interactions, which decay as 1/r, this approximation is certainly reasonable.
It is also found to hold true for all particles and interactions, which can so far be prepared
as initial states, or detected as final sates.
The final, or outgoing, state, has to have the same total energy, and the same masses.
Thus, the total final momenta ~qi has the same size, qi = pi , but may have a different
direction. However, there is only one undetermined quantity, since the scattering is copla-
nar: There is a possible scattering angle, which can be defined between any incoming and
outgoing momenta, e. g. p1 q1 cos = p~1 ~q1 . There are many conventions in use.
This scattering angle depends on the interaction with which the two particles scatter.
Measuring it and comparing it to the prediction is one of the possibilities on how a theory
can be falsified.
2.1.2 Cross section

In a quantum theory, a scattering process will be statistically distributed, and therefore
the measurement of an individual scattering angle is meaningless. Thus, experiments are
10 2.1. Non-relativistic
repeated many (many billion) times to obtain a distribution of scattering angles. The
number of particles scattered into a solid angle d = sin dd is then defining this cross
section as
dN
(, ) = .
d
In practice, there are two quantities, which can be derived from the cross section, which
are of particular importance. One is the total cross section
Z
t = d,
the other are differential cross sections
(, , x)
x = ,
x
where x can be any kind of variable, including also and themselves, on which the
cross-section can depend. Also higher order differentials could be useful, so-called multiple-
differential cross sections.
The unit of a cross section is an area. A useful unit to measure it in particle physics
is barn, abbreviated b, corresponding to 100 fm2 . Although nowadays usual interesting
cross-sections are of the order of nb to f b.
Predicting such cross sections is one of the most important tasks when one wants to
test a given theory against modern particle physics experiments. However, this is quite
an indirect process. The cross-section is calculated as a function of the parameters of a
theory, which includes also the type and number of elementary particles. This result is
then compared to the experimental result, an by comparison of how the cross-section is
then depending on external parameters (e. g. the angle or the type of particles involved),
the theory is either supported or falsified.
This becomes complicated in practice, if the cross section has contribution not only
from the theory in question, but especially also from other known origins. E. g., when
searching for yet unobserved physics, one has to subtract any contribution of known physics
from the measured cross-section. Especially if the known contribution is the dominant
one, this is difficult, as it is practically impossible to calculate it to arbitrary precision.
It is also not possible to derive it from experiment; there is no possibility to continue
unambiguously an experimental result from one case to another, without performing a
theoretical calculation. Thus, this background reduction of known processes has become
one of the major challenges in contemporary particle physics.
2.1.3 Luminosity
The definition of a cross-section yields immediately another figure of merit for a modern
experiment, the luminosity. Technically, it is defined as the number of particles in two
beams, N1 and N2 , which interact in an (effective) area A with frequency f ,
N1 N2 f
L= ,
A
i. e. this is the number of processes per unit time and unit area. In modern particle physics
experiments, the beam is usually bunched into n packages, which act like there would be
several beams, and thus multiply the luminosity by n.
More important is the integrated luminosity,
ZT
L= dtL,
that is how many collisions per unit area occur during an experiment of temporal length
T . From this quantity, the number of events N occurring with a cross-section can be
calculated as
N = L.
Especially, to obtain one event occurring with cross-section during the time of the
experiment, an integrated luminosity of L = 1/ is necessary. This permits to directly
estimate whether an experiment is able to measure a process at all. Typical values for
integrated luminosity at the LHC are 25 fb1 until the end of 2012, and at least 300 fb1
until 2018.
A concept which will become important later on is the so-called partial (or later parton)
luminosities. Assume that the beams consist not out of a single type of projectiles, but
several. Than one can define for each possible pairing an individual partial luminosity.
2.2 Relativistic
2.2.1 Repetition: Relativistic notation
A non-relativistic description is, unfortunately, not sufficient in particle physics. There are
three reasons for this limitation. One is that there are massless particles, which necessarily
move with the speed of light, and therefore require a relativistic description. The second
is that many regularities of particle physics are very obscure when not viewed from a
relativistic perspective. Finally, in the reactions of elementary particles the creation and
12 2.2. Relativistic
annihilation of particles, a conversion of energy to mass and back, is ubiquitous. Only

(special) relativity permits such a conversion, and is therefore mandatory.
To just repeat the most pertinent and relevant relations, the starting point are the
Lorentz transformations for a movement in the 3-direction,
x1 = x1
x2 = x1
x3 = (x3 x0 )
x0 = (x0 x3 )
1
= (1 2 ) 2
which leave the length x2 = x20 ~x2 invariant. The quantity x denotes the corresponding
four-vector with components x in contrast to the spatial position vector ~x with compo-
~ is in natural units just the speed.
nents xi . Note that = ||
The scalar product of two four-vectors can be obtained using the metric g = diag(1, 1, 1, 1),
i. e.,
x2 = x x = x g x ,
i. e., the metric lowers and raises indices, which distinguish covariant and contravariant
vectors. Though there are profound geometric differences between both, these have es-
sentially no bearing on particle physics in the context of the standard model of particle
physics.
From this follows the definition of the relativistic momentum and energy,
p = m
E 2 = p~2 + m2 (2.1)
where m is the (rest) mass of the particle, an intrinsic and immutable (at least within
known particle physics theories) property of any given type of particle. The concept of a
relativistic mass, defined as m is actually not necessary nor relevant in particle physics,
since rest mass, and spatial momentum completely characterize the state of a particle.
Note that for ~p = ~0 the relation (2.1) implies E = m, and thus rest mass and energy in
the rest frame are exchangeable concepts. Especially, this relation implies that mass and
energy can be freely converted into each other in special relativity.
An important special case are massless particles, which always move at the speed of
light, i. e. = 1. In that case, E = |~p| precisely. Note that is neither well-defined, nor
needed, when describing massless particles.
The relation (2.1) is only true for real particles. The virtual particles stemming from
quantum fluctuations introduced earlier do not fulfill it. Such particles are called off-
shell, since the relation (2.1) defines a surface in momentum space, called the mass-shell.
Particles fulfilling (2.1) are consequently called on-shell.
One further consequence relevant to particle physics is time dilatation, i. e. the time
(difference) observed in a frame moving relative to the rest frame of the particle is longer
than the eigentime in the rest frame as
= . (2.2)
Of course, this also implies that a moving particle experiences in its own rest frame a
shorter time than outside. This will be very relevant for objects with a finite life-time: They
exist the longer the faster they are. This has especially consequences for the experimental
accessibility of massive particles.
Elastic scattering proceeds then in the relativistic case just as in the non-relativistic
case. The only exchange is that the separate conservation of spatial momentum and energy
is replaced by the conservation of four momentum,
p1 + p2 = q1 + q2 .
Still, the identities of the particles are conserved in an elastic scattering, and therefore
their rest mass is not changed. Also, there is still only one scattering angle characterizing
this (also still coplanar) reaction.
However, an interesting new possibility are now inelastic scatterings where particles
are produced or destroyed1
2.2.2 Inelastic scattering

In particle physics, it is no longer necessary that the particles scattered retain their identity
after the scattering process. Especially, they can be transformed into a number of other
particles, a so-called inelastic collision. Of course, the final products must still conserve
4-momentum. Thus,
X
p1 + p2 = qi .
n
1
In principle, e. g. by breaking up into more than two fragments, there are also non-relativistic inelastic
scattering processes possible, where the total mass is conserved. Though this plays an important role in,
e. g., chemistry, this possibility will be ignored here .
14 2.2. Relativistic
This implies especially

q q Xq
2 2 2 2
p~1 + m1 + p~2 + m2 = ~qi2 + m2i
i
X
~p1 + p~2 = ~qi .
i
In the center-of-mass frame, where the initial 3-momentum vanishes, it is thus in principle
possible to transform the rest energy and the kinetic energy of the initial particles only
into the rest energy of a number of new particles. This is called an on-resonance produc-
tion. Especially, it is possible to create particles which are heavier than the original ones,
provided the kinetic energy is large enough. This explains why higher and higher energies
are required in particle experiments to obtain heavier and heavier particles.
Of course, this can be generalized to more than two incoming particles. However,
in practice it is almost impossible in an experiment to coordinate three particles such
that they collide simultaneously. Hence, this will not play a role in the following. It
is not irrelevant, though. If the number of collisions becomes much larger than currently
technically achievable, this can happen for an appreciable fraction of the events. Inside the
sun such huge numbers are possible, and then three-body interactions become important
for the thermonuclear processes. In fact, without a certain reaction of three helium nuclei
to a carbon nucleus, the production of heavier elements in the sun would like not occur in
the way it does in nature.
This also yields another important way of how to deconstruct cross sections. Assume a
collision of two particles of type x as the so-called initial state. As a quantum mechanical
process, the reaction will not only lead to a single outcome, a single final state, i. e. only
one set of final particles. If it is possible, given the conservation of four-momentum and any
other constraints, that there are more than one possible outcomes yi , they all will occur.
Their relative frequency is given by the details of both the reaction and the underlying
interactions. These rates ni for n final states yield the partial cross-sections,
ni
2xi = lim Pn .
i ni
ni
Partial cross sections can also be defined in a broader sense. Instead of having precisely
defined particles in the final state it is e. g., possible to define a partial cross-section for n
particles of any type.
Partial cross-sections play an important role in experiment and theory alike. Measuring
a total cross-section is called an inclusive measurement, while identifying some or all
final particles and their properties are called exclusive measurements. At the same time,
theoretical calculations permit to make a judicious choice of a final state in which the
background due to known physics is small or vanishing, thereby increasing the signal-to-
noise ratio. Determining partial cross-sections is therefore a very important task.
2.3 Decays
In a sense a very particular type of inelastic cross-section is the decay of a single particle,
i. e. a cross-section with a single initial particle, and two or more final particles. A decay
occurs if a massive particle can, in principle, be converted into two or more lighter particles.
For this to be possible, the rest mass of the original, or parent, particle must be larger
than the sum of the rest masses of the daughter particle, plus possible a certain amount of
energy to satisfy spatial momentum conservation. Furthermore, there may be additional
constraints like that electric charge is conserved.
Since the decay of a particle is a quantum-mechanical process, it does not occur after
a fixed decay time , but occurs at a time with an exponentially decaying probability,
exp( /2), where is the so-called decay width. The value of the decay width is
determined by details of the interaction, but is in most cases the larger the larger the
difference between the rest mass of the initial particle and the total rest mass of the final
particles is. Furthermore, the ratio /m is used to characterize how unstable a particle is.
If this ratio is small, the particle is rather stable, as it can traverse many of its Compton
wave-lengths before decaying. Such particles are also known as resonances. If the ratio is
close to or larger than one, the particle is not really behaving as a real particle, and there
is no final consensus on whether such objects should be regarded as particles or not. Most
particles well accessible in experiments have small widths.
However, quantum mechanically all possible outcomes can always be realized, just
with differing probabilities. Therefore, if it is kinematically, and with respect to all other
constraints, possible for a particle to decay into different final states, this will occur. Each
possible outcome is called a decay channel, and is characterized by a corresponding partial
decay width i , The sum of these widths together can be shown to yield the total decay
width, which determines the life-time of the particle.
For the experimental measurement of the properties of resonances time dilatation (2.2)
plays an important role. Since in experimental particle physics decay products are often
produced at considerable speeds, their life-time in the experiments rest frame is usually
much larger than the life-time in their respective rest-frame. As a consequence, they can
travel over considerable distances away from where the original particle has been produced,
which permits to disentangle the processes of resonance production and decay. This also
permits to much better determine the decay products, which in turn permit to identifying
16 2.4. Cross-sections and particles
the particle which has decayed.
2.4 Cross-sections and particles

The formation of resonances yields, at least to leading order, a unique signature in the
cross-section. Consider the situation that two particles of momenta p1 and p2 are collided,
and react, elastically or inelastically, with each other, yielding two final particles with
final momenta q1 and q2 with p1 + p2 = q1 + q2 . If this process occurs such that the two
original particles annihilate and form a resonance, which consecutively decays into the
final products, it can be shown that the cross section behaves as
1
(2.3)
((p1 + p2 )2 m2 )2
If the initial momenta are tuned such that the center of mass energy s fulfills s = (p1 +
p2 )2 = m2 , the cross-section diverge, and thus the name resonance for the intermediate
particles. Away from this condition, the cross-section quickly drops. Thus the formation
of an intermediate resonance is signaled by a sharp peak in the cross-section. Since the
relevant variable is s, this is also called an s-channel process.
In contrast, later processes will be encountered, where one particle can be transmuted
into another by emitting a third particle. In these cases the relevant variables to be tuned
are differently, either in a so-called t = (p1 q1 )2 channel or, if the final particles are
exchanged, in the u = (p1 q2 )2 channel. These three variables together, s, t, and u are
denoted collectively as Mandelstam variables. Observing a resonance in any of the channels
is already providing insights into its production process, as the different kinematics can
be shown to stem from different physical origins.
Of course, in nature there is not a real divergence in the cross-section, since any res-
onances has a finite, though possible small, width. Including this width leads to a Breit-
Wigner cross section in, e. g. the s-channel
2
p 2
. (2.4)
( (s) m)2 + 4
Thus, the divergences becomes a peak with a width determined by the decay width of the
particle, which is therefore also accessible to an experimental measurement.
Of course, this is an idealized situation. Higher order processes or processes of a dif-
ferent origin can all contribute to the cross-section. Already a second resonance close by
could distort the cross-section appreciably. Furthermore, quantum mechanically interfer-
ence processes could suppress a peak or, even worse, create a fake peak. This has been
observed, and is therefore not just an academic possibility.
Finally, if the process is inelastic, there could be more than just two particles in the
final states. Then the cross-section becomes more involved, and the presence of resonances
is not just a simple peak. There are advanced techniques to deal with such situations, like
the Dalitz analysis, which now lead too far.
One should not that the formula (2.4) is only accurate if /m is small. For resonances
with a large width, the shape can be significantly distorted to the point where no real
peak is observable at all.
2.5 Feynman diagrams

As has now been repeatedly emphasized, scattering experiments are the central experi-
mental tool in particle physics. Accordingly, theoretical calculations of them have become
the single most important technique. The usual task is therefore to have some initial state,
e. g. two protons at the LHC, and then determine the partial cross-section for a given final
state, inclusive or exclusive.
The most well-known tool for doing so is perturbation theory, in which the interactions
between the particles is expanded in a power series of its strength. Though it is well-known
that such an expansion is not able to capture all features of such a theory, there are very
many experimentally accessible processes in which already low orders in perturbation
theory, usually one to four, yield the dominant part of the relevant cross-section. The
lowest order is under the assumption of quick convergence of the series, usually called
leading order, abbreviated by LO. The next order is called next-to-leading order, NLO,
and so on as N...NLO.
Such perturbative cross-section calculations in quantum field theories can be organized
similar to quantum mechanics, i. e. as a mathematical series. Such a series can be graphi-
cally represented, which is called Feynman diagrams. In such diagrams a line corresponds
to the propagation of a particle, a so-called propagator, while a vertex corresponds to an
interaction. Thus, the number of vertices corresponds directly to the order of the series
expansion. It can then be shown that each graph can be uniquely mapped to a mathemat-
ical expression describing the contribution of a given physical process to a cross-section.
Thus one can use the mathematical tool of graph theory to construct all contributions of
a given order to a physical process.
However, the possible vertices and propagators are not arbitrary, but are fixed for
each theory in the form of so-called Feynman rules, which uniquely determine all the
possible lines, vertices, and composition rules. These have to be determined for each theory
anew, though this is an algorithmic procedure which a simple computer algebra system
18 2.5. Feynman diagrams
can perform. Especially vertices are restricted by conservation laws, like four-momentum
conservation.
However, in practice for a reasonable complicated theory the number of diagrams
grows factorial with the order of the expansion. Thus higher-order calculations require
sophisticated methods to deal with the logistical problems involved.
Despite the power of these expression, it should always be kept in mind that already a
hydrogen atom cannot be described with these methods.
Finally, it has turned out that also beyond perturbation theory a graphical language
can very often be introduced in a mathematical precise way. Hence, Feynman-graph-
like representations are ubiquitous in particle physics, though may mean very different
mathematical objects. Especially, they are not necessarily restricted to a series expansion,
and it is possible to encode also the full content of a theory graphically. Thus, one should
always make sure what entities a given set of graphs corresponds to.
Chapter 3
A role model: Quantum

electrodynamics
The first example of a particle physics theory, the first part of the modern standard model
of particle physics, to emerge was quantum electrodynamics, QED. It already contains
many of the pertinent features of the more complex standard model, while at the same
time is much simpler. It is therefore the ideal starting point to get acquainted with a
particle physics theory.
Quantum electrodynamics is describing the existence and interaction of electromag-
netically active particles. As such, it is the quantum generalization of the classical elec-
trodynamics, which already had been implementing special relativity.
3.1 Electrons and photons

QED, like any other particle physics theory, implements two roles for particles, though
particles may later also implement both roles simultaneously.
The first role is that of charge carrier. Classically, these are objects which carry an
electromagnetic charge. They therefore constitute the quanta of the electric current j =
(, ~j). The prime example of these charge carriers are the electrons. These elementary
particles are fermions, having spin 1/2, and have a rest mass of m 511 keV. To the
best of our current knowledge they are point-like1 , or at least smaller than 107 fm. Each
electron carries exactly one unit of electric charge, which is in natural units dimensionless,
and of a size of roughly e = 0.3. Its sign is chosen by convention, giving electrons a
negative charge, in accordance with classical physics.
1
The problem of infinite energy of point-like objects is resolved in quantum field theory, at least in an
effective way. This will be discussed below in section 3.7.
19
20 3.1. Electrons and photons
The second role is that of the force carrier. Such particles are exchanged by charge
carriers to transmit the force. The force carriers are thus the quanta of the electromagnetic
field, which transmit electromagnetic interactions classically. Thus, the particles carrying
the electromagnetic force are the quanta of the electromagnetic field, called photons. These
are massless2 bosons of spin one, and therefore so-called vector bosons.
~ and B,
In classical electrodynamics it is the electromagnetic fields E ~ which are associ-
ated with the force. These fields can be derived from the vector potential A,
Bi = ijk j Ak
Ei = i A0 0 Ai .
Classically, this potential does not appear in the Maxwell equations
~ E(~
~ x) = i Ei = 4(~x)
~ B(~
~ x) t E
~ = ijk j Bk (~x)~ei t E~ = 4~j(~x)
~ E(~
~ x) + t B(~
~ x) = ijk j Ek (~x)~ei + t B(~
~ x) =0
~ B(~
~ x) = i Bi =0
~ ~j + t =
i ji + t = 0,
which describe the interaction between the electromagnetic fields and the sources. How-
ever, already in the quantum-mechanical Hamilton operator of a (spin-less) electron
1
H= (ii eAi )2 + eA0 , (3.1)
2m
it is instead the vector potential that couples to the particle. Thus, the photons are
actually the quanta of the vector potential A, instead of the E and B fields3 . Classically,
~ and B
for any given E ~ fields the corresponding vector potential is not uniquely determined.
It is always possible to perform a gauge transformation
A A + , (3.2)
2
The concept of mass in quantum field theory is a non-trivial one. What precisely is meant by the
statement that the photon has zero mass involves a significant amount of technical subtleties, which will
be glossed over here. A full discussion requires a treatment of quantum field theory beyond perturbation
theory.
3
It is actually possible to write down a quantum theory also in terms of the E ~ and B~ fields. However,
such a theory is much more complicated, especially it contains non-local interactions, and is almost in-
tractable analytically. Hence the formulation in terms of the vector potential is used essentially exclusively.
It is called the localized version, since no non-local interactions are present.
Chapter 3. A role model: Quantum electrodynamics 21
with some arbitrary function . This is also true quantum-mechanically, except that also
the wave-function of the electron needs to be modified as
exp(ie). (3.3)
Thus, the quanta describing electrons and photons are not uniquely defined, but can
be altered. It can, however, be shown than any experimentally observable consequence
of the theory does not depend on the the function , and thus on the choice of gauge.
Theories with these features are called gauge theories. All theories, which have been
experimentally supported, which we know today contain at least one gauge field. They
therefore represent the archetype of particle physics theories. QED is just the simplest
example for such a gauge theory. Later on, much more complicated gauge theories will
appear, which eventually form the standard model of particle physics.
Since the gauge symmetry requires that the ordinary derivative in the Hamilton (3.1)
is replaced by the combination of the ordinary derivative, , and the gauge field Ai ,
Di = i eAi , this combination is called the covariant derivative. The name covariant
stems from the fact that the whole operator transforms in such a way under a gauge
transformation, as to make the whole Hamiltonian gauge-invariant4 .
It is worthwhile to mention that the gauge symmetry is essentially what was called a
redundant variable in classical mechanics. In classical mechanics, the usual first step is to
eliminate all redundant variables by enforcing all constraints. This is not done in particle
physics, and the presence of gauge transformations (3.2-3.3) is just a manifestation of this
redundancy. Keeping this redundancy is, in contrast to classical mechanics, technically
much more advantageous than eliminating it. The presence of this gauge degree of freedom
is thus rather a mathematical convenience, than a genuine feature of nature.
As noted before in section 1.5, there are anti-particles to every particle, and especially
the positron as an anti-particle to the electron. It is thus also a spin 1/2 fermion with the
same mass, but opposite electric charge e = 0.3. The photon is uncharged. It is therefore
its own anti-particle, and there is no anti-photon which could be distinguished by any
means from the photon.
3.2 Formulation as a field theory

So far, the formulation at the level of (3.1) is quantum-mechanical. As noted, quantum
mechanics cannot deal with relativistic effect, and one has to pass to quantum field theory
4
Note here the distinction between gauge-invariance, i. e. unchanged under a gauge transformation,
and gauge-independent, i. e. not transforming at all under a gauge transformation.
22 3.2. Formulation as a field theory
(QFT). QFT is technically much more complicated than ordinary quantum mechanics,
and is beyond the scope of this lecture. However, it is very useful to understand and use
the way QFTs are formulated, and this provides a quick way of representing a theory.
Furthermore, significant insight can be gained already by treating the QFT classically,
and thus just like classical electrodynamics, which is a classical field theory.
This should now be exemplified for the case of QED.
The simplest part is just starting with Maxwell theory, i. e. the theory of free electro-
magnetic fields. Since the Hamilton operator is not a Lorentz-invariant it is usually not
the best way to formulate a relativistic theory. A better starting point is the Lagrange
function L. For a field theory, this changes to a Lagrangian density L, or briefly denoted
by Lagrangian. Classical Maxwell theory takes the form
1
L = F F (3.4)
4
F = A A , (3.5)
which is explicitly invariant under the gauge transformation (3.2). The anti-symmetric
tensor F is the field-strength tensor, with components formed by the electromagnetic
fields. The fact that the photon is a spin one boson can be read off from the Lagrangian
since it carries a single Lorentz index, and therefore transforms like a vector, and is de-
scribed by an ordinary field.
Adding fermions, and thus electrons, is subtle. Fermions obey the Pauli principle.
There is no classical object, which does the same. To formulate a field theory for them,
it was necessary to construct mathematical entities which encode this property, so-called
Grassmann numbers. A Grassmann number has the property that it squares to zero, i.
e. 2 = 0, and anti-commutes with any other Grassmann number , = . This looks
rather strange at first,but can be dealt with quite straightforwardly in practice. Here,
however, it is only necessary to remember that one cannot commute fermion fields, which
are fields of Grassmann numbers, at will.
Spin 1/2 fermions, like electrons, are described by four complex Grassmann variables,
which are collected into a column matrix . The anti-particle can then be shown to be
described by the field = + 0 with 0 a four-dimensional matrix, which is unitarily
equivalent to diag(1, 1, 1, 1). The total Lagrangian of QED then becomes
1
L = F F + (i ieA ) m, (3.6)
4
where m is the mass of the electron. The fields representing the electron, the so-called
spinors , transform under gauge-transformations like the quantum-mechanical wave-
functions, and therefore this Lagrangian becomes gauge-invariant. These spinors are four-
dimensional vectors of complex Grassmann numbers. Though they are four-dimensional
they do not transform like vectors under Lorentz transformations, but differently, hence
the name spinors. The details of this will not be relevant in the lecture, and are therefore
skipped.
Furthermore, the matrices i form together with 0 the Dirac matrices , and fulfill
the so-called Clifford algebra
{ , } = 2g .
Note that despite the suggestive notations these matrices are invariant under Lorentz
transformations, but play an important role in showing that this theory is, indeed, Lorentz-
invariant. Nonetheless, the metric can be used to define -matrices with raised indices as
well. The details o this are beyond the present scope. However, it is important to note
that the quantity is a scalar and is a vector under Lorentz transformations. It
is furthermore possible to define the matrix 5 = i0 1 2 3 , which can be used to create
a pseudo scalar 5 and an axial vector 5 .
Besides the presence of photons and (massive) electrons and positrons another property
of the theory can be read off the Lagrangian (3.6). There is a term which involves not
two, but three fields, the interaction term ie A . This is the first example of an
interaction vertex, which specifies an interaction of three particles, an electron, a positron,
and a photon. They interact with the coupling strength e, the electromagnetic coupling.
The remainder factors i ensure that this respects Lorentz invariance, and that it yields
a meaningful quantum theory. Note that there is no term describing interactions with
more than three particles in this Lagrangian. This will change later on. Furthermore, the
relation between positron and electron requires them that a mass term includes both of
them, and there are no separate mass terms for both particle species.
It is a further remarkable feature that the involved fields are functions not only of
space, but also of time. Thus, a field configuration, i. e. any given field, represents a
complete space-time history of a universe containing only QED.
As emphasized before, the Lagrangian is a classical object, i. e. the fields are not
operator-valued. The quantization yields a QFT, a topic treated in a separate lecture.
Here, it suffices to use the language of Lagrangians to specify a theory.
3.3 Indistinguishable particles

The formulation in terms of the Lagrangian (3.6) also answers the question why elementary
particles cannot be distinguished in any way. This Lagrangian contains only a single
electron field, a single positron field, and a single photon field. Each of these fields can
24 3.4. Gauge symmetry
describe an arbitrary number of particles of the corresponding type. Thus, since the
corresponding particles are only excitations of the same field, they cannot be distinguished.
How can such particles then be identified at all? This requires to understand what a
particle in the context of a field theory really means. A single particle is usually identified
with an (almost) Gaussian excitation of the field5 , and therefore exponentially localized.
Two separate particles are therefore two Gaussian excitations, where the two peaks are
separated far compared to the widths of the peaks. Thus, the two particles can be iden-
tified, but since they are just the same excitation type of the field, they are nonetheless
indistinguishable. This can be repeated for as many particles as desired, as long as the
separation is large compared to the widths, creating many indistinguishable particles.
If the two particles come close to each other, the peaks overlap, and it no longer
makes sense to speak of individual particles. Interactions play a role, and particles in
the conventional sense only reemerge when the particles have moved far away from each
other. In a perturbative description of a scattering process, the starting point are two such
Gaussians sufficiently far apart as to ignore any remaining overlap. The same applies to
the final states. That is what is called asymptotic states.
3.4 Gauge symmetry

The gauge transformations (3.2-3.3) have indeed a quite deeper structure than visible at
first sight. Especially, they implement a symmetry, the gauge symmetry. The electron
fields are modified by an arbitrary phase factor at every space-time point. This phase
factor is a complex number, and therefore corresponds to a two-dimensional rotation. It is
therefore an explicit implementation of the two-dimensional rotation group SO(2)U(1).
As a consequence, the gauge symmetry is called a U(1) gauge symmetry, and the group
U(1) is called the gauge group. Since the elements of U(1) commute, this is also called an
Abelian gauge theory.
The electron field transforms like an element of this group, i. e. by multiplication.
The electron is therefore said to be in a representation of the gauge group, the so-called
fundamental representation.
The photon field does not transform by a phase factor, but instead by an additive
function, which is related to the phase factor by exponentiation. It can be shown that
these functions implement an algebra, the so(2)u(1) algebra. Therefore, the photon
field is said to be in a different representation of the gauge group, the so-called adjoint
5
As usual, the reality of field theory is far more complicated than this, but this picture is sufficiently
close to the true mechanism to give the correct idea.
representation.
Such structures will again reappear for the other parts of the standard model, and are
also present in gravity. So far, only the fundamental and adjoint representation play a role
in particle physics, though several proposals exist for theories beyond the standard model
in which particles in different representations would appear. They would then transform
differently under gauge transformations. The precise form is given by the relevant gauge
group, and is mainly an exercise in group theory.
For comparison, gravity can also be thought of as a gauge theory, with as gauge group
the non-Abelian Poincare group SO(3,1)T, where T is the (Abelian) translation group.
This is, however, beyond the scope of this lecture.
3.5 Bound states and the positronium

Already the simple system of QED in form of (3.6) yields a plethora of interesting phe-
nomena, which will be discussed in the following. The first one is that besides the electrons
and photons6 it is possible to construct bound states. The simplest bound state usually
encountered is an atom. Lacking a proton or nuclei so far, the simplest bound state, which
can be constructed in QED is positronium. This bound state consists out of an electron
and a positron. In principle, it is just like a hydrogen atom, but its nuclei is just as heavy
as the electron. This state has also been observed in experiments.
Of course, such a bound state is not stable. As stated, matter and anti-matter can
annihilate each other, and so can electrons and positrons. Thus, after a while, the electron
and the positron will annihilate each other into two or three photons (one is not permitted
due to four-momentum conservation). The two different options depend on the relative
alignment of the electrons and positrons spin, ether parallel or anti-parallel.
Thus, even such a simple theory as QED has already highly non-trivial structures.
3.6 Muons and Taus

Before delving further into the theoretical features of QED, it is time to introduce a
property of nature which was not foreseen at the time QED was invented, and for which
till today no convincing explanation has been found. This is the fact that the electron has
two heavier siblings, and likewise does the positron. One is the muon, having a mass of
about 105 MeV, and the other the tauon, with a mass of around 1777 MeV. Thus, both
6
Again here some subtleties arise in a quantum-field theory when one talks about electrons or photons
as particles. This is also beyond the scope of this lecture.
26 3.7. Radiative corrections and renormalization
are much heavier than the electron. Otherwise their properties are the same as for the
electron: They are fermions with spin 1/2 and negative electric charge. They can be easily
included into the Lagrangian (3.6),
1 X
L = F F + f (i ieA mf )f , (3.7)
4 f
where f now counts the so-called flavor, i. e. type, of the fermions, and can be electron,
muon, and tauon. In QED, this flavor is conserved, i. e. it is not possible to convert the
heavier muon into an electron. Later, additional interactions will be encountered, which
change this. Then neither the muon nor the tauon are stable, and thus are encountered
in nature only as short-lived products of some reaction.
The electron, muon, and tauon are otherwise exact replicas of each other. They form
the first members of three so-called families or generations of elementary particles, which
will have, except for the mass, otherwise identical properties. Their presence, and the
relative sizes of their masses form one of the most challenging mysteries of modern par-
ticle physics, subsumed under the name of flavor physics. However, several interesting
phenomena we observe in nature are only possible, if there are at least three such families,
and some will be encountered later on.
To distinguish the electron, muon, and tauon from other particles to be discussed later,
they are also collectively denoted as leptons.
3.7 Radiative corrections and renormalization

When one tries to perform perturbative calculations in QED, this is actually rather
straightforward to leading order. However, a serious problem occurs if an attempt is
made to go beyond tree-level. These were first encountered when people attempted to
calculate corrections due to the emission and absorption of photons from electrons bound
in atoms. Therefore, such contributions are nowadays collectively called radiative correc-
tions, though in most cases this name is less than accurate.
The problem encountered in the calculation stems from the combination of two features
of quantum field theories. One is that energy may be borrowed for some times in the form
of virtual particles. The other is that quantum mechanically everything what could happen
happens and mixes.
Therefore, in the next order of perturbation theory generically the possibility appears
that an electron can emit a virtual photon and then reabsorb it later. Or a photon can
split into a virtual electron-positron pair, which later annihilates to form again the photon.
So far, this is not a problem. But these virtual particles can borrow energy, and therefore
may be created with arbitrarily large energies. These arbitrarily large energies then induce
divergences in the mathematical description, and all results become either zero or infinite.
This is, of course, at first sight a catastrophe, and for a while it seemed to mark the
end of quantum field theory. It is actually no an artifact of perturbation theory, as one
might suspect; though there are theories where this is the case, this is not so for those
relevant for the standard model of particle physics.
Today, this failure has been understood not as a problem, but rather as a signal of the
theory itself that it cannot work at arbitrarily high energies. It signals its own breakdown,
and the theory is, in fact, only a low-energy approximation, or a low-energy effective
theory, of the true theory. This is then as expected, since gravity is not yet part of the
standard model, but is expected to play an important role at very high energies. Hence
the prediction of failure is consistent with our experimental knowledge on the presence of
gravity.
This insight alone does not cure the problems, however. Fortunately, it turns out that
a certain class of theories exist, so-called renormalizable theories, where this problem can
be cured in exchange for having a finite number of undetermined parameters. In case
of the full QED Lagrangian (3.7) these are four: The strength of the electromagnetic
coupling and the three masses. We exchange the divergences in favor of having these four
numbers as an input to our theory, which we have to measure in an experiment. Once
these parameters are fixed, reliable predictions at not too high energies can be made,
where not too high for the standard model covers an energy range larger than our current
experimental reach, probably much larger.
This exchange, a process known for historical reasons as renormalization, was the final
savior of quantum field theory, making it again a viable theory. It also implies that a
renormalizable theory cannot be a complete description of physics: The parameters of the
theory cannot be predicted from inside the theory, and external input is needed. Whether a
completely parameter-free theory can be formulated, which describes nature, is not known.
3.8 Fixing the inputs and running quantities

In the previous section 3.7, it was sloppily said that it is necessary to fix the parameters,
in case of QED three masses and one coupling. This is a far less trivial process than it
seems at first.
The problem is most easily seen when it comes to fixing the electric charge. To do so
requires to define what is meant by the term electrical charge. Classically, this is simple.
Together with the electron mass, it can be determined by the deflection of an electron in
28 3.8. Fixing the inputs and running quantities
both an electric an a magnetic field. Quantum-mechanically, this is not so obvious. As

stated, what is observed in nature is not just the electron. What actually looks like an
electron, is an electron which is constantly emitting virtual photons, which themselves may
emit virtual electron-positron pairs. It is this electromagnetic cloud, which surrounds the
electron, what is measured. When the electron is probed at shorter and shorter distances,
the probe travels farther and farther into this cloud. Therefore, what is measured, depends
on the distance traversed, and thus the energy of the probe. Hence, it is necessary to make
a definition of what electrical charge is, since it is not possible to measure only the process
at tree-level.
For QED, this is usually done by measuring Thompson scattering. This is the scattering
cross-section of a photon of small energy from an electron at rest. Aside from some trivial
geometric and kinematic factors, the size of this scattering is determined by the fine-
structure constant = e2 /(4), and thus by the electric charge. By fixing the parameters
in the Lagrangian of the theory (3.7) such as to reproduce this scattering cross-section,
one of the parameters is fixed, in this case to the value of roughly 1/137.
This at first sight simple prescription has two inherent subtleties. The first is rather
of a technical nature. For no relevant theory it is possible to calculate things like a
scattering cross-section exactly. As a consequence, the dependence of the calculated cross-
section on the parameters of the Lagrangian is only approximate, and therefore their
determination can only be approximate. Moreover, two different calculations done with
different approximations will not necessarily yield coinciding values for the parameters. It
is therefore mandatory to make sure that any comparison has to be done in such a way as
to take such approximation artifacts into account.
The second subtlety is far more complicated. It has just been defined what the electric
charge is. However, it is not forbidden to instead define the electric charge by measuring
the scattering cross-section for an electron and photons having all a certain energy, say
. This is the so-called symmetric configuration. In general7 , the scattering cross-section
will depend on this energy , and hence the so-defined electric charge will be different,
and depend on the actual value of , = (). Thus, there is no unique definition of the
electric coupling, but a coupling depending on the energy where it is measured. This is
called a running coupling, a concept of great importance.
The change of the running coupling with the energy scale is also called the renor-
malization evolution, and it is possible to write down a mathematical group where the
elements are changes in this scale. This is the so-called renormalization group, a very
7
There are some theories, so-called conformal theories, where this is not the case. None of the theories
realized in nature is of this type, though. This can actually be excluded exactly from experiment.
powerful tool when performing actual calculations in quantum field theory.

Similarly, the electromagnetic cloud of an electron will be deformed, if it is probed at
higher and higher energies. As a consequence, the effective mass of an electron is also
not independent of the energy8 , and the mass depends similarly on the energy me (), a
running mass.
This running of quantities is a characteristic of quantum field theories. In fact, testing
the energy-dependence of the running quantities is one of the most stringent tests, which
can be performed on the validity of a theory. For the standard model, this has been done
for the running over several orders of magnitude, a marked success of the theory.
It now may appear that one parameter has been traded in for a function. This is not
the case. It is actually sufficient to measure the value of four running quantities at a single
energy to fix all parameters of QED uniquely. The remainder of the functions is then
uniquely determined by the theory and the four parameters. It is thus sufficient to probe
a theory at a single energy scale, and it is then possible to make predictions about it at
completely different energy scales.
A final word of caution must be added. In the present case, the electric charge was
defined at the symmetric point, to obtain a single running coupling. There is no reason to
make this choice. An equally well acceptable choice would be to chose the photons energy
twice as large as the one of the electron, and deduce everything from this configuration.
In the end, physics must be independent of such choices. This is indeed what happens.
Such different choices are called renormalization schemes, and the change between two
such schemes is called a renormalization scheme transformation. Though a mathemati-
cal well-defined process, it is important to compare only results in the same scheme, as
otherwise the comparison is meaningless.
3.9 Landau poles

One of the mainstays of modern particle physics remains perturbation theory. However,
perturbative calculations can, strictly speaking, never be entirely correct9 , and there are
always non-perturbative contributions to any quantities at all energies. Still, in many
relevant cases, the perturbatively calculable contribution is by far the dominant part.
One of the arguably most famous results of perturbation theory is the running of the
8
Here, again, a subtlety of advanced quantum field theory plays a role when it comes to define what
an electron is. This will also be skipped, as it is of little effective importance.
9
Formally, a perturbative series is expected to break down roughly at order O(1/x), where x is the
expansion parameter.
30 3.9. Landau poles
coupling (). It shows that the running coupling satisfies the differential equation
de e3
= (e) = 0 + O(e5 ), (3.8)
d ln 16 2
where is the so-called -function, and i are its Taylor coefficients. The latter can be
calculated in perturbation theory. Truncating the series at the lowest order, the differential
equation can be integrated, and yields
e(2 )2 (20) 4
(2 ) = = 2) 2 , (3.9)
4 (
0 ln 2
2
1 + 40 0 ln 2
0
where 0 is the energy where the experimental input (0 ) has been obtained. The so
introduced quantity is often called the scale of the theory, and is a characteristic energy
of the described interaction. The value of 0 for QED is -4. This implies that the running
coupling becomes larger with increasing energy, until it eventually diverges at a very large
energy scale.
This divergence is the so-called Landau pole. Similar problems arise also in the other
running quantities. The presence of such a Landau pole can be traced back to the use of
perturbation theory. Before reaching the Landau pole, the running coupling, which is the
expansion parameter, becomes larger than one, and therefore the perturbative expansion
breaks down. Thus, a Landau pole is a signal of the breakdown of perturbation theory.
When this point is reached, it is necessary to resort to non-perturbative methods. Even
for QED such a full non-perturbative treatment is not simple. The results, however, indi-
cate that QED is also breaking down beyond perturbation theory, and the only quantum
version of QED which make sense is the one with e = 0, a so-called trivial theory. This is
the triviality problem. It is assumed that this effect is cured, once QED is embedded into
a larger theory, in this case the standard model of particle physics. This is a recurring
problem, and will be discussed in more details later.
Chapter 4
Invariances and quantum numbers
4.1 Symmetries, groups, and algebras

As has already been seen in section 3.4, symmetries and groups play an important role in
particle physics. This becomes even more pronounced when eventually the whole standard
model will be formulated. A technically convenient and tractable formulation of particle
physics appears so far only possible by using gauge symmetries. However, these are no
the only kind of symmetries which play an important role in particle physics. The most
important ones will be discussed in the following sections.
Invariances have two important consequences. One is that the observation of an in-
variance strongly restricts how the underlying theory can look like: The theory must
implement the invariance. The second is that if such an invariance is implemented, then
as a consequence different processes can be connected with each other. E. g., one of the in-
variances to be encountered below is connected with the electric charge. As a consequence
the scattering of two particles with given charges will have the same cross section as for
the anti-particles with electric charge reversed. Hence, invariances are powerful tools in
theoretical calculations, and powerful experimental tools to unravel the structure of the
underlying theory.
An important classification of symmetries should be mentioned beforehand. In particle
physics, there are three most important classes of symmetries. The first are discrete
symmetries, like reflection in a mirror. They act, and do not involve any parameters.
In contrast, continuous symmetries have transformations which depend on one or more
parameters. If this parameter is constant, this is called a global symmetry. If it depends
on space-time, as for the gauge transformations of QED, it is called a local or gauge
symmetry.
Since symmetries imply the existence of an operator which commutes with the Hamil-
31
32 4.1. Symmetries, groups, and algebras
ton operator, they lead in general to conserved quantum numbers. These can be either
multiplicatively or additive. This means that a for a composite state the quantum num-
bers of the constituents are either multiplied or added. Discrete symmetries usually lead
to multiplicative ones, while continuous ones to additive ones. Note that not all states
have well-defined quantum numbers, even if a symmetry exists, and that a composite state
can have well-defined quantum numbers even if the constituents do not have.
Furthermore, it is necessary to introduce a few notions of group theory. Later, in
section 5.15, this will be further elaborated. For now, it is sufficient to introduce a few
concepts. Most of these concepts are already familiar from quantum mechanics, especially
from the theory of atomic spectra, and thus angular momenta. However, these concepts
require generalization in the context of particle physics.
The basic property of a symmetry is that there exists some transformation t, which
leaves physics invariant. Especially, such a transformation must leave therefore a state |si
invariant, up to a non-observable phase. Thus, a transformation has to be implemented by
a unitary or anti-unitary operator, T . Especially in cases of continuous symmetries, it is
possible to act on any state repeatedly with different values of the relevant parameter(s),
T (1 ) = T1 , T (2 ) = T2 yielding T2 T1 |si. There is also always a unit element, i. e. one
transformation which leaves every state invariant. It is also furthermore for a unitary
transformation possible to reverse it. Hence, such transformations have a group structure,
and it is said that they are the corresponding symmetry group.
An example has been the group of the gauge transformations from section 3.4. There,
the group was SO(2)U(1). The group elements were the phase rotations exp(i(x)), with
space-time-dependent parameters. It was thus a local symmetry. The unit operator is thus
just the one, with = 0. The inverse one is just exp(i(x)), the complex conjugated
one, as expected for a unitary operator.
Since the application of two gauge transformations commute, this symmetry group is
Abelian. However, a consecutive application of more than one group element need not be
commutative,
T1 T2 |si =
6 T2 T1 |si.
If it is not commutative, the symmetry group is called non-Abelian.

The gauge field did not transform in the same way, see equation (3.2). In fact, the
transformation did not seem to be unitary at all. It is, however, only the manifestation of
a different mathematical version of the symmetry. Since interesting symmetry transfor-
mations are (anti-)unitary, they can always be decomposed into a complex conjugation or
a one, for anti-unitary or unitary transformations, respectively, and a unitary transforma-
Chapter 4. Invariances and quantum numbers 33
tion. Any unitary transformation can be written as1

i
T = eii ,
where the i are Hermitian operators. These operator form the so-called generators of
the symmetry group, and form themselves the associated symmetry algebra. Hence, the
gauge fields transform not with the group, but with the algebra instead.
To separate both cases, it is said that the electron states and the gauge fields transform
in different representations of the gauge algebra.
4.2 Noethers theorem

One of the central reasons why symmetries are so important in quantum physics is that it
can be shown that any continuous symmetry entails the existence of a conserved current.
Such conserved quantities are of central importance for both theoretical and practical
reasons. Practical, because exploiting conservation laws is very helpful in constructing
experimental signatures which are not too contaminated by known physics or making
theoretical calculations more feasible. Theoretical, conserved currents mainly conserve
quantum numbers. Since states of different conserved quantum numbers do not mix, they
can be used to classify states.
The connection of symmetry and conserved currents is established by Noethers theo-
rem. It is already a classical statement. Essentially it boils down to the fact that if the
Lagrangian is invariant under a variation of the field, the variation can be used to derive
a conserved current. To keep it simple, start with a Lagrangian L(, ) of a single
real field , and its derivative . The Lagrangian is assumed to be invariant under the
transformation
+ ,
i. e.
L = L( + , + ) L(, ) = K , (4.1)
where K is a total divergence, which is neither classically nor quantum-field theoreti-

cally able to alter the behavior of the system2 . The corresponding equation of motions of
the field are the usual Euler-Lagrange equations
L L
= 0. (4.2)

1
Note that this relation is in general only unique up to a discrete group.
2
A least in flat space-time and for most quantum-field theories.
34 4.2. Noethers theorem
The explicit form of the variation (4.1) is then

L L
L = + ( )

L L
= + ( )

L
= ,

where the equation of motion (4.2) were used in the derivation and differentiations have
been exchanged. This implies that the Noether current
L
j = K ,

or its corresponding extension to arbitrary sets of fields
L L
j = K i i
i i
is conserved, j = 0. Hence, the associated charge
Z
Q = d3~xj0 ,
is conserved and thus time-independent,

dQ
Z Z Z
3 ~~
= dV ~xt j0 = dV ~xj = d2~n~j,
3
dt
V
where ~n is the unit vector on the surface S, stemming from the application of Gauss law3 .
Thus, the only change in the total charge is by the migration of charge carriers through
the surface in which the charge is enclosed. Hence, if the surface is send to infinity, where
no more charges can come from the outside, and the charge becomes time-independent.
In this way, conserved currents are linked to symmetries of the theory. This will be
encountered throughout the standard model of particle physics. Another useful feature
of symmetries in quantum-field theories is that they entail relations between different
quantities, so-called Ward-Takahashi identities for global symmetries and Slavnov-Taylor
identities for local (gauge) symmetries. This is a rather technical topic in detail, and will
therefore not be addressed further here.
3
Note that it was throughout tacitly assumed that the current and fields vanish sufficiently fast at
infinity to permit the integral manipulations. In quantum-field theory, this is not obviously true anymore.
But appropriate generalizations hold true in most cases, but not always, and care has to be taken. For
the content of this lecture, this can be assumed to be true.
4.3 Electric charge

It is worthwhile to apply this mathematical exercise to derive the electromagnetic current
of (Q)ED. There are now three terms, the variations of the fermions, anti-fermions, and
the photon- field. In this case, there appears no surface current K . The three terms are
L
= (ie) (i )

L
= (ie ) 0

L
A = ( ) F = 0,
A
where the transformation function has been dropped, as it is arbitrary, and all identities
have to hold anyway for any arbitrary choice. The product symbolizes the adequate
composition rules for the various tensors involved. Note that the term for the gauge field
vanishes because of the antisymmetry of the field-strength tensor. Thus the current is
j = e , (4.3)
where use has been made of the fact that the fermion fields anticommute, and there is
an implicit 0 in the anti-fermion field. As expected, the current is carried only by the
fermion fields, and the photons do not contribute. It thus reproduces the expected form.
Thus electric charge is conserved in QED, as it is in classical physics. This will also
not be altered when passing to the standard model of particle physics. However, there is a
remarkable feature. Since the gauge-transformation of the photon field (3.2) is independent
of the electric charge, it is in principle possible that the three generations, i. e. electron,
muon, and tauon, could have different and arbitrary electric charges. Furthermore, there
is no reason that the proton has the opposite and positive charge as the electron, as it is
not made of of positrons, as discussed below. Still, all experimental results agree to very
good precision with this equality, especially the latter one. In QED, there is absolutely
no reason for this fact. A discussed later, the standard model of particle physics is only a
consistent quantum-field theory if and only if all the electric charges fulfill certain relations.
However, the fact that a theory only works if certain experimental facts are taken into
account is supporting the theory. But is does not dictate that the experimental facts have
to be this way. If a small deviation of these relation would be observed tomorrow, it is
the theory which has the problem. Therefore, the condition that the electric charges are
as they are for the theory to work is not an explanation of why this has to be. It is an
experimental fact which can be described, but not explained inside the standard model.
36 4.4. Implications of space-time symmetries
Though there are several proposals why this could be the case, especially so-called grand-
unified theories. There is not yet any experimental support for any explanation, and it
remains one of the great mysteries of modern particle physics.
4.4 Implications of space-time symmetries

Noethers theorem is not limited to so-called internal symmetries, i. e. symmetries which
leave the space-time unchanged. It can also be applied to space-time transformations like
translations and rotations. The consequences of both is that four-momentum and total
angular momentum are conserved.
The spin of a particle is also a consequence of space-time symmetry, but in a much
more subtle way. It is possible to show that the Poincare group only admits certain values
of spin for particles, half-integer and integer4 . This does not yet imply anything about
the statistics which these particles obey. This requires further symmetries to be exploited
later.
All this depends on having Minkowski space-time implementing the Poincare symmetry.
These are two inputs to particle physics, and therefore the above listed properties are again
not a prediction but a feature of particle physics. Other space-times can endow a theory
with quite different properties, but for the subject of this lecture, it is fixed.
It also should be noted that the individual components of the spin have no meaning as
conserved quantities in particle physics. This is most immediately visible when performing
a Lorentz boost, under which they are not invariant. But for massless, and thus light-like
particles, there is an exception. Since there is no way by making a Lorentz-boost to move
into their rest-frame, the projection of the spin upon their momentum is always the same.
This projection is called helicity, and is sometimes a useful concept in particle physics,
when dealing with massless, or nearly massless, particles.
Note that space-time symmetries are global. Making them local leads to general rela-
tivity, which is not yet part of the standard model.
4.5 Parity
Another symmetry, closely related to space-time symmetry, is the discrete parity symme-
try. The discrete transformation parity P is essentially a reflection in the mirror, i. e.
4
In two space-time dimensions this is not the case, and the spin can take on any real number, leading
to so-called anyonic particles. They are of no relevance in the standard model, but play a certain role in
solid-state physics.
~x ~x. Time is unaltered. A distinction between space and time appears here, because
of the signature of the Minkowski metric, which makes time and space (in a loose sense)
distinguishable. Note that applying a parity transformation twice yields again the original
system. In fact, a parity transformation inverts the sign of each vector, e. g., coordinate
r or momenta p
P p = p.
Pseudo-vectors or axial vectors, however, do not change sign under parity transformation.
Such vectors are obtained from vectors, e. g., by forming a cross product. Thus the prime
example of an axial vector are (all kind of) angular momenta
P L = P (r p) = P r P p = r p = L.
As a consequence, fields can either change sign or remain invariant under a parity trans-
formation5 . Hence, a quantity can be assigned a positive or a negative parity, +1 and 1.
Generically, when there are only these two possibilities, the quantities are called even or
odd, respectively, in this case under parity transformations.
The fields describing particles have definite transformation properties under parity.
It is necessary to define the parity of some states, to remove some ambiguities regarding
absolute phases. Thus, the absolute parity of a particle is a question of convention. Usually,
the electrons, and thus also muons and tauons, are assigned positive parity, while the
photon has negative parity. Note that the anti-particles have the opposite parity for
fermions, but the same for bosons. The reason is the different statistics.
In QED, parity is conserved. Classically that can be read off the Lagrangian (3.7),
though there are some subtleties involved concerning fermions when performing the trans-
formation. Furthermore, it it is important to note that also differential operators have
definite transformation properties under parity transformation, and this is the same as
the one of the photon field. As a consequence all terms have a definite positive parity, and
the total Lagrangian has even parity. In principle, this could be changed in the quantiza-
tion procedure, but this does not occur for QED. Later, additional interactions of particle
physics will break this symmetry already classically.
4.6 Time reversal

Of course, the corresponding partner to parity is the discrete time-reversal symmetry
T , i. e. the exchange of t t, without changes to the spatial coordinates. This is
5
For fermions, there is a multiplication by 0 involved.
38 4.7. Charge parity
physically equivalent to reversing all speeds and momenta, and as a consequence also
angular momenta: Either objects move backward or the time moves backward. Time-
reversal symmetry is anti-unitary, and therefore involves besides a factor of 1 or -1 also a
complex conjugation of the object it acts upon. Similarly to parity, fields can be classified
as being even or odd under time-reversal. However, because time-reversal is anti-unitary,
there is no quantum number associated with it.
As with parity, time reversal symmetry is respected both classically and quantum-
mechanical by QED, and again later on theories will be encountered which do not respect
it.
4.7 Charge parity

In classical electrodynamics, it is possible to reverse the sign of all electric charges without
changing the physics. This is a classical example of charge parity.
In particle physics, it is the presence of particles and anti-particles, and the fact that
all their intrinsic (additive) properties are identical, that permits to define another discrete
symmetry, the quantum version of charge parity C. Charge parity defines how a quan-
tity behaves under the exchange of particles and anti-particles. Fields are also complex-
conjugated, making charge-parity an anti-unitary operation. Note that properties like rest
mass, momentum, spin, angular momentum or parity are not affected by charge parity.
It is again possible for a system to either change sign or not, defining even and odd
charge parity. But it is also possible to not have a particular property at all, which
is particularly true for charged state. That can be understood in the following way:
To have a definite charge parity, a state must be an eigenstate under the charge parity
transformation. However, charge parity transforms a particle with, say, electric charge +1
into a state with charge -1, and since such states are not identical, these states cannot
have a definite charge parity. However, neutral states like the photon can have. It charge
parity, e. g., is -1. Also, states made up from multiple charged states which are in total
neutral can have a definite charge parity.
Again, QED respects charge parity, and again, later on theories will be encountered,
where this is not the case.
4.8 CPT
It is now a very deep result that quantum-field theories on Minkowski space-time have
the following property: A combination of charge parity transformation, parity and time
reversal, i. e. the application of the operator CP T (or any other order, as they commute)
will always be a symmetry of the theory6 . Especially, causality is deeply connected with
this property, and one implies the other.
This implies that if one mirrors physics and exchanges particles by anti-particles, and
runs everything backwards there will not be any difference observed in any physical process.
From this, it can also be shown that necessarily all integer spin particles respect Bose-
Einstein symmetry, while all half-integer spin particles will respect Fermi-Dirac statistics.
The connection is made using special relativity: On the one hand it implies that only
certain spins may exist, and at the same time it implies that certain properties under
the exchange of particles have to be respected. Connecting this leads to the aforemen-
tioned classification. Therefore, in the context of particle physics, integer-spin particles
are synonymously called bosons and half-integer spin particles fermions. Since this is a
particular property of three-(or higher) dimensional quantum-field theory endowed with
special relativity on flat Minkowski space-time, which is the arena of the standard-model,
this association is fixed in particle physics. One must be careful, if one moves to a different
field.
As a consequence of this so-called CPT-theorem the behavior of a state or field under
one of the symmetries is fixed once the other two are given. Conventionally therefore
only the quantum numbers under P and C are recorded for a particle, which leads to the
J P C classification, where J is the total, orbital and intrinsic spin, angular momentum of
a particle, and P and C are only denoted as + or . E. g. a photon has J P C = 1 . The
charge parity quantum number of particles with anti-particles of half-integer spin is not
entirely trivial.
4.9 Symmetry breaking

Before continuing to further symmetries a brief intermission about breaking symmetries
is necessary. Take a Lagrangian L0 with a symmetry at the classical level. There are now
three possibilities how this symmetry may be affected.
The first is a somewhat obvious possibility, the explicit breaking. In this case an
additional term is added to the Lagrangian, L0 + Lb , with a parameter, which does not
obey the symmetry. Then , of course, already at the classical level the theory no longer
6
There are some attempts, whether it is possible to formulate quantum-field theories which do not
respect this, as experimentally there are only upper limits for any violation. However, so far none emerged
with a convincing concept of how this really defines a reasonable field theory. It is, of course, possible
that when going beyond the framework of quantum-field theory such possibilities may exist.
40 4.9. Symmetry breaking
has the symmetry. It can be shown that such an explicit breaking can only be performed
for global symmetries. Any attempt to do this for a local symmetry, though classically
possible, cannot be quantized consistently.
The relevance of explicit symmetry breaking comes about in two special cases. One
is that of soft symmetry breaking. In this case the terms in Lb are such that they are
only taking effect in a certain energy regime, but become negligible in another one. The
most common case will be seen to be mass terms, which break symmetries of most particle
physics theories at low energies, but become irrelevant at large energies. Therefore, if
the theory is probed at sufficiently large energies, the symmetry appears to be effectively
restored.
The second possibility is, if is small. In this case the symmetry may not be exact, but
relations due to the symmetry may still hold true approximately also in the full theory.
In such a case it is said that the symmetry is only weakly broken. This concept plays an
important role later for the strong nuclear interaction.
The next possibility is that a classical symmetry is no longer a symmetry of the quan-
tized theory. Again, this only leads to a consistent quantum theory if the affected sym-
metry is a global one. Such an effect is also called an anomaly. In that case the symmetry
is just not present at the quantum level. There are several examples to be encountered
in particle physics. The simplest one is given by massless QED. The classical theory is
invariant under a scale transformation, i. e. when rescaling all dimensionful quantities by
a fixed number. This symmetry, the so-called dilation symmetry, is broken in the quan-
tization. As a consequence, the theory has an intrinsic mass scale. This mass scale is e.
g. the one encountered in the running of the coupling, see section 3.8. If the dilatation
symmetry would be unbroken, the coupling would be independent of energy. Incidentally,
it can be shown that this is deeply connected to the masslessness of the photon.
If a symmetry exists both classically and survives the quantization process, it may still
be broken spontaneously. It can be proven that this cannot occur in usual particle physics
theories for local symmetries, but only for global symmetries. This will manifest itself in
the possible outcomes of the theory. E. g., if a symmetry predicts that two states should be
the same, they will no longer be after spontaneous symmetry breaking. The spontaneous
magnetization of a magnet below th Curie temperature is an often cited example for this
phenomenon. Again, there is some fine-print in a full quantum-field theory7 , but the
simpler picture will suffice for this lecture.
7
E. g. the total magnetization of a magnet is always zero, because there is no preferred direction in
space-time. Spontaneous magnetization manifests itself rather in the relative magnetization of two neigh-
boring spins in the magnet. These are uncorrelated without spontaneous magnetization, but correlated
in the case of spontaneous magnetization.
A more detailed treatment of what symmetry breaking really is will follow in section
5.7.
4.10 Fermion number

An inspection of the QED Lagrangian (3.7) shows that there is another symmetry. If the
fermion field is multiplied by a constant phase exp(i) and the anti-fermion with exp(i),
the Lagrangian remains unbroken. This is a continuous U(1) symmetry, and therefore a
Noether current has to exist. Calculating it, its is found to be proportional to the electric
current (4.3). However, it is not the electric current itself, but is actually the current of
fermion number, which just in the particular case of all fermions having the same electric
charge coincides with the former.
Therefore, the number of fermions, i. e. the difference of the number of fermions minus
the number of anti-fermions, is conserved. This does not require that these fermions
remain of the same type. E. g. a state with one electron or with one muon both have
the same fermion number, one. Any process respects this property. In fact, within the
standard model this number of fermions is always respected.
Note that there is no similar condition for photons, and they can be created and
destroyed at will, provided other conditions like conservation of total angular momentum
and four-momentum conservation are satisfied. This is not so because they are bosons,
but because they are their own anti-particles. Later on bosons will be encountered, which
carry charge, and there charge conservation will restrict this possibility. Still then no
analogue in the form of a boson number conservation exist.
4.11 Flavor
QED has, however, not only a single fermion species, but three ones, so-called three flavors.
Multiplying only one of them with a phase still is a symmetry of the Lagrangian. Each
of these three symmetries has its own conserved current. Therefore, the total number
electrons, muons, and tauons are conserved in QED separately, it constitutes a flavor
quantum number. This flavor quantum number is, like electric charge or fermion number,
an internal symmetry, and therefore flavor is an internal quantum number. QED is said to
respect flavor symmetry, where the flavors are uniquely identified by their mass. Formally,
this is a U(1)3 symmetry, i. e. a product of three independent U(1) groups.
If the three flavors would be mass-degenerate, this symmetry would be enlarged. Since
all particles have the same mass and the same electric charge, they become indistinguish-
42 4.12. Chiral symmetry
able, and they can also be exchanged. Thus, it is possible to perform rotations in the
internal flavor space. If the leptons would be described by real fields, this would be just
the usual orthogonal group of three-dimensional rotations, O(3). But because they are
complex, this group is enlarged to SU(3). The total flavor symmetry of mass-degenerate
QED is hence SU(3), or, when supplemented by the additional fermion number symmetry,
SU(3)U(1).
4.12 Chiral symmetry

The QED Lagrangian (3.7) exhibits one interesting additional symmetry if the masses of
all the fermions are set to zero, besides the then manifest flavor symmetry. This additional
symmetry emerges by the combination of a flavor or fermion number transformation and
an axial transformation. Axial transformations are a special property of fermions, and
there is no analogue for bosons of arbitrary spin. The combination of fermion number and
the axial transformation is the simplest case. It is mediate by multiplying every fermion
field by exp(i5 ), where is a real parameter, and 5 = i0 1 2 3 is a combination of
the -matrices. This can be shown using the fact that 5 anti-commutes with all . The
anti-fermion field is transformed by the corresponding hermitian conjugated phase factor.
This phase symmetry adds an additional U(1) symmetry to the theory, which is called
an axial symmetry UA (1). In addition, like the generalization of the fermion number
symmetry U(1) to the flavor symmetry SU(Nf )U(1) for Nf flavors, it is possible to en-
large the axial symmetry to an axial flavor symmetry, called the chiral symmetry. This
name stems from the fact that it turns out that it connects fermions with spin projec-
tions along and opposite to their momentum direction, i. e. of different helicities. Since
these projections yield classically a left-handed and right-handed screw8 , the name chi-
ral, Greek for handedness, is assigned. The total symmetry of the theory is therefore
SU(Nf )SUA (Nf )U(1)UA (1) for Nf flavors of leptons.
Of these symmetries, the axial symmetry is actually broken by an anomaly during
quantization. Non-zero lepton masses break the chiral symmetry, and the non-degenerate
lepton masses then finally break the flavor symmetry just to a flavor number symmetry.
Hence, little is left from the classical symmetries of massless QED.
This symmetry is therefore not realized in nature, where these masses are non-zero
and different. Furthermore, since the masses are large compared to the strength of elec-
8
Note that the question of what is left-handed or right-handed depends upon whether you look at a
screw from the top or the bottom. Since different conventions on how to look at a screw are in use, care
should be taken.
tromagnetic interactions, the symmetry is so baldy broken that it has almost no relevance
for QED. This will change later when discussing the strong nuclear interaction.
4.13 Generators
A final remark on symmetries is the notion of generators, as it appears throughout the
literature. It is a technically very useful observation that for a continuous symmetry, which
changes + , the associated Noether charge Q plays a special role. It can be shown
that
[, Q] ,
where additional constants of proportionality may or may not appear, depending on con-
ventions, and the indicates that this may be a commutator or anti-commutator, de-
pending on whether the involved quantities are bosonic or fermionic, respectively. As a
consequence, the charges Q are also referred to as the generators of the symmetry. In this
sense, the electric charge creates the electromagnetic (gauge) symmetry.
Chapter 5
Strong interactions
It was very early on recognized that QED cannot be used to describe atomic nuclei, since
QED can never sustain a bound state of the positive charge carriers which make up the
nuclei, the so-called protons with their mass of roughly 938 MeV, and electrically neutral
particles, the neutrons, with their just about 1.5 MeV smaller mass. Since also gravity was
not an option, there had to be another force at wok, which is called the strong nuclear force.
This became an ever more pressing issue, as it became experimentally clear that neither
protons nor neutrons can be elementary particles, as they have a finite extension, about 1
fm, and behave in scattering experiments like having a sub-structure. To understand the
details was, however, a rather complicated challenge.
The formulation of the strong interactions in its currently final form of quantum chromo
dynamics (QCD) is, in a sense, the latest addition to the standard model. Because of the
phenomenon of confinement, discussed below in section 5.8, it was much longer formulated
than experimentally established. The reason behind is that the strong nuclear force is, as it
names suggest, strong. It is therefore only in very limited, though relevant, circumstances
possible to use perturbation theory to calculate anything in QCD. This made theoretical
progress slow. The mainstay of QCD calculations today are in the form of numerical
simulations, which for many problems in QCD physics has become the method of choice.
Still, since computational power is limited, though such simulations being among the top
ten contender for computation time in the world, it is not yet possible to calculate complex
objects like even a helium nucleus. Still, for many purposes, many of the more fundamental
features of the theory remain an area of active research today. In this context, QCD and
theories which are (slightly) modified versions of QCD also serve as role models for generic
strongly interacting theories in particle physics.
44
Chapter 5. Strong interactions 45
5.1 Nuclei and the nuclear force

As noted, all atomic nuclei consists out of differing numbers of protons and neutrons, the
former having a positive electric charge of the same size as the electron and the latter
no electric charge at all. The simplest atomic nucleus is a single proton. However, single
neutrons are not stable, but decay in a few minutes, and are therefore only appearing
bound in nuclei. Because of electromagnetic repulsion it became quickly clear that the
nuclear force must be very much stronger than QED to create quite compact nuclei, not
much larger than the protons and neutrons, with more than one proton. It must also act
on different charges, as the neutron is electrically neutral.
One additional observation was made, when investigating nuclei and the nuclear force.
It was not only because of its strength very different from QED and gravity, but it was also
not of an infinite range. Though being much stronger than electrodynamics, it dropped
almost to zero within a very short range of a few Fermi.
How can such a short-range force emerge? One suggestion comes from the relativistic
description of bosons. The simplest example is the Klein-Gordon equation for a scalar
boson of mass m. It is obtained when replacing in the relativistic energy-momentum
relation (2.1) the energy and momentum canonically by it and ii .
( 2 + m2 ) = 0, (5.1)
for the wave-function . Investigating a static situation yields the solution

g m|~r|
(~r) = e , (5.2)
4|~r|
where g is an integration constant. Thus, the wave-function decays exponential towards
large distances, where the characteristic distance of propagation is given by the inverse
mass of the particle. Hence, a short-range interaction can arise if the exchange particles
are massive. This lead to the prediction of an exchange boson for the nuclear force of
order a few hundred MeV, given the range of the nuclear force of about a few fm. Such a
potential is called a Yukawa potential.
The interpretation of the integration constant g becomes clear when the zero-mass limit
is considered. In this case the wave-function becomes Coulomb-like, with g corresponding
to the fine-structure constant. Thus, also in the massive case g can be interpreted as a
coupling constant, which characterizes the strength of the strong nuclear force. It turns
out that characteristic values for it are about two orders of magnitude larger than for the
electromagnetic force.
Before taking this as more than a hand-waving motivation to look for a massive particle,
one should add a word of caution with respect to the Klein-Gordon equation. It is,
46 5.2. Mesons
strictly speaking, only valid if the energy scales involved are small compared to the mass.
Furthermore, it does not lead to a stable vacuum state, and therefore it can only describe
physics relative to an (undetermined) vacuum. Both sicknesses are only cured when moving
to quantum field theory, though this is beyond the scope of this lecture.
Nonetheless, the Yukawa potential yields the right idea. Instead of having massless
photons as force particles, massive particles must mediate the strong nuclear force. They
were indeed found in the form of the mesons.
5.2 Mesons
While the protons and neutrons are fermions with spin 1/2, the force carrier of the nuclear
force were identified to be actually bosons. The lightest of them are the pions with quantum
numbers J P = 0 , i. e. they are pseudoscalars. They come as a neutral one, 0 , and two
oppositely charged ones, . The range of the nuclear force is about 1 fm, which indicates
that the mass of the force carrier, according to the Yukawa potential (5.2), should have
a mass around 100-200 MeV. Indeed, the pions are found to have masses of 135.0 and
139.6 MeV for the uncharged and charged ones, respectively, and are thus much lighter
than either protons or neutrons. These pions are not stable, but decay either dominantly
electromagnetically into photons for the neutral one or like the neutron for the charged
ones. Their life-time is of the order of 108 seconds and 1017 seconds for the charged and
uncharged ones, respectively. Therefore the charged ones live long enough to be directly
detectable in experiment.
One of the surprises is that the neutral one decays into two photons, as usually pho-
tons are expected to couple only to electromagnetically charged objects. While this can
be thought of as a neutral pion virtually splitting into two charged pions, and then anni-
hilation under emission of photons, this is somewhat awkward. A more elegant resolution
of this will be given in the quark model below in section 5.4.
With these pions it was possible to describe the overall properties of nucleons, especially
long-range properties. At shorter range and for finer details it turned out that a description
only with pions as force carriers was impossible. This was resolved by the introduction,
and also observation, of further mesons. Especially the vector meson with a mass of 770
MeV, spin one, and a very short life-time of roughly 1024 seconds and the vector meson
with a mass of about 780 MeV, but with a 20 times longer life-time than the , play an
important role. This larger number of mesons is also at the core of apparent three-body
forces observed in nuclear interactions, which are, e. g., necessary to describe deuterium
adequately. In fact, many more mesons have been discovered, and some more will appear
later.
Describing how these various mesons create the strong nuclear force is in detail very
complicated, but it in principle can be systematically performed, e. g. in the form of the
so-called chiral perturbation theory. This will lead too far astray from particle physics
itself, and will therefore not be detailed here. What is, however, remarkable is that out
of nowhere appear several different mesons, all contributing to the nuclear force, and
actually all of them also affected by the nuclear force. Such a diversity of force carriers is
distinctively different from the case of QED, where only the photon appears.
5.3 Nucleons, isospin, and baryons

In the endeavor to find the carriers of the nuclear force, several other observations have
been made. The first is that most nuclear reactions show an additional approximate
symmetry, the isospin symmetry. This symmetry is manifest in the almost degenerate
masses of the proton and the neutron. Both particles can therefore be considered to be a
doublet of this new symmetry, which is similar to the flavor symmetry of QED, just much
less broken because of the very similar masses. It is furthermore found that also the three
pions fit into this scheme.
That once two particles and once three particles appear can be understood in the
following way, very similar to the case of angular momentum or spin in quantum mechanics.
There the rotational symmetry existed, and states could have any number of total spin l,
and then there existed a degeneracy of the states into 2(l + 1) states with different third
components. Similar for the isospin, a state can have a particular isospin value, and then
there are different charge states. In fact, isospin can be related to an SU(2) group once
more, just like spin in quantum mechanics, hence the name. The proton and neutron then
belong to the case of isospin I being 1/2, and thus there are two states with different third
components of the isospin I3 , which are essentially charge states. Such a situation is called
a doublet, and therefore proton and neutron are sub-summed under the name of nucleons.
Since the symmetry is only approximate, the different charge states are not degenerate.
This is similar to the Zeemann effect in quantum mechanics, only that the mass now takes
the role of the magnetic field.
The pions then belong to a state with I = 1, and thus three states. This is called
a triplet, with approximately the same mass. Generically, such collections are called
multiplets. In group theory, they correspond to different representations.
The natural question is then, whether there are higher representation, e. g. I = 3/2,
with four states. Based on the fact that isospin seems to be related to electric charge, since
48 5.4. The quark model
the different states of an isospin multiplet all have different charge, and that half-integer
or integer isospin corresponds to fermions or bosons, the properties of such a quadruplet
should be predictable. In fact, the relation
Q = I3 + S (5.3)
seems to hold so far, where I3 is the charge of the state, and S its spin1 . For the proton
and neutron, the assignments I3 = 1/2 and I3 = 1/2, respectively, yield the correct
result. For the three pions, the I3 assignments of 1, 0, and 1 reproduce the correct
electric charge for , 0 , and + .
However, this fails for the meson, which has spin one, and is an isospin singlet, but
in contrast to the rule (5.3) is uncharged. This can be remedied by replacing the spin S
by B/2, where B is the so-called baryon number,
B
Q = I3 + (5.4)
2
which is one for the nucleons and zero for the mesons. This is so far a phenomenological
identification, but will become quite relevant in section 5.4. Of course, the anti-particle of
the nucleons, the anti-proton and anti-neutron, carry negative baryon number.
According to this rule, it is possible to attempt to construct a quadruplet, having
four states with I3 = 3/2, 1/2, 1/2, 3/2. To get integer charges, it must then have a
baryon number, like the nucleons. These particles should therefore have electric charge
1, 0, +1, and +2, and corresponding anti-particles. These particles have been observed
experimentally, and again the different states have almost the same mass. They are called
, have masses of about 1232 MeV, and are fermions, as are the nucleons. However,
their spin is 3/2. Since both nucleons and baryons carry baryon number, there are called
commonly baryons, to distinguish them from the mesons. In fact, they are not the only
baryons, and many more have been found experimentally.
Together, mesons and baryons are denoted by hadrons, and are identified as those
particles directly affected by the strong nuclear force.
5.4 The quark model

The number of baryons and mesons found by now numbers several hundreds. Already
decades ago, when only a few dozens were known, it appeared unlikely that all of them
should be elementary. This was very quickly confirmed by experimental results which
1
The spin will already in the next step be replaced by a different quantum number.
showed that the proton had a finite size of about 1 fm, and Rutherford-like experiments
found that there are scattering centers inside the proton, which appeared point-like. How-
ever, in contrast to the atoms these constituents were not almost isolated in essentially
free space, but very tightly packed. Furthermore, while the neutron is electrically neutral,
it was found to have a magnetic dipole moment, a feature beforehand believed to be only
existing if there is an electrically charged substructure present.
This evidence together suggested that the elementary particle zoo could possibly be
obtained from simpler constituent and put into a scheme like the periodic table of chemical
elements, which originates from just three different particles.
Playing around with quantum numbers showed a number of regular features of the
hadrons. This gave rise to the quark model, where in the beginning two quarks were
needed to explain the regularities observed, the up quark and down quark, abbreviated
by u and d, as well as their anti-particles. Since both the bosonic mesons and fermionic
hadrons must be constructed from them, it requires them to be fermions themselves.
Since all of the hadrons have an extension, none of them can be identified with a single
quark, just like the periodic table does not contain a single proton, neutron or electron.
However, in contrast to the latter no free quarks are observed in nature, a phenomenon to
be discussed below in section 5.8.
The simplest possibility to construct then a hadron would be from two quarks. This
must be a boson, as two times a half-integer spin can only be coupled to an integer spin, and
therefore a meson. Since no free quarks are seen, the nucleons must contain at least three
quarks to get a half-integer spin. These considerations turn out to be correct. However,
they lead to the conclusion that quarks cannot have integer electric charges. This is most
easily seen by looking at the nucleons.
Scattering experiments identified that the nucleons have no uniform sub-structure, but
have a two-one structure, that is two quarks of one type and one quark of the other type.
Since it is found that the down quark is heavier than the up quark, the heavier one, i.
e. the neutron, should have two down quarks. This yields a composition of uud for the
proton and udd for the neutron. The only solution for the observed electric charges of
the proton and the neutron are then an assignment of 2/3 of the (absolute value of the)
electron charge for the up quark, and 1/3 of the (absolute value of the) electron charge
for the down quark. This consistently yields the required positive and neutral proton and
neutron charges. This also explains the magnetic dipole moment of the neutron. At the
same time, the baryon number of quarks must be 1/3 for up quark and down quark. This
implies also that the isospin of the up quark is +1/2 and that of the down quark is 1/2.
The pions are then constructed as a combination of a quark and an anti-quark, ud for
50 5.5. Color and gluons
for the 0 , i. e. the state of the 0 is

the + , ud for the , and a mixture of uu and dd
,
0
= cos |uui + sin dd
where is a mixing angle. This mixing angle can be experimentally or theoretically

determined. Experimentally, this is possible by using the different decay patterns for
different mixing angles. Theoretically, by calculating the left-hand side and the right-
hand side in different bases. However, in practice both possibilities are highly challenging.
An assignment of two quarks instead of a quark and anti-quark is not possible, as this
cannot give the required baryon number of zero.
Particles like the meson are then also combinations of a quark and anti-quark, but
where the quarks have relative orbital angular momentum, creating their total spin of one.
The , however, turns out to pose a serious challenge.
5.5 Color and gluons

At first glance, the appears simple enough. The double-positive state ++ is just three
up quarks, and with decreasing charge always one up quark is replaced by one down quark,
until reaching the with three down quarks. To obtain the observed 3/2 spin requires
to align the spin of all three quarks. Of course, it could be possible that there would
be a relative orbital angular momentum, but experimentally this is not found. In fact,
there exists an excited version of the with such an orbital angular momentum and total
angular momentum of 5/2, which is also experimentally confirmed.
And this is, where the problem enters. Since the is a fermion, its wave-function must
be totally antisymmetric. Since the spins are aligned and all three quarks are of the same
type in the ground-state, no wave-function can be constructed which is anti-symmetric.
Thus the existence of the appears to violate the Pauli principle at first sight. But this
is not so. Originally introduced to resolve that problem, and later experimentally verified,
another conserved quantum number is attached to quarks: Color. The wave-function can
then be anti-symmetric in this new quantum number, saving the Pauli principle and the
quark model at the same time.
Since this new quantum number of the quarks is not observed for the , or any other
hadron, the hadrons must all be neutral with respect to this new quantum numbers. For
the mesons, consisting of a particle and an anti-particle, this is simple enough, as just both
have to have the same charge. This is not the case for baryons. Assigning just positive
or negative charges, like the electrical charge, it is not possible to construct neutral states
out of three particles. Attempts to do so with fractional charges also do not succeed in
the attempt to make the proton and neutron color-neutral simultaneously. It is therefore
necessary to depart from the simple structure of the electromagnetic charge.
As a consequence, it is assumed that there are three different charges, suggestively
called red, green (or sometimes, especially in older literature, yellow), and blue. It is
furthermore assumed that not only a color and the corresponding anti-color is neutral,
but also a set of each of the colors is neutral. Then there are three quarks for each flavor:
red, green, and blue up quarks, and red, green, and blue down quarks, totaling six quarks.
A color-neutral baryon is then containing a quark of each color, e. g. a proton contains
a red and a blue up quark, and a green down quark. In fact, since the total charge of a
proton is zero, it is a mixture of any possible combination of color assignments to each
three quarks, which are consistent with neutrality and the Pauli principle. Similar, the
++ now consists of a red up quark, a green up quark, and a blue up quark.
This construction is rather strange at first sight, but it can formulated in a mathemat-
ically well-defined way. This will be done below in section 5.16, after collecting the other
ingredients of the strong interactions.
One other important ingredient, now that there is a new charge, is what mediates
the force between the charges. In electromagnetism it was the massless photons. It is
therefore reasonable to assume that there is also a mediator of the force between color
charges. These were indeed found, and named gluons. As the photons these are massless2
bosons with spin one. However, they differ from photons in a very important property.
While photons are only mediating electromagnetic force, they are not themselves affected
by it, since they carry no electric charge. But gluons carry color charge. In fact there are
8 different charges3 carried by gluons, and none of these eight are either the quark charges,
nor is their any simple relation to the quark charges. Especially, it is impossible to add a
single quark charge with any combination of the gluon charges to obtain a neutral object.
To achieve this, at least two quarks have to be added to one or more gluons.
Nonetheless, the idea of gluons has been experimentally verified, and they have been
identified as the carrier of the strong interaction, binding quarks into color-neutral hadrons.
The exchange of mesons to bind nucleons into nuclei can be viewed as a high-order effect
of the gluon interaction. This is similar to Van-der-Waals force, though the details are
different, as here not a color dipole moment enters, and the details are not yet fully
resolved. Still, the effect can be traced back to the gluons.
2
Again, the notion of mass is murky here, and the precise details are only becoming clear in a full
non-perturbative treatment of the corresponding quantum field theory. However, for most purposes the
notion of massless is sufficient.
3
One can think of four of them as particles and four as anti-particles, though this is not really how it
works, see section 5.16.
52 5.6. Chiral symmetry breaking
Hence, the combination of quarks, gluons, and colors, can explain the structure of
all known hadrons, similar to the periodic table. Unfortunately, the strong force binding
quarks by gluon exchange is not accessible using perturbation theory, at least when it
comes to describing hadrons. Its treatment is therefore highly non-trivial. Because of the
color, this underlying theory of hadrons is called chromodynamics, it quantum version
quantum chromodynamics, or QCD for brief.
5.6 Chiral symmetry breaking

No word has yet been said about how the masses of the hadrons relate to the masses of the
quarks. The mass of the proton is known very precisely to be 938.3 MeV, and the proton
to be 939.6 MeV. This implies that the mass difference between up and down quarks must
be tiny. The is somewhat heavier, about 1230 MeV. This can be understood as an
excited state, and is therefore heavier. Most ground state mesons have a mass of about
600 MeV or more. All this suggest a mass of about 300 MeV for the up quark and down
quark, with very little difference4 . But two mysteries appear. One is that the pions are
very light, just about 140 MeV. The second is that any attempt to directly measure the
quark masses yield consistently a mass of about 2.3(7) MeV for the up quark, and 4.8(5)
MeV for the down quark. Though the difference is consistent, the absolute values are
much smaller than the suggested 300 MeV from the nucleon properties.
The resolution of this puzzle is found in a dynamic effect of the strong interactions.
To understand it, recall that massless free fermions have a chiral symmetry, see section
4.12. I. e. they can be multiplied by a phase factor exp(i5 a ), where the a are the Pauli
matrices for two flavors of quarks. This symmetry is, at first sight, approximately valid
for up and down quarks as their intrinsic masses (often called current mass) is very small
compared to other hadronic scales, like the hadron masses. Thus, naively it is expected
to hold. This chiral symmetry has a number of particular consequences. E. g. it implies
that bound states of opposite parity, but otherwise identical content, should have the same
mass. But the mass splitting between such bound states for hadrons is large. E. g., the
parity partner of the nucleon is called the N(1535) and has, as its name suggests a mass
of 1535 MeV, about 50% heavier than the nucleons. This is much larger than expected
due to the explicit breaking of the chiral symmetry because of the small current masses.
Thus, chiral symmetry must be much stronger broken than just from the current quark
masses. The strong interactions must either spontaneously or explicitly break chiral sym-
4
Though this difference is crucial for making the proton lighter and thus (more) stable (than the
neutron), and therefore chemistry possible.
metry. Experiments show that at sufficiently high energies, many times the hadronic
scales, the quarks again behave as if they have only their current masses. Thus, the break-
ing must be spontaneous. This will be confirmed when writing down the field theory of
QCD in section 5.16. Hence, the strong interaction break chiral symmetry spontaneously.
This proceeds in the following way. The gluon interaction create a strong binding of the
quarks, which creates a so-called vacuum condensate huui + hddi 5. Since the condensate
is made from the quarks, any other quark will interact with it. Since the strong force is
attractive, otherwise protons, neutrons, or nuclei would not exist, this will slow down any
movement of a quark. Thus, quarks gain in this process inertial mass6 , and thus effectively
a larger mass. This additional mass is the same for up quarks and down quarks, and
approximately 300 MeV. This explains how the heavier mesons and the baryons gain a
mass of this size. It does not yet explain two other facts: The lightness of the pions nor
why this is no longer the case at high energies. These two questions will be answered in
the next two sections.
Before this, some remarks should be added. It is easy to wonder how the universe
should be filled with such a condensate, but no effect appears to be visible. Here, one
notes how everyday experience can be deceiving. Of course, the effect is visible, since
without it, all nucleons would be very much lighter, and so would be we, since almost all
our mass is due to nucleons. Thus, whenever we feel our mass, we feel the consequences of
this condensate. In fact, this condensate is responsible for essentially 99% of the mass of
everything we can see in the universe, i. e. suns, planets, asteroids, and gas, the so-called
luminous mass of the universe.
5.7 The Goldstone theorem

The lightness of the pions is a very generic feature of particle physics theories with spon-
taneous symmetry breaking. It is formulated in Goldstones theorem. To lay out this
theorem, it is helpful to investigate a very simplified model.
Take two scalar particles7 , which can interact with themselves. Such particles can either
be described by two scalar fields 1 and 2 or with a single complex field = 1 + i2 .
Such a theory has a classical Lagrangian of
L = U( ), (5.5)
5
The omission of the state generally denotes that the expectation value is taken with respect to the
vacuum.
6
That this also acts as a gravitational mass is not at all clear. Only the equivalence principle implements
this as a law of nature, an axiom.
7
With only one scalar field it is not possible to construct a simple example with a continuous symmetry.
54 5.7. The Goldstone theorem
where U describes the potential for the field, which can only come from interactions
between the two fields. The simplest (and in known particle physics so far exclusively
appearing) possibility to obtain a symmetry for this theory is that the potential depends
only the product .
To obtain the simplest example for a symmetry, take the potential
2 v 2 2 + 2
U( ) = + 2 ( )2 .
4 2 4v
The pre-factors, i. e. coupling constants, as well as the irrelevant constant term have been
chosen judiciously such that the result will be looking simple. This potential, as well as
the kinetic term, is invariant under the phase rotation exp(i). Therefore, this
theory has a global (U(1)) phase symmetry. It is a bit odd theory, as the quadratic term is
usually associated with a mass. This can be seen when determining the classical equation
of motion, which read
L L 2 2 2 2
0 = = + . (5.6)
2 2v 2
The equation of motion for the field is, up to conjugation, identical. Ignoring the
interaction term, this is just again the Klein-Gordon equation (5.1), which describes the
movement of a free scalar particle, if the sign of the second term would be positive. In
this case, it looks like the particle has an imaginary mass, what would be called a tachyon.
This is, however, not a problem. The sign of the quartic term is positive. Therefore, the
energy is bounded from below classically, and the theory remains stable. It is therefore
just an odd term in the potential energy.
To proceed, an interesting question is what the classical lowest energy state is. Since the
kinetic term is positive, any spatial or temporal variation would increase the total energy.
Hence, the state of lowest energy is necessarily a field 0 constant throughout space and
time, which minimizes the potential (5.6). This constant is found to be 0 = vei , where
is an arbitrary phase, i. e. all values of are a minimum. Thus, the solution manifold is
highly degenerate. This is a consequence of the symmetry: Any change in can be offset
by a symmetry transformation, without changing the physics.
One can proceed by specifying further. However, any choice is physically indistinct,
and therefore arbitrary. For further calculations keeping manifest is awkward, and
therefore in the following the explicit choice = 0 is made. Since the choice is arbitrary,
the symmetry is not violated. However, the symmetry is no longer manifest either. One
therefore speaks of hiding the symmetry, or a hidden symmetry.
To make the situation more transparent, the next step is to shift the field by its
vacuum expectation value, i. e. replace
(x) v + (x) + i(x), (5.7)
making now the complex structure manifest with the two real fields and . In this way,
fluctuations of the fields around the classical vacuum can be studied. Inserting this into
the Lagrangian (5.5) yields
2 3 2 2 2 2 2
L = + 2 2 + + + 2 2 + 2 4 + 2 4 . (5.8)
v v 2v 4v 4v
This Lagrangian shows now a number of very interesting features, which are very generic.
The first is that the two fields and behave differently. While there is a mass term,
now with the correct sign, for , giving it a mass of , there is no mass for . Pictorially,
one can think of as excitations which describe fluctuations out of the minimum, while
, which is orthogonal to the direction of the chosen vacuum, moves between the different
minima of the potential Since the vacua all have the same energy, this does not cost any
energy, and therefore the mode is massless. This is a generic feature of such situations,
and is known as Goldstones theorem. In a nutshell, it is the statement that there as many
massless particles as there are directions in which the minima are equivalent8 .
The second is that there are now many different interactions between the fields
and . However their couplings, i. e. their pre-factors, are not all different, but completely
determined by the original parameters. The reason is that the symmetry is just hidden. To
ensure that any symmetry transformation is still valid requires that the various interactions
cannot have arbitrary pre-factors, because otherwise it would no longer be invariant under
the symmetry transformation
v + (x) + i(x) ei (v + (x) + i(x)), (5.9)
where it should be noted that the vacuum solution v is also transformed accordingly. For
that to work out, it is necessary to keep track when changing from (5.5) to (5.8), which
occurrence of v stem from the original coupling constants in (5.5), and which from the
shift (5.7), since only the latter are affected.
If at any point a term is added to the Lagrangian, which violates the symmetry, the
symmetry becomes explicitly broken. The most obvious way is to add a mass term for
the field to the Lagrangian (5.8). Then, the Lagrangian is no longer invariant under the
symmetry transformation (5.9). Of course, this can be translated back into the original
8
The precise formulation is that there are as many massless particles as there are generators of the
symmetry group minus the number of generators of symmetry transformations after all possible remaining
choices have been made, here this is one minus zero, and thus one massless particles.
56 5.7. The Goldstone theorem
Lagrangian (5.5), where it takes the form of an additional quadratic term in the potential
(5.6) of type ()2 . The effect is essentially that the potential is tilted, and the vacuum
state has now a unique solution v, which has no longer an invariance. This gives also
a physical explanation for the mass: Since there are no degenerate vacuum solutions
anymore, any movement increases the energy.
This now illustrates how pions gain their small mass. The breaking of chiral symmetry
would lead to a number of Goldstone bosons. In case of two quark flavors9 , there will
be three Goldstone bosons, which are the three pions. These would be massless, if the
quarks would be massless. However, because of the small current mass of the quarks, the
symmetry is not exact, but rather explicitly broken. This gives the mass to the pions.
That the masses of the pions are still large compared to current masses of the quarks is a
dynamical effect. Approximately, the pion masses scale linearly with the current masses
of the quarks, the so-called Gell-Mann-Oaks-Renner relation, but the pre-factor is large.
Before continuing on, a few words about subtleties and semantics must be said. When
going to the quantum theory, quantum effects do what they always do, and they will mix
all the degenerate vacuum states, and no vacuum will be preferred. Therefore, the vacuum
state will exhibit a perfect symmetry, in contrast to the classical case. But the quantum
system also carries the seed of the classical physics within, as it is metastable. If any
arbitrarily small external perturbation, e. g. an infinitesimal mass for the from some
other physics process, arises, it will immediately have a unique vacuum state, in which
the symmetry is no longer realized. Hence, though strictly speaking the system without
external influence is perfectly symmetric, the presence of this metastability has lead to the
expression that the symmetry is nonetheless spontaneously broken.
In fact, even though the symmetry is exact, a full non-perturbative calculation shows
that the system has both an ordinary massive and a massless excitation, and hence the
most pertinent feature of the Goldstone theorem are realized even with the symmetry
present. Of course, in a perturbative calculation this will not show. Since perturbation
theory only permits very small deviations from the vacuum state, the particles will still
appear to be tachyons, as the relevant Lagrangian is (5.5). To cure this problem, one
can introduce a weak external perturbation to the theory, which prefers a single vacuum,
perform perturbation theory around this vacuum, i. e. using the Lagrangian (5.8) instead,
and remove the external perturbation at the end10 . Then, also in perturbation theory
the system exhibits a massive and a massless particle. Especially because of this trick,
which is extremely useful in the standard model, the exact notion of hidden symmetry is
9
It can be shown that the number of colors does not matter.
10
There are some subtleties associated with non-analyticities in this case, but in most cases they can
be dealt with.
nowadays very rarely used, and almost always the situation will be denoted by spontaneous
symmetry breaking.
5.8 Confinement and asymptotic freedom

This explains the light mass of the pion. It does not explain why the quarks appear
to become lighter at high energies. The reason is that the strong interaction work very
differently than electromagnetism or gravity. These two forces have the property that
the potential11 between two particles decays as 1/r, and thus becomes weaker the longer
the distance between the two particles is. Similarly, the force between two color-neutral
particles, the residual effect of the strong interactions due to meson exchange, even decrease
quicker because of the Yukawa potential (5.2) between them.
The situation for the interactions between two colored objects is drastically different. In
fact, this was one of the reasons why it took very long to identify the underlying theory for
the strong interaction, as its behavior is quite unique: The strong force becomes stronger
with distance. An appropriate modeling of the inter-quark force is given by the potential
4 s
V = + r, (5.10)
3 r
i. e. it has two components. There is a short-range Coulomb-like interaction, with the
strong fine-structure constant s = gs2/(4) defined by the strong coupling constant, sim-
ilar to the electromagnetic force. But there is a second component, which rises linearly
with distances, with a constant of proportionality . This latter contribution is of purely
non-perturbative origin, which also made it very hard to discover its theoretical origin.
That we do need such a potential for the strong interaction is an experimental fact. Only
that s and k are not unrelated can be understood so far, but why this is necessary is just
an observation.
This now explains why quarks appear to be lighter at high energies and thus shorter
distances. At long distances, the interaction is very strong, which also manifests itself
in chiral symmetry breaking and the condensation of quarks There the quarks are thus
behave as described, and are quite heavy. However, this interaction becomes weaker at
higher energies, i. e. quarks moving through the quark condensate at high enough energies
will no longer be slowed down by it. The quarks will thus behave as if the only have their
11
Note for the following that the notion of potential has no real meaning anymore when particles can
be created or destroyed, and it is necessary to pass to quantum field theory for a more precise description.
But for the present purpose the analogy is sufficiently good as to give a qualitative correct picture, and is
even quantitatively surprisingly good.
58 5.9. Hadronic resonances
current mass. This explains the difference of both masses. This effect that the strong
interaction becomes weaker at higher energies is called asymptotic freedom. Of course, the
remaining force is still as strong as the electromagnetic one, due to the classical Coulomb
potential in (5.10), but this is much weaker than at long distances, and therefore effectively
free. Moreover, quantum corrections reduce this interaction further: The running strong
coupling decreases at high energies, because the leading coefficient of the -function 0
in (3.9) is negative, with a scale of roughly 1 GeV. Hence, at a very high energy, the
electromagnetic force becomes actually stronger than the strong one, above roughly 1015
GeV.
The potential (5.10) also explains why no individual quarks are seen, at least to some
extent. Since the potential energy between two quarks rises indeterminate when increasing
the distance, one needs enormous amounts of energy when trying to separate them, much
more than available for any earthbound experiment for any experimentally resolvable
distance. This energy is stored in the gluonic field between the two quarks. Investigations
have shown that this field is highly collimated between the two quarks, it is like a string
connecting them both. Since the tension of this string is characterized to lowest order
by the parameter in (5.10), this parameter is therefore also called string tension. This
explains12 why colored quarks are not observed individually, but are confined in hadrons,
and the phenomenon is denoted as confinement.
However, in practical cases the energy limits of experiments is not a real concern. Long
before the relevant energy limit is reached, there is so much energy stored in the string
between the two quarks that it becomes energetically favored to convert it into hadrons.
Hence, when trying to separate the two quarks inside, e. g., a meson by pumping energy
into it, at some point it will just split into two mesons. This is called string breaking
or screening, and is a 100% efficient mechanism, i. e. there is no remaining quark. The
reason is that color, like electric charge, is conserved. Hence, overall the system must
remain color-neutral. Thus, there would be at least two colored objects left over, which
would again create immediately a string. This is also the mechanism why gluons are not
observed. When trying to separate gluons from a hadron, again a string is created, which
breaks and yields just additional hadrons.
5.9 Hadronic resonances

A consequence of the potential (5.10) is that for each hadronic state, there can be further
excited states. These are the hadronic resonances, and exist for both baryons and mesons.
12
Again, the true QFT version of this is much more subtle, and not yet fully understood.
E. g. the first excited state of the nucleon is the N(1440), with a mass of 1440 MeV,
as the name indicates. In total, about 28, experimentally more or less well established,
resonances have been observed for the nucleons, and likewise for the other hadrons.
An interesting observation, which can be understood on basis of the string picture
discussed in the previous section, is the presence of so-called Regge trajectories. A Regge
trajectory is that the mass squared of the resonances are roughly proportional to its total
angular momentum, i. e. the combination of spin and orbital angular momentum. This
has been experimentally observed before the inception of QCD, and has led to the first
formulation of a string theory as a theory of the hadronic interactions. Due to consistency
problems, and the lack of ability to describe other experimental observations, this approach
of a quantized string has been dropped later in favor of QCD.
All hadronic resonances are unstable, and decay in various ways. Most quickly are those
where the decay proceeds by splitting into hadrons, which usually involves the decay into
a lower resonance and some pions. E. g. the N(1440) decays in about 50-75% into a proton
or neutron and a pion, with a decay time of roughly (200 MeV)1 . Some resonances can
kinematically not decay into two hadrons, and they de-excite by emission of photons or
other particles. This happens on a much longer time scale. Such states are often considered
as excited states in contrast to the resonances. These are under the decay threshold into
purely hadronic final states. The much longer decay time is of course due to the fact
that the strong force is, indeed, the strongest one, and all processes subjected to it occur
very quickly. The only exceptions are those processes close to, but above, the hadronic
decay threshold. There kinematic reasons, the absence of a large phase space, statistically
suppress decays13 .
Th study of the decays of hadrons has been very useful in the understanding of the
strong force. As will be seen later in section 5.12, today decays of hadronic resonances
have also become an important window into the study of other processes.
5.10 Glueballs, hybrids, tetraquarks, and pentaquarks

When studying the quark model and the possible color charges, it becomes quickly clear
that three quarks or one quark and one anti-quark are not the only possibilities how to
create color-neutral objects. Tetraquarks, made from two quarks and two anti-quarks,
as well as pentaquarks, made from four quarks and one anti-quark, are equally possible.
There is indeed no a-priori reason why such bound states should not exist. However, the
13
Quantum interference effects can occasionally modify those statements, but they are rather good
guidelines for estimates of decays.
60 5.10. Glueballs, hybrids, tetraquarks, and pentaquarks
are not yet any unequivocally observed tetraquarks or pentaquarks. The reason for this
are due to two effects.
The first effect is mixing. E. g. for a tetra-quark, it is almost always possible to
construct a meson with the same quantum numbers, i. e. the same spin, parity, charge-
parity and electric charge. There is also the possibility to construct equally well a dimeson
molecule. One of the most infamous examples is the meson14 . It is a light neutral meson,
with quantum numbers compatible with the states uu, dd, uudd, uuuu, dd
dd,
+ , and
0 0 . Since it is a quantum state, it follows the quantum rules, which in particular
imply that all states with the same quantum numbers mix. It is therefore a superposition
of all these seven states. The question which of these states contribute most is highly
non-trivial. It can, in principle, be experimentally measured or theoretical calculated.
There is no really reliable way of estimating it. The results found so far indicate that the
combination of two pions is most dominant, it is therefore likely a molecule. For most
other states the two-quark components appears to be the dominant one. Similarly, almost
all baryons turn out to be completely dominated by the three-quark state. There are only
few cases left, where there is still some doubt.
The second possibility is to investigate one of those possibilities where the quantum
numbers of the tetraquarks cannot be created using a two-quark system. Such cases are
rare, but they exist. In principle, it would therefore be sufficient to just observe such a
state. Unfortunately, almost all of these states are highly unstable. They are therefore
experimentally hard to observe, and only very few candidates have been found so far,
though recently a few candidates have been identified, to be discussed later in section
5.12.
This problem becomes more complicated due to the gluons. Though it is not possible
to create a color-neutral state from a quark and a gluon, it is possible to combine a quark,
an anti-quark and one or more gluons to obtain a colorless state. It is similarly possible to
combine three quarks and a number of gluons to obtain a colorless state. Such states are
called hybrids. However, the gluons can at most add angular momentum, but no other
charges to the state. Therefore, there is always a state with the same quantum numbers,
but just made from quarks. Since adding a particle or orbital angular momentum to a
state usually increases its mass, these states are unstable against the decay to a state
with just the minimum number of quarks. Though these hybrids are therefore formally
admixtures to any state, it is essentially always a small one, and therefore hybrids are very
hard to identify both experimentally and theoretically.
The last class of states which can come into the mix are bosonic glueballs, which
14
The official name is f0 (500), though the historical name of meson still is commonly used.
combine only gluons to a colorless objects. The usual counting rules of the quark model
do not apply to them, but as a rough estimate even the simplest state is made out of four
gluons. Such states carry no electric charge, and most of them have the same J P C quantum
numbers as mesons, and therefore mix. However, there are some candidates, particularly
the so-called f(x) mesons, with x around 1500 MeV, which appear to have a large admixture
from glueballs. This is experimentally identified by the possible decays. Since gluons are,
in contrast to quarks, electrically neutral, decays into electrically neutral decay products,
except for photons, should be favored if there is a large glueball contribution in the state.
This has been observed, especially when comparing the partial decay widths of decays
to uncharged particles to the one to photons. However, even at best these state ar only
partially glueballs.
There are some glueball states which have quantum numbers which cannot be formed
by only quarks, at least not in the simple quark model. Unfortunately, all theoretical
estimates place theses states at masses of 2.5-3 GeV, and therefore far above any hadronic
decay threshold. They are therefore highly unstable, and decay quickly. Hence, there is
not yet any experimental evidence for them, though new dedicated searches are ongoing
or are prepared.
5.11 Flavor symmetry and strangeness

So far two quark flavors, up and down, have been introduced. As discussed in section 4.11,
this implies the presence of an approximate SU(2) flavor symmetry, since the masses of
both quarks are small. It was possible to describe all hadrons introduced so far using just
these two quarks with the quark model.
However, already before QCD was formulated, hadrons were observed, which do not
fit into this picture. The most well-known of them are the kaons K , K 0 , and K 0 ,
four mesonic hadrons of masses 494 MeV for the charged ones and 498 MeV for the two
uncharged ones. Most remarkably, these new mesons were more stable than those of similar
masses made from the two quarks inside the quark model.
The resolution of this mystery is that there are more than the two quark flavors nec-
essary to construct the proton and neutron. These additional quark flavors, which will be
discussed now and in the following two sections, do not occur in naturally observed atomic
nuclei15 but can be produced in natural or artificial collisions.
The quark to obtain the kaons in the quark model has been called the strange quark s.
Its current mass is 90-100 MeV, and therefore its full mass about 400-450 MeV. It follows
15
With the possible exception of the inner core of neutron stars, though this is not yet settled.
62 5.11. Flavor symmetry and strangeness
then that the kaons, with only 500 MeV mass, are also would-be Goldstone mesons,
obtaining their mass from the with the strange quark enlarged chiral symmetry breaking.
Since the strange quarks current mass is already not too small compared to usual hadronic
scales, the kaons are much closer to their expected mass of about 700 MeV in the quark
model obtained from a construction of a u or a d quark and one of the new s quarks.
Indeed, the s quark has an electric charge of 1/3, just like the d quark. The charged
kaons are therefore the combinations us, us, and the neutral ones ds and sd, explaining
their small mass difference, and their multiplicity. The Goldstone theorem actually pre-
dicts that there should be 8 Goldstone bosons. These are six. The other two are the
and mesons, which are made from ss combinations, and some admixtures from neutral
combinations of u and d quarks. They are therefore even heavier, the having a mass
of 550 MeV. Somewhat peculiar, the mass of the is much higher, about 960 MeV. The
reason is that the also receives mass from another source, the so-called axial anomaly.
The latter will be discussed below in section 6.11.
Besides the broken chiral symmetry, there is also the now enlarged flavor symmetry.
Since it involves up, down, and strange quarks, it is an SU(3) group. Since the quarks have
different masses, the group is not unbroken, but reduced to three counting symmetries,
i. e. U(1)3 . Hence, the individual quark flavors are conserved, but bound states with
differing quark content have differing masses. This conservation of quark flavor by the
strong interaction is also at the origin of the name strangeness. When the kaons were
discovered, the quark model was yet to be established. The kaons, and also baryons,
called hyperons, with a single or more strange quarks included, showed a different decay
patterns than ordinary hadrons, due to the conservation of strangeness. Thus, they did
not fit into the scheme, and were therefore considered strange.
The presence of the strange quark, which has an effective mass of about 400 MeV, and
thus close to the masses of the up and down quarks of 300 MeV, adds many more possible
combinations to the quark model, which all have very similar masses. Thus, there is a very
large number of hadrons with masses between 500 MeV, and roughly 3000 MeV, where the
states become too unstable to be still considered as real particles. In fact, the number of
states N turns out to rise exponentially with mass, N exp(M/TH ). This is a so-called
Hagedorn spectrum, where TH is called the Hagedorn temperature. The reason for the
name is that naively a system with such a spectrum has the property that at a finite
temperature, the Hagedorn temperature TH , an infinite number of particles is created,
and thus the system cannot be heated beyond this point. For the strong interactions,
the Hagedorn temperature is about 160 MeV. Of course, it is in practice possible to go
beyond this temperature. What happens is that at this point the quark substructure can
no longer be ignored, and this effect limits the number of states growth. This has originally
lead to the idea that at this temperature a phase transition has to occur, which signals
the change from a hadronic system to one were the quark substructure becomes more
important. However, the most reliable results to date rather indicate that it is a cross-
over. This transition plays nonetheless a role in the development of the early universe,
though a rather small one. This will be discussed in detail in section 5.18
The experimental and theoretical study of this transition is an important part of mod-
ern particle physics. Nowadays it is not restricted to the question of large temperatures,
but also small temperature and large densities. This is the situation which is encountered
during supernovas and in the interior of neutron stars and during neutron star mergers.
At theses densities, which become larger than the density of nuclei, nucleons can no longer
exist without starting to overlap with each other. Similarly to the high-temperature case,
such an overlap of nucleons is expected to change the properties of matter dramatically,
as the quark substructure can no longer be ignored. However, much less is known about
this situation, as it is both experimentally and theoretically much more demanding to
investigate. Some hope is provided by the possibility to use gravitational waves to probe
the interior of neutron stars during neutron star mergers, but this will crucially depend
on the developments in the next few years, whether this turns out to be correct. One of
the most interesting questions is, whether strange quarks play a role for neutron stars.
Another aspect were strange quarks play an important role are the quests for tetraquarks
and pentaquarks. Since strangeness is a conserved quantum number of the electromagnetic
and strong interaction, it is possible to construct states which do not have the quantum
numbers of an ordinary state, e. g. a meson with total strangeness of 2, which therefore
must contain two strange quarks, and cannot be a simple quark-anti-quark state, or a
baryon with strangeness -1, due to a single anti-strange quark, which therefore must be a
tetra-quark. Searches for such signatures are intensively pursued.
5.12 Charm, bottom and quarkonia

It appears at first rather surprising that there should be just one other quark, which has
the same electric charge as the down quark. This appears unbalanced, and another quark
with the electric charge of the up quark appears to be necessary. Indeed, this is correct,
and there is also a heavier copy of the up quark, which is called charm16 quark. However,
16
The name originates from the fact that it solves several experimental mysteries observed in the weak
interactions, to be discussed in chapter 6, and because at that time it appeared to complete the quark
picture.
64 5.12. Charm, bottom and quarkonia
while the strange quark has a rather similar mass, despite its larger current mass, as the
light quarks, the charm quark has not. Its current mass is 1275 MeV, and thus similar
to its effective mass of roughly 1600 MeV. As a consequence, hadrons involving a charm
quark are much heavier than hadrons containing only the lighter quarks. Furthermore,
though chiral symmetry is even more enlarged, its so badly explicitly broken that would-
be Goldstone bosons involving charm quarks have essentially the same mass as without
the breaking of chiral symmetry. The same is true for the flavor symmetry, and the only
remnant is that charm is again a conserved quantum number in both the electromagnetic
and strong interaction.
This conservation of charm has very interesting consequences. Of course there are
hadrons were only one of the quarks is a charm quark, which are called open charm
hadrons. The best known ones are the D mesons, with masses of about 1870 MeV mass
and having the structure of a single charm quark and either an up or down quark. These
are the lightest particles with a charm quark.
But there are also particles, especially mesons, which consists only of charm quarks. In
the meson case, where the total charm is zero if they consist out of a charm and an anti-
charm quark, these are called hidden charm. The latter states are particularly interesting,
because they show a very interesting mass spectrum. In fact, the lightest cc states have
a mass which is just below threshold for the decay into two hadrons with a charm quark
and an anti-charm quark each, the DD threshold. They can therefore not decay directly.
Of course, the charm and anti-charm quarks can annihilate. But because of how quark
and gluon color charges are arranged, such a process is substantial suppressed in QCD
compared to the decay with a production of an additional quark-anti-quark pair. Hence,
decays occur very slowly. Hence, these hadrons are extremely stable compared to hadrons
made from lighter hadrons, where the pions offer a simpler decay channel. Thus, these
meta-stable state carry the name of charmonia, which have masses of about 3 GeV, but
decay widths of around a few 100 keV.
Because of this fact, the charmonia states turn out to present a very good realization of
the possible states permitted by the potential (5.10). Similarly to the hydrogen atom, this
potential creates states distinguished by a main quantum number and orbital quantum
numbers. The most well-known state is the J/, at about 3097 MeV with a decay width
of 93 keV, which is a state with one unit of angular momentum. However, the ground
state of the system is the c , with a mass of 2984 MeV and a decay width of 320 keV.
That the ground state decays quicker is mainly due to kinematic effects from the angular
momentum. Simply put, the ground state is in an s-wave, and thus the wave-functions of
the two charm quarks have a large overlap. Thus an annihilation into photons is much
more likely than in the case with angular momentum, where the overlap of the wave
functions is much smaller. Right now about 8 states are known, which are below the DD
threshold, the heaviest the so-called (2S), with a mass of 3690 MeV and a decay width
of 303 keV.
These charmonia have been very instrumental in understanding the potential (5.10),
and thus the strong interactions. The very well-defined spectrum, which provides the
opportunity of a true spectroscopy, including many angular momentum states, permits a
much cleaner study than in case of the light hadrons, where the ubiquitous decays into
pseudo-Goldstone bosons make resonances decay very quickly.
However, not all of the states in this spectrum are easily explained within the framework
of the quark model and the potential (5.10). These are the so-called X, Y, and Z mesons,
with masses above the DD threshold, and some also with open charm. These states do
not fit into the spectroscopic scheme, and especially some may have quantum numbers,
which are not in agreement with a simple quark-anti-quark system. This is still under
experimental and theoretical investigation. However, it already shows that the simple
quark model is not also able to explain all hadrons.
With the charm quark, it may appear that everything is complete and symmetric.
However, nature did not decide to stop at the charm quark, but added another quark:
The bottom (or in older text beauty) quark. It is a heavier copy of the down quark, and
has therefore an electric charge of -1/3. Its mass is about three times that of the charm
quark, with a current mass of 4.2 GeV. It therefore introduces another quark flavor, but like
for the charm quark, this does not play any role for dynamical chiral symmetric breaking,
as the explicit breaking is far too large.
Other than the mass and the electric charge, the bottom quark behaves essentially as
the charm quarks. Especially, there is a rich spectroscopy with open and hidden beauty17 ,
the latter also called bottonium in analogy to charmonium. Similar to the case for the
charm quark, the lightest meson with open beauty is rather heavy, B and B 0 being 5.3
GeV. As a consequence, the bottonium spectrum has a large number of quasi-stable states,
the lightest being the b with a mass of 9.4 GeV and a decay width of roughly 10 MeV, the
playing the role of the J/ with a mass of 9.5 GeV and a width of 54 keV, and then even
15 states to the heaviest b (3P ) with 10.5 GeV observed so far. There are also heavier
states, including bottom versions of the X, Y, and Z mesons, which do not fit easily into
a simple quark model explanation. Thus, an even richer spectroscopy is possible, though
the production of bottonia in so- called beauty farms, requires more resources than for the
charmonia.
17
For the flavor quantum number the term beauty still survives.
66 5.13. Top
Of course, for both bottom quarks and charm quarks there exist also baryons, with
one or more of these quarks, also with both charm and bottom quarks. These are rather
complicated to produce, but have been observed, though baryons with multiple charm or
bottom quarks only very recently. These baryons are not as stable as the mesons, but are
still sufficiently stable that they production and decay take place so far apart from each
other, a few m, that both processes can be experimentally resolved. They play therefore
an important role to identify the production of charm and especially bottom quarks in
high-energy collisions (so called c- and b-tagging).
Together, charmonia and bottonia are usually referred to as quarkonia. Studying theses
states are also interesting for other reasons than to understand QCD. Because the J/ and
are very long-lived, they are very well suited for precision measurements. Furthermore,
as will become evident in section 6.5, it appears plausible that new phenomena will be
influencing heavy quarks stronger than light quarks. Searching therefore for deviations of
the decays of quarkonia, especially bottomonia, from the standard-model predictions has
become an important branch in experimental physics, which is also called flavor physics.
5.13 Top
With the introduction of the bottom quark the situation appears again as unbalanced as
with the introduction of the strange quark. This is indeed true, and the picture extended
by the top (or in old texts truth) quark. This is the last quark, which so far has been
found, though there is no a-priori reason to expect that there may not be further quarks,
and searching for them is a very active field in experimental physics.
The top quark is a heavy version of the up quark, and thus has an electric charge of
2/3. The fact which is really uprising about it, is its mass of 173 GeV. It is therefore forty
times heavier than the bottom quark, and this is the largest jump in masses in the quark
mass hierarchy. It is an interesting side remark, and a triumph of theory, that this mass
has been established within 10 GeV before the direct observation of the top quark, by
measuring other processes sensitive to the top quark very precisely, and then using theory
to make a prediction.
The enormous mass of the top quark makes it the heaviest particle detected so far. Due
to its large mass, it decays much quicker than the lighter quarks, with a decay width of 2
GeV. This is a consequence of effects to be introduced in section 6. Hence, the formation
even of short-lived hadrons with a top quark is almost impossible, and no hadron with a
top quark has so far been directly observed. Whether a quasi-stable toponium is possible
is not clear, but so far there is no experimental evidence for it. But, due to the mass,
it is also not trivial to produce large numbers of top quarks, and thus the study of top
quarks is rather complicated. Hence, the top quark is so far more of interest for its own
properties, particularly its mass, rather than for its relation to QCD.
5.14 Partons and scattering experiments

Due to confinement, quarks (and gluons) cannot be isolated as individual particles. This
even applies to the top quark though this is of little practical relevance as it decays so
quickly. This has quite important consequences for the experimental detection of quarks
and gluons, as it implies that only hadrons are ever observed.
This also implies that a hadron is, in fact, a complicated structure of the quarks
providing its quantum numbers, the so-called valence quarks, and many virtual particles.
The latter can be either also quarks, then called sea quarks to distinguish them from the
valence quark, and (also virtual) gluons. Collectively, the valence quarks and the virtual
particles inside a hadron are called partons, a term originating from a time as it was
not yet clear whether there are different particles inside a hadron, or just a single type.
Furthermore, since quarks are indistinguishable, there are also processes where a sea quark
and a valence quark are exchanged. Thus, valence quarks are not immutable.
The collective of all partons carries the total quantum numbers, like momentum, elec-
tric charge, spin etc., of a hadron. Especially, each parton carries a momentum fraction
x (called Feynman x) of the total hadron momentum. If there would be only the valence
quarks, each of them would have x = 1/3. Because of the virtual particles, the average
x for each particle is smaller, since there are more particles, though quantum fluctuations
permit also values of x larger than 1/3, up to 1, though rarely. Since gluons are massless,
they contribute the majority of particles at small x. As a consequence, many properties of
hadrons can already reasonable well be approximated when the sea quarks are neglected,
the so-called quenched approximation.
5.14.1 Fragmentation functions and jets

One of the most direct consequences of confinement is that the quarks and gluons are not
observed. In fact, when they are produced in a high-energy collision they behave, due
to asymptotic freedom, for a short time as quasi-free particles. But confinement kicks
in after time or distance scales of about or less than a fermi. Since the initial state has
been color neutral, the final state is also color neutral. Thus, any produced partons are
connected, in the string picture by strings, with all the other colored particles. While
the distances are small, the string is short and little energy is required. When, after the
68 5.14. Partons and scattering experiments
collision, the particles start to move apart, energy must be invested into the string, and it
breaks eventually. This produces a spray of hadrons. It is said that the partons hadronize.
In the rest frame of the produced quark or gluon, this hadronization will be isotropic,
if sufficiently many quarks and gluons are present that the color charge is roughly isotrop-
ically distributed. That is a rather good approximation, due to the light gluons, for most
cases. But in high-energy collisions the quarks and gluons are usually produced at high
energies. Hence, in the laboratory frame the produced hadrons will be highly collimated
along the direction of the original quark and gluon. Such a collimated spray of hadrons
originating from a single quark or gluon is called a jet.
This jet, in principle, carries all the information of the original quark and gluon, like
electric charge, momentum, invariant mass, and flavor. In practice, some particles may
still be not emitted very close to the original direction, i. e. not in the jet cone, and there
may be other jets in an event. Thus, such information is not always easy to reconstruct.
Still, by identifying such a jet, especially if it carries very much energy, it is possible to
infer information about the original quark or gluon. E. g. top quarks are identified by the
total invariant mass of the jet.
A further information can be obtained from which hadrons are present in the jet. Since
quarks and gluons have different flavors, masses, spins, and electric charge, they produce
different hadrons. The mathematical description of the probability for a hadron of type
i to emerge from a quark or gluon of type j is called a fragmentation function. Such
fragmentation functions are very hard to calculate, but can be measured experimentally,
to some extent.
5.14.2 Parton distribution functions

Even more important in actual experiments is the inverse question: How probable is it to
find inside the projectile hadrons (e. g. the protons at the LHC) a certain quark flavor or
a gluon with a given momentum fraction x of the total hadron momentum? The answer
to this question are given by the parton distribution functions Phi (x) for a hadron of type
h and a parton of type i with momentum fraction x. The practical importance of these
functions becomes immediately clear when considering some reactions which lead to a
given final state f . The actual cross-section is then a convolution18
XZ
hhf = dxdyijf Phi (x)Phj (y)(1 x y). (5.11)
ij
18
This is somewhat symbolic, and the indices i, j, and f include many further properties, like spins,
polarizations etc.
Hence, the actual cross-section gets contribution from all possible parton combinations
inside the projectile hadrons.
The calculation of the parton distribution functions, or PDFs for short, is highly com-
plicated. It is, especially for small values of x a non-perturbative question, and has not yet
been performed satisfactorily. Hence, they are usually parametrized using experimental
input, where theoretical considerations determine the parametrization. The experimental
results are most easily obtained when instead of two hadrons electrons and hadrons are
collided, as has been done at DESY in Hamburg with electrons on protons.
The importance of the PDFs cannot be underestimated, since they vary substantially
with x. Especially, except for the valence quark flavors they drop very quickly with x.
As a consequence, it is rather unlikely to find a parton with large energy fraction inside
a hadron. Since the cross-section for particle production are usually peaked around the
mass of the particle to be produced, see equation (2.4), the effectiveness of producing
new particles drops quicker with the energy than would be expected from just the energy
dependency of the elementary cross section. Hence, the parton luminosities, i. e. the
luminosity folded with the PDFs, is the actual figure of merit for a hadron collider. On
the other hand, this means that increasing the energy for a hadron collider can increase
the effectiveness faster than the naive energy scaling of the elementary cross-section.
PDFs have also a theoretical importance. Because of asymptotic freedom, it is often
possible to obtain perturbatively a rather good approximation to the elementary cross-
section ijf . The involved PDFs are then known experimental input, and the interesting
process can be obtained directly. That is the standard procedure used for most calculations
in high-energy experiments. The assumption is that (5.11) is valid, i. e. that the process
can be factorized into the PDFs and the elementary interaction. Because of asymptotic
freedom, this is a reasonable assumption so far as the fractions x and y yield energy
fractions for the corresponding partons much larger than 1 GeV. Since an integration is
performed over all x and y, however, this will only work if the product of the PDFs and
the cross-section becomes small at small x and y. For many cases of practical relevance,
this is still a good approximation. Otherwise, so-called factorization violations are found,
and this approach breaks down.
Formally, due to the presence of virtual particles, any elementary particle has to be
described in terms of a PDF. E. g., an electron is surrounded by virtual photons. There is
hence a PDF describing the probability to observe a photon with an energy fraction x of
the total electron momentum inside an electron, Pe (x). Of course, also quarks or gluons
can be found in this virtual cloud. However, since the electron is not strongly interacting,
the PDFs for non-partons can be often calculated to a reasonable accuracy perturbatively.
70 5.14. Partons and scattering experiments
Still, since especially at low energies also hadrons can be encountered inside an electron
or a photon, this is not completely possible. This is not an esoteric effect. E. g. in the
interaction of photons with nuclei the PDF P (x) describing the mutability of protons
into mesons plays an important, and experimentally verified, role.
5.14.3 Monte-Carlo generators
As is visible, the experimental description of collisions is usually performed in a three step

process, as far as factorization is viable. The first step is to obtain the initial particles for
the interesting process. These are described by the various PDFs of the initial projectiles.
The second step is to calculate the elementary process, usually performed perturbatively,
to a certain order in perturbation theory. For the LHC at this time most such hard
processes are calculated to orders between tree-level and NNLO. The third step is then to
transform the final state of the elementary reaction into the particles finally observed in
experiments. This involves usually first a decay cascade of the particles obtained in the
elementary process into electrons, positrons, muons, photons, and other particles stable
and weakly interacting on the scale of the experiments or into quarks and gluons. In the
latter case also the creation of jets and/or fragmentation must be calculated, to obtain
the final state.
While the initial state is rather simple, already the possibly involved intermediate
particles before the hard collisions can easily number dozens. Moreover, even at moderate
orders of perturbation theory, very many contributions appear. The final state is even
more complex, and can involve many final particles, and several jets - up to eight jets and
about a dozen quasi-stable particles are currently feasible for experiments.
To deal with this complexity requires computers. For many theories, the outer part
of the event, the creation and fragmentation of the particles entering into the elementary
hard process, is the same. Hence, this part can be automatized. Such programs, in which
only the elementary cross-section is used as an input, are called Monte-Carlo generators.
Current examples are publicly available codes like Sherpa, Pythia, Herwig, or Whizard,
and can be downloaded e. g. at www.hepforge.org.
Even the elementary cross-sections can be calculated in many cases automatically,
especially at tree-level. In these cases only the Lagrangian has to be provided, to permit
an automated calculation from the initial to the final state.
5.15 Some algebra and group theory

Up to now, QCD has not been specified. This will be done in the next subsection 5.16.
The first question to be answered, however, is how to to implement the color structure
discussed in section 5.5 into a theory. The fact that three colors should create a neutral
state, but not any other combination without anti-colors, already tells that it is impossible
to do so just with real numbers. Vectors can indeed be arranged to have such a property.
Since in two dimensions at most two vectors can be linearly independent, the space must
be three-dimensional, to have three independent quark colors. But also the eight gluons
colors, and the anti-colors must be fitted in.
It turns out that there is a unique mathematical structure which offers all these prop-
erties. This is the so-called Lie algebra su(3), and its associated Lie group SU(3). These
are special cases of so-called Lie algebras and Lie groups. These are in turn special cases
of symmetry groups and algebras, as introduced in section 4.1. In particle physics, some
theory of these algebras and groups is inevitable, as it is not only necessary for the strong
interactions, and QCD, but also for the weak interactions. In this brief section, the most
basic concepts for Lie groups and algebras will be given. Much more group theory than
the following is usually not necessary, though quite useful.
In fact, Lie groups are ubiquitous in particle physics. Examples have already been
encountered in the context of flavor in section 4.11 and gauge symmetry 3.4, but there the
details were not yet necessary.
To understand the concept, return to QED. QED implemented gauge-fields and matter
which were related by the gauge-transformation to the u(1) algebra and the U(1) group.
Thus, the fields could be regarded as tensor products of a function of space-time times
an element of the group. To deal with QCD, it will turn out that it is only necessary to
replace this u(1) by a different group, the aforementioned su(3), which is able to faithfully
represent the concept of color. To proceed requires more insight into this group structure
than for the flavor, and therefore some basic elements of the mathematical foundations of
non-Abelian Lie algebras, which will be introduced before formulating QCD.
Note that the fact that such mathematical structures describe particle physics is an
assumption, which must be experimentally supported. This was indeed possible for the
standard model of particle physics.
The basic element in this representation will be to represent a quantity carrying a
gluon-like color charge by the base vectors of a particular algebra. Such base vectors are
called generators of the algebra. The type of algebra needed in particle physics are so-
called Lie algebras, G. Hence, if there should be N independent gluon-like charges, there
must be N independent base vectors a with a = 1...N and N = dim G, and the Lie
72 5.15. Some algebra and group theory
algebra must therefore be N-dimensional.

The defining property of such an algebra are the commutation relations
a b
, = 2if abc c . (5.12)
with the anti-symmetric structure constants f abc . Noe that sometimes the factor of 2 is
included in the f abc . This algebra implies already that everything is non-commutative,
hence the name of non-Abelian Lie algebra. These structure constants fulfill the Jacobi
identity
f abe f cde + f ace f dbe + f ade f bce = 0. (5.13)
These base vectors have to be further hermitiean, i. e., a = a . Note that the position
at top or bottom (covariant and contravariant) of the indices is of no relevance for the
Lie algebras to be encountered in the standard model, but can become important in more
general settings: For Lie algebras, the metric is just ab .
Such a Lie algebra can be represented, e. g., by a set of finite-dimensional matrices.
An example is the su(2) algebra with its three generators, which can be chosen to be the
Pauli matrices,
!
0 1
1 =
1 0
!
0 i
2 =
i 0
!
1 0
3 = .
0 1
Furthermore, to each algebra one or more groups can be associated by exponentiation, i.
e.,
a
a = ei , (5.14)
provides base vectors for the associated group, which are by definition unitary and thus
1
a = a . For su(2), these are again the Pauli matrices, generating the group SU(2).
However, the relation 5.14 is not necessarily a unique relation, and there can be more than
one group representation. E. g., for su(2) there are two possible groups related to the
algebra by the relation (5.14), the group SU(2) and the group SU(2)/Z2 SO(3), where
matrices which differ only be a negative unit matrix are identified with each other.
Because of the exponential relation, a generic group element exp(ia a ) with real
numbers a can be expanded for infinitesimal a as 1 + ia a . Thus the algebra de-
scribes infinitesimal transformations in the group. This will play an important role when
introducing gauge transformations for non-Abelian gauge theories.
There is only a denumerable infinite number of groups which an be constructed in

this way. One are the N-dimensional special unitary groups with algebra su(N), and
the simplest group representation SU(N) of unitary, unimodular matrices. The second
set are the symplectic algebras sp(2N) which are transformations leaving a metric of
alternating signature invariant, and thus are even-dimensional. Finally, there are the
special orthogonal algebras so(N), known from conventional rotations. Besides these,
there are five exceptional algebras g2 , f4 , and e6 , e7 , and e8 . The u(1) algebra of Maxwell
theory fits also into this scheme, the u(1) group is the special case of all f abc being zero,
and the algebra being one-dimensional. This is equivalent to so(2).
The two algebras relevant for the standard model are su(2) and su(3). The su(2)
algebra has the total-antisymmetric Levi-Civita tensor as structure constant, f abc = abc
with abc = 1. The algebra su(3) has as non-vanishing structure constants
f 123 = 1

3
f 458 = f 678 =
2
1
f 147 = f 156 = f 246 = f 257 = f 345 = f 367 = ,
2
and the corresponding ones with permuted indices. There is some arbitrary normalization
possible, and the values here are therefore conventional.
From these, also the generators for the eight-dimensional algebra su(3) can be con-
structed, the so-called Gell-Mann matrices,

0 1 0 0 i 0 1 0 0
1 = 1 0 0 2 = i 0 0 3 = 0 1 0

0 0 0 0 0 0 0 0 0

0 0 1 0 0 i 0 0 0
4 = 0 0 0 5 = 0 0 0 6 = 0 0 1

1 0 0 i 0 0 0 1 0

0 0 0 1 0 0
7 8 1
= 0 0 i = 0 1 0 . (5.15)

3
0 i 0 0 0 2
In general, there are N 2 1 base vectors for su(N), but the dependency for the other
algebras is different. For the sake of simplicity, in the following only the expressions for
su(N) will be given.
Generators, which are diagonal as matrices, and therefore commute with each other,
are said to be in the Cartan sub-algebra or sub-group of the algebra or group, respectively.
74 5.16. Yang-Mills theory and QCD
For su(2), this is only one generator, for su(3) there are two. This also gives the rank of
the group.
These lowest-dimensional realization of the commutation relations is called the fun-
damental representations of the algebra or group. Since the commutation relations are
invariant under unitary transformations, it is possible to select a particular convenient
realization. Note, however, that there may be more than one unitarily inequivalent fun-
damental representation. E. g., for su(3) there are two, called fundamental and anti-
fundamental, while there is only one for su(2).
It also possible to give representations of the algebras with higher-dimensional matrices.
The next simple one is the so-called adjoint representations with the matrices
(Aa )ij = if aij ,
which are three-dimensional for su(2) and eight-dimensional for su(3). There are cases in
which the fundamental and the adjoint representation coincide. This can be continued to
an infinite number of further representations, which will not be needed here, though they
may appear in the context of beyond-the-standard-model physics.
This completes the required group theory necessary to describe the standard model.
5.16 Yang-Mills theory and QCD

It is now possible to return to QCD. It turns out that three quark colors (and three anti-
quark colors) as well as eight gluon colors are precisely related to group SU(3), i. e. the
group of unitary three-by-three matrices with unit determinant. The three quarks can
then be assigned to live in the corresponding three-dimensional complex vector space, the
anti-colors in the corresponding hermitiean conjugated space, and the eight gluon colors
can be associated with the eight of the nine base matrices, the ninth being the unit matrix.
The eight non-trivial base matrices are the Gell-Mann matrices a , given in (5.15).
More mathematically, following and generalizing section 3.4 for QED, the gluon fields
live in the corresponding algebra, and are connected to the generators, and thus belong
to the adjoint representation. The quarks and anti-quarks belong to the fundamental and
anti-fundamental representation, respectively19 .
Is it thus sufficient to just generalize the QED Lagrangian (3.6)? The answer to this
is both yes and no. The quarks are now vectors in color space, i. e. the spinors carry an
19
The actual group in the standard model is furthermore SU(3)/Z3 , that is all elements divided out
which commute with every other elements, i. e. the center of the group, except the unit matrix. This is
of no relevance here.
additional color index like i . A gauge transformation acts now as a rotation in this color
space, and is thus a matrix G as
g

i = Gij j = exp i a a j ,
2 ij
where the i are space-time-dependent functions, and g is playing the same role as the
electric charge, and is thus a coupling constant. The corresponding covariant derivative to
form a gauge-invariant Lagrangian has to take this matrix structure into account, yielding
g
Dij = ij i Aa ( a )ij , (5.16)
2
where the Aa are now the eight gluon fields. These have to transform under an infinitesimal
gauge transformation like
1
Aa = Aa + Dab b , (5.17)
g
Dab = gf abc Ac .
ab
(5.18)
A finite transformation takes the form

i a Aa = G a Aa G1 + ( G)G1 ,
were the combination Aa a is often abbreviated as just A . In the infinitesimal version

(5.17) a second covariant derivative appears (5.18), where additional constants f abc appear.
These are just the structure constants of SU(3) from the defining algebra (5.12), and are
therefore related to the Gell-Mann matrices as
a b
, = 2if abc c ,
forming the algebra su(3).

Replacing the spinors by color vectors of spinors in the QED Lagrangian (3.7), and the
flavors by the six quark flavors, is already sufficient to obtained the quark part of the QCD
Lagrangian. However, it turns out that just taking the same field-strength tensor as in
QED (3.5) with replacing the fields by their generalization with an index is not sufficient.
In fact, in this case the Maxwell term (3.4) is not gauge-invariant.
This can be solved by introducing a more complicated field strength tensor
a
F = Aa Aa + gf abc Ab Ac . (5.19)
It is a straightforward, though tedious, calculation to show that with this replacement

the Maxwell-term (3.4) becomes gauge-invariant. However, in contrast to QED, this field-
strength tensor in itself is still not gauge-invariant, but trF F is.
76 5.16. Yang-Mills theory and QCD
QCD is therefore described by the Lagrangian

1 a X i
L = F Fa + f (iDij mf )fj . (5.20)
4 f
This theory has a number of further remarkable features, which an be read off this La-
grangian.
The first is that the interaction between quarks and gluons is given by a vertex very
similar to the QED one, it is just adorned by a Gell-Mann matrix. Since the interaction
potential between quarks (5.10) is very different than the Coulomb potential, this implies
that there must be substantial radiative corrections. In fact, since there is a qualitative
change, these must be of non-perturbative origin.
Second, in QED it is in principle possible that each flavor has its own electric charge.
This is not the case here. Due to the appearance of the second term in the field-strength
tensor (5.19) the coupling constant in the gauge transformation in the gauge field (5.17)
is fixed. But the covariant derivative (5.16) will then only yield the necessary cancella-
tions between the transformations of the gluons and the quarks if the coupling constant
appearing in it is the same as for the gluon fields. Thus, every flavor has to couple to the
gluons with the same coupling constant g. This feature, which arises from the underlying
group structure, is called coupling universality, and has been experimentally verified.
Third, since the field strength tensor (5.19) is not gauge-invariant, so are not the
chromoelectric and chromomagnetic fields, and then neither can be the chromo version
of the Maxwell equation nor the color current. Hence, color and colored fields are not
gauge-invariant, and therefore cannot be observable. Because of confinement, as described
in section 5.8, this was not expected. But this makes confinement a necessity. Still, this
leaves open the question why confinement operators in the way it does, with the potential
(5.10), and not in any other way.
Fourth, the appearance of a quadratic term in the field-strength tensor (5.19) implies
that there are cubic and quartic self-interactions of the gluons with each other. This
coupling strength, due to the underlying group structure, are uniquely fixed, again a
consequence of coupling universality. But this also implies that gluons, in contrast to
photons, are not ignorant of each other. A theory with only gluons in it is therefore not a
trivial theory, but an interacting one. In fact, it turns out to be a strongly interacting one,
exhibiting features like confinement or the formation of glueballs. This reduced version of
QCD is called Yang-Mills theory. It is sometimes also denoted as quenched QCD, in the
sense that any conversion of gluons in quarks (and thus sea quarks) is fully suppressed, i. e.
quenched. It is possible to calculate in this reduced theory also approximately quantities
like hadron masses. They are then found to be roughly of the same size as in (full) QCD,
indicating that most of the physics in QCD resides in the gluon sector.
Fifth, a careful investigation of the vertices shows that the gluon-self-interaction and
the quark-self-interaction are opposite. While quarks tend to reduce the interaction, i.
e. they screen color charges, gluons tend to increase them, they anti-screen it. The ex-
perimental consequences of this fact actually lead originally to the discovery of the group
structure of QCD, as this behavior is almost exclusive to theories with a non-Abelian Lie
group as gauge group. Because anti-screening dominates this leads to the difference be-
tween QED and QCD, especially the feature of asymptotic freedom discussed in section
5.8. In the terms of section 3.8, the coupling does not become stronger but weaker with
energies.
Especially, it is possible to calculate the function (3.8) also of QCD. To leading
perturbative order only the first Taylor coefficients is relevant, which takes for QCD the
value
11 3
0 = Nc Nf = 2,
3 2
where Nc = 3 is the number of colors and Nf = 6 is the number of flavors. The value is
positive, in contrast to QED. Thus, because of the formula for the running coupling to
leading order (3.9), the interaction strength in QCD diminishes with increasing energy, a
manifestation of asymptotic freedom. The scale turns out to be roughly 1 GeV, and the
size of the coupling at about 2 GeV is 0.3, to be compared with the one of QED being
about 0.0075. QCD is thus indeed much stronger interacting at every-day energies than
QED.
However, at energies of the order of the proton mass, the perturbative coupling diverges,
and a Landau pole arises. In this case, this clearly signals the breakdown of perturbation
theory. In fact, even a value of 0.3 is already shaky for perturbative calculations, and
strong deviations from perturbation theory have been encountered up to 25 GeV. Thus,
only if all involved energy scales are in the range of (a few) 10 GeV or more QCD can be
treated perturbatively.
However, it is possible to treat QCD at low energies also non-perturbatively. There
are various methods available. The most successful one is arguably numerical simulations,
which confirms that the QCD Lagrangian (5.20) describes objects like the proton and
other hadrons. QCD is thus, to the best of our knowledge, the correct theory of hadrons,
and hence of nuclear physics.
To obtain the full Lagrangian of QCD and QED, it is just necessary to add the QED
Lagrangian to the QCD Lagrangian (5.20), and modify the covariant derivative of the
quarks to include the electromagnetic interaction as
Dij = ij igAa (a )ij ief A ,

78 5.17. Cosmic radiation
where now the electric charge depends on the flavor, and is either 2/3 or 1/3 in units
of the electron charge. This ratio of the electron to quark charge is actually very well
established experimentally, though why it is the case is unclear. However, as will be
discussed in section 6.11 it is necessary for the formulation of the standard model in its
current form.
5.17 Cosmic radiation
Before adding further elements to the standard model it is worthwhile to discuss two
examples where QCD (and partly also QED) plays an important role, which are usually
not directly associated with particles physics.
One of them is cosmic radiation. There are many sources in the universe which create
high-energetic particles. These are leptons and photons, but also protons. Electrically
neutral particles can be most easily tracked back to their origin, as their paths are not
changed due to galactic or intergalactic magnetic fields, while protons are affected. These
cosmic particles can have energies many orders of magnitude larger than created by any
experiment. Energies of up to almost 109 TeV have been observed20 . Possible sources may
be active galactic nuclei.
When such a high-energetic particles hits earths atmosphere, it reacts with the atoms
and atomic nuclei. As a consequence, a jet, called in this context a cascade, of hadrons
and electrons, muons, and photons is created, which an be observed with observatories on
earth. To be able to decide whether the original particle was charged or uncharged requires
to understand the evolution of this jet. This is a complicated problem, as it requires to
understand how the various particles interact and are produced. But this is decisively
important, so that it becomes possible to understand whether the direction of the cascade
could have been influenced by magnetic fields, and thus whether its origin is meaningful
or not.
Similar to the Monte Carlo generators of section 5.14.3 for particle collisions, this can be
usually only done by simulations. This is an important task for astrophysical observation,
and our knowledge of the universe.
20
Such very high-energetic particles are only observed extremely rarely, at rates which can be as low as
one per km2 and year, and are thus not useful to test particle physics.
5.18 The QCD phase diagram

As has already been noted in section 5.11, aspects of QCD play an important role for
the physics of neutron stars. In fact, a neutron star can be considered as a gigantic,
stable nucleus, though this is an oversimplification. Since the density inside a neutron star
increases from the outside to the inside, the physics will change. While the outer part
is indeed essentially a nuclear system, the situation in the interior is not clear. Because
QCD is so strongly interacting, it is very hard to treat in such an environment, and even
numerical simulations fail21 . It is therefore still an open field what actually occurs near
the core of a neutron star, and whether the state of matter there is still just nuclei, or
whether other hadrons, including strange ones, play a significant role.
The situation in neutron stars is only a part of a wider topic, the QCD phase diagram,
i. e. what is the state of matter at densities of similar or larger size than in nuclei and
temperatures of size of hadron masses.
Experimentally, these are very hard to address questions. For neutron stars, possibly
gravitational wave astronomy, as remarked in section 5.11, together with the X-ray septum
of neutron stars as a function of time, as well as their sizes and masses, will be the only
available experimental input for a long time to come.
At much smaller densities, the situation becomes essentially the one of nuclear physics.
here, experiments with nuclei, especially collisions, can be used. In this way, it was possible
to find that there is a phase transition between a gas of nucleons and (meta-)stable nuclei
at roughly the density of nuclei, which permits to form nuclei. This phase separation
persist for a few MeV in temperature, up to about 15 MeV, where it ends in a critical
end-point. Thus, both phases are not qualitatively distinguished, just like in the case of
water.
From the point of view of particle physics, the involved energy scales of nuclei are very
small, and the distinction between the liquid and gaseous phases is essentially irrelevant.
Thus both phases are not regarded as different. This common phase is denoted as the
hadronic or vacuum phase, the latter as for all practical purposes the phase consists of far
separated hadrons separated by vacuum.
There are now several interesting directions to move on. Usually, they are signified by
21
The reason is actually not the strongness of QCD. It is rather a problem which has to do with the
distinguishability of particles and anti-particles, and how this can be treated if there are more particles
than anti-particles. For some theories, like QCD, but there are also such systems in nuclear physics and
solid-state physics, this entails algorithmic problems, the so-called sign-problem, which prevents so far
the development of efficient simulation algorithms, as the computation time scales exponentially with the
system size. It is a technical problem, and not a physics problem.
80 5.18. The QCD phase diagram
temperature and the baryo-chemical potential, i. e. the chemical potential distinguishing

baryons and anti-baryons. Thus, zero (baryo-)chemical potential denotes a situation in
which there is the same number of baryons and anti-baryons, which includes the possibility
of none of either type.
When neutron stars are interesting, the axis with zero temperature and finite baryo-
chemical potential is most interesting. Neutron stars do have some temperature, but it
is of the order of one MeV, and even during their formation in a supernova explosion
or during a merger it does not exceed 10-20 MeV. On hadronic scales, this is essentially
negligible. Thus, to good approximation it is a movement only along the baryo-chemical
potential direction. Starting from the vacuum, i. e. zero baryo-chemical potential, it can
be shown exactly that nothing will happen before reaching a chemical potential of roughly
a third of the nucleon mass. This is due to some analyticity properties of the free energy
as a function of the masses of the lightest baryon, and the fact that it is made out of three
quarks. This feature is called the silver-blaze feature. However, any small temperature
will change this, and it is then only approximately true for chemical potentials much larger
than the temperature, but smaller than this silver-blaze point.
After this point soon the nuclear liquid-gas transition is encountered. After this exper-
imentally established point, as noted above, no fully reliable results are available. There
are some reasons to believe that at least one further phase transition will be encountered,
though even this is not sure. It is furthermore unclear whether this will happen at densities
sill relevant for a neutron star, or significantly above it. However, model calculations, i.
e. calculations using simplified versions of QCD, as well as comparisons to other theories
which are similar to QCD, indicate that there could even exist many different phases, some
amorphous, and some crystalline, in which besides the nucleons also other hadrons, like
pions, kaons, and hyperons, may play a role.
The situation is much better regarding zero chemical22 potential. This situation is
relevant in the early universe, where though all matter in the universe is already present,
and there is thus a sizable amount of baryons, the temperature is high enough to thermally
produce baryon-anti-baryon pairs, reducing the chemical potential very close to zero.
This situation is good accessible experimentally by high-energy heavy-ion collisions,
e. g. at the LHC with up to 2.4 TeV kinetic energy per nucleon for lead nuclei. In such
22
As in most cases only the baryo-chemical potential is present, the word baryo is dropped. The only
other chemical potentials of relevance in most cases are an isospin-chemical potential, which gives the
difference between up and down quarks, and is therefore relevant in neutron stars, though it is even
then only small because of the close similarity of up and down quark, strange-chemical potential, which
determines how many more strange than anti-strange quarks appear, and electro-chemical potential, which
indicates the presence of electrons.
an experimental setup, temperatures as high as 600-700 MeV with almost zero chemical
potential, despite the 416 nucleons in the original nuclei, can be achieved. Furthermore,
this situation poses no serious problems to numerical simulations. Hence, the knowledge
of this axis is rather good.
It turns out that the physics depends significantly on the masses of the up and down
quarks. Though this is also suspected for the remainder of the phase diagram, it is ev-
ident in this case. If both quarks are very heavy, there is a first-order phase transition
at a temperature of about 250-300 MeV. As the quark masses become lighter, this tem-
perature decreases, and the transition becomes a rapid cross-over at a temperature23 of
about 150-160 MeV, the aforementioned Hagedorn temperature. This is the situation for
up and down quark masses observed in nature, the physical masses. The mass of the
strange quark influences the precise values of the temperatures, but does not provide any
qualitative influence, and the heavier quarks have even less relevance. If the quark masses
are decreased further, the temperature still drops a little bit. More importantly, at some
point the cross-over turns again into a phase transition, this time of second order, where
it stays until zero quark masses.
What happens can be understood already in a simple picture. Temperature is classi-
cally nothing more than the kinetic energy of particles. In a quantum theory, temperature
is just energy, which can also be converted to new particles. This will be exponentially
suppressed with the mass of the created particles. Hence, the lightest particles will be
most copiously produced. In QCD, these are the pions. These particles will have large
kinetic energies, and will rapidly and repeatedly collide. Because of the asymptotic free-
dom of QCD, these scatterings will mainly be dominated by hard partonic scatterings, and
thus almost perturbative. Thus, QCD becomes essentially as it behaves at high-energies.
Especially, this implies that the effects of chiral symmetry breaking become reduced, and
the quarks lose their effective mass, though not their current mass, at the phase transition
or cross-over. In fact, in the limit of zero quark mass, the second order transition becomes
a symmetry transition where chiral symmetry becomes restored. At the same time, since
most collisions are hard and partonic, excited states become very unstable and in most
cases it does not matter anymore that quarks are confined into hadrons. They act effec-
tively as if they would no longer be confined. Thus, one speaks also of a deconfined phase,
and calls the transition a deconfinement transition. However, since it is a cross-over, it is
clear that qualitatively nothing has changed, but quantitatively it is a completely different
situation.
23
There is no unique definition of a cross-over temperature. This is just the temperature were most
changes occur.
82 5.18. The QCD phase diagram
As a consequence of this dominance of the partonic degrees of freedom and asymptotic

freedom actually the high-temperature thermodynamic behavior of the theory is essentially
that of a free gas of quarks and gluons, a so-called Stefan-Boltzmann gas. The reason is
mainly that the hard processes contribute to the free energy like the temperature to the
fourth power, while all other effects contribute like the cube of the temperature, or even
less. Hence, they become for thermodynamic bulk properties, i. e. extensive properties,
irrelevant. Still, there are certain observables which are sensitive to non-trivial effects.
Furthermore, the transition is very slow, and even at a few times the transition temperature
even the bulk quantities are not yet fully dominated by the partonic processes. The system
then behaves not like a gas, but rather like an almost ideal fluid.
This is the situation encountered in the early universe. While it cools done, it will
go through this cross-over. Before that, it is essentially dominated by the quarks and
gluons, and only afterwards it starts to be dominated by the hadrons. However, because
the transition is a cross-over, it seems that the transition had little quantitative influence
on the evolution of the universe.
The situation in the remainder of the phase diagram is not yet clear. It is possible
to map out parts of it with heavy-ion collisions at lower energies. Because then less
energy is available, less particles are produced, and therefore the baryon chemical potential
is larger. Still, the accessible region is that of rather high temperatures, above those
characteristic for neutron stars, and likely below the relevant chemical potentials. Also,
numerical simulations start again to fail the larger the chemical potential becomes. Hence,
the situation becomes less and less clear. What seems to be certain at the current time
is that for quite some distance into the chemical potential direction little changes, and
the cross-over remains at a temperature only slowly decreasing with increasing chemical
potential. There are some speculations about a critical end-point, from which a phase
boundary starts, which eventually meets with the chemical potential axis, but this is not
yet settled. Other than that, the field is still wide open.
Chapter 6
Weak interactions
The last ingredient of the standard model are the weak interactions, which will again be a
gauge interaction, though of quite different character than both the electromagnetic and
the strong interaction. The name weak is also somewhat misleading, as the interaction
is actually stronger than the electromagnetic one, but it is short-range, as is the strong
interaction, though for entirely different reasons.
It is impossible to separate from any description of the weak interactions the recently
established Higgs sector, as without it the weak interactions would be very different in
character. In fact, without the Higgs sector, it would be very similar to the strong interac-
tion. The Higgs effect introduces the only non-gauge interactions into the standard-model.
Depending on personal taste, the interactions of the Higgs are not counted at all as true
interactions, as many believe these are just low-energy effective realizations of a yet un-
known gauge interaction, a single interaction, or, by counting the number of independent
coupling constants, 13 interactions. This will become clearer in the following.
One of of the most interesting facts about the weak interactions is that they live on
a quite different scale than electromagnetism and the strong interactions. While electro-
magnetism acts mainly on the scales of atomic physics, and the strong interactions at the
scale of nuclear and hadronic physics, the typical scale of the weak interactions will be
found to be of the order of 100 GeV. Since this is the highest known scale in particle
physics yet, apart from the gravitational scale, it is widely believed that the study of the
weak interaction will be the gateway to understand whether there is any other physics
between this scale and gravity. This is the main reason it plays such a central role in
modern particle physics.
One of the reasons why the weak interactions, or more commonly denoted as the
electroweak sectors for reasons to become clear later, is rather unwieldy is that it is an
aggregation of several, intertwined phenomena. Though each of them can be studied on
83
84 6.1. decay and parity violation
its own, essentially none of them unfolds its full relevance without the others. This makes
the electroweak sector often at first somewhat confusing. The different elements appearing
will be a new gauge interaction, a mixing between this new interaction and QED, parity
violation, the Higgs effect, and flavor changes.
6.1 decay and parity violation

The weak interaction surfaced for the first time in certain decays, particularly the -
decay of free neutrons and of bound nuclei, which could not be explained with either the
electromagnetic force nor the then newly discovered strong force. That this is impossible
is directly seen from the quark content of the proton and the neutron: A transmutation
of an up to a down quark is not possible in either interaction, because both are flavor-
conserving. Furthermore, no hadrons are produced in the process, so this cannot be an
exchange reaction. hence, another, third force has to be involved. Due to the small rate
and also the very rarely seen consequences of this force in ordinary physics, this force was
called weak force.
Though no additional hadrons are produced, other particles are involved. Especially
the decay conserves electric charge by adding an electron to the final state, i. e. n p+e .
However, for a free neutron this violates spin conservation. Furthermore, it is found that
both the proton and the electron exhibit a continuous energy spectrum, and therefore
energy conservation would be violated. The resolution of this is that the decay is indeed
into three particles, and involves a new particle, the so-called electron-neutrino e , actually
an anti-electron-neutrino e , n p + e + e . Since the neutrino turns out to be a fermion
of spin 1/2, this fixes both spin and energy conservation. The neutrino itself is a quite
mysterious particles, and will be discussed further in section 6.2.
The first attempts of a field-theoretical formulation were based on a four-fermion in-
teraction of type
GF (p n)(e e ), (6.1)
where p, n, e, and e represent the fields of the involved proton, neutron, electron, and
the (anti)neutrino and are four-component spinors. The characteristic scale for the weak
process was set by the Fermi constant, which is of order 1.14105 GeV2 . This is
not a renormalizable interaction, and as such should be only the low-energy limit of an
underlying renormalizable theory.
Besides the appearance of a new particle, the -decay shows another quite surprising
property: It violates parity. Thus, a -decay is not operating in the same way as it would in
the mirror. Experimentally, this can be observed by embedding the process into a magnetic
Chapter 6. Weak interactions 85
field. A parity-conserving interaction, like electromagnetism, should yield the same result,
irrespective of whether the spins are aligned or anti-aligned with the magnetic field1 , due
to parity invariance. However, experimentally a difference is observed, indicating parity
violation. More precisely, it was observed that a polarized neutron which decays will emit
the electron preferentially in one direction. Therefore, the interaction must couple spin
s and momenta, and would therefore have a contribution proportional to sp. However,
the momenta of the decay products also depend on the invariant mass, p2 , and thus on
a scalar contribution. Since therefore both scalars (scalar products of two vectors or two
axial vectors) and pseudoscalars (products of a vector and an axial vector) appear imply
that the interaction is not having a definite transformation behavior under parity, and is
thus parity violating. In fact, it turned out that it is maximally parity violating.
To give this a more formal version, it is necessary to consider how coupling of fermions
transform under parity. The parity transformation of a spinor is obtained by multiplying
it with 0 , the time-Dirac matrix. So, for spinors the parity transformation is given by
P = 0 .
Since furthermore 0 0 = , the four-fermion coupling (6.1) will indeed transform as

a scalar
P (( )( )) = (0 0 )(0 0 ) = ( )( ).
To obtain a pseudoscalar coupling, one of the vectors would have to be replaced by an
axial vector. This can be obtained if there would be a matrix such that 0 5 0 = 5 . In
fact, such a matrix is given by
5 = i 0 1 2 3 .
This matrix anticommutes with all Dirac matrices,
{5, } = 0.
As a consequence, the current

5 ,
is an axial vector, and can be used to obtain a pseudoscalar coupling.
However, this does not yet determine to which extent the weak interactions should be
parity violating, and this can also not be predicted on basis of the standard model to be
developed. Experimentally, however, it is found that parity is maximally violated. For
massless particles (e. g. for neutrinos to a very good accuracy) this would imply that only
1
In a practical case, the spin of the originating nucleus plays also a role, and a suitable choice is
mandatory to see the effect simply. It was first observed in the -decay of 60 Co.
86 6.2. Neutrinos
one of the helicity states would be affected by the weak interactions. The helicity state
for a spinor is projected out as
1 5
.
2
The sign is determined by whether left-handed or right-handed states should be selected.
Experiment finds that only left-handed states are involved, and thus a minus sign is appro-
priate. Furthermore, the weak interactions are found to violate also the charge conjugation
symmetry maximally. Hence, the sign is not reversed for the anti-particle state. Therefore,
the correct four-fermion interaction version of the weak interactions would be (appropri-
ately normalized)

GF 1 5 1 5 1 5 1 5
,
2 2 2 2 2
which therefore exhibits maximal violation of C and P individually, but conserves CP.
Hence, the new interaction is very different from either the strong or electromagnetic
force: There are particles only responding to it, the neutrinos, and it violates both parity
and flavor conservation. Furthermore, it is found to be very short range, much more than
the strong interaction. On the other hand, electrons and neutrinos are both sensitive to
it, but are not confined. Thus, the way how this range limitation is operating has to be
different from the confinement mechanism of QCD.
6.2 Neutrinos
Before discussing the weak interaction itself. It is worthwhile to introduce the neutrinos
first. Some further properties of them will be added later, in section 6.9, when the weak
force has been discussed in some more detail.
So far, only one neutrino has been introduced, the electron-neutrino, a fermion of
spin 1/2. There are actually two more flavors of neutrinos, the muon-neutrino and
the -neutrino , which are again fermions of spin 1/2. As their name suggest, they are
partnered with one lepton each, forming the last member of the particle generations, which
are conventionally assigned to be the sets {u, d, e, e}, {c, s, , }, and {t, b, , }. These
combinations are somewhat arbitrary, and mainly motivated by the masses of the particles.
Another motivation is that flavor is, though violated, still approximately conserved for
leptons and neutrinos. In then turns out that, if electrons carry an electron flavor number
of one and positrons of -1, electro-neutrinos likewise carry an electron flavor number of 1
and anti-electron neutrinos of -1. Hence, e. g. the -decay, conserves this flavor number.
Likewise, there are approximately conserved flavors for muon-like and tauon-like particles.
As a consequence, neutrinos are usually also included in the term leptons, though not
always.
When it comes to masses, the neutrinos actually are very surprising. To date, it is
only known for sure that their mass is below2 2 eV. A direct measurement of the mass of
the neutrinos is complicated by the fact that they are only interacting through the weak
force, which reduces cross-sections by order of magnitudes compared to the leptons and
the quarks. However, it was possible to measure the mass differences of the neutrinos in a
way to be explained in section 6.9. But so far, it was only possible to measure differences
without knowing which of the possible differences are measured. The two values are about
0.009 eV and 0.05 eV. This implies that at least one of the neutrinos has a mass of 0.05
eV, though it is not clear, which of them. It is furthermore not yet clear, whether the
other two to or just one is much lighter. If the electron neutrino would be the heaviest,
there are very good chances to measure its mass directly within the next decade. Since the
other neutrino species are much harder to produce in larger numbers, this will take much
longer if it is is either of the other ones. It is also, in principle, still possible that one of
the neutrinos could be massless, though there is yet no strong evidence for this possibility.
6.3 Flavor-changing currents

As already observed, the weak interaction changes the flavor of down quarks to up quarks.
Likewise, processes have been observed that changes strange to charm and bottom to top,
and vice versa, kinematics always permitting. There are also processes observed which
changes, say, charm to up quarks. However, these processes are strongly suppressed, and
will be neglected for now. They will be added below in section 6.8.
After neglecting these processes, the flavor changes only occur within one generation.
Since in this process also one unit of electric charge is transferred to leptons, these are called
flavor-changing charged current, or FCCC for short. In addition, there are also flavor-
changing neutral currents, FCNC, but they belong to the class of processes postponed.
This observation strongly suggest that the quark flavors in any generation can be
regarded just as two states, a doublet, under the weak interaction, similar to the isospin.
Hence, a new symmetry is introduced, the weak isospin symmetry, under which doublets
are formed from the two quarks of each generation, and also the electron and the neutrino
of each generation. This doublet structure has been already observed with the isospin
relation (5.4), and it will be seen later how the electromagnetic charge enters this. Note,
2
This is a conservative upper bound. There are good indications that it is at least an order of magnitude
smaller.
88 6.4. W and Z bosons
however, that in FCCC also the masses play a role, and for reasons to be discussed later
massless particles would not participate. Therefore flavor changes for the leptons occur
much more rarely because of the very small neutrino masses, though have been observed
nonetheless.
6.4 W and Z bosons

The doublet structure can be regarded as a charge, the weak isospin charge. The weak force
couples to this charge. The interaction which mediates the neutron decay appears to be
described by the interaction of four fermions, an up quark, a down quark, an electron, and
an anti-electron neutrino, as given in (6.1). The strength of this interaction is characterized
by a new coupling constant, the Fermi constant GF , which has units of inverse energy
squared, and a value of 1.16 105 GeV2 , corresponding to an energy scale of about 293
1
GeV, though usually this energy scale is reduced by a factor of 2 4 to 246 GeV for reasons
of conventions. At any rate, the energy scale is much larger than any hadronic scale. The
only other object encountered so far with a similar scale is the top quark, though this
appears at the current time unrelated.
Since weak charges are not confined, another mechanism is required to give the weak
interaction a finite range. According to section 5.1, a possibility is that the charge carriers
are massive. Another problem with this theory was that such a theory is not renormal-
izable, in the sense of section 3.7. Hence, it would not be possible to obtain meaningful
predictions from it at energies of similar size or larger than 246 GeV, which would already
cover any top quark physics. It was therefore early on argued that this cannot be the
underlying theory, and at a scale of around 100 GeV there must be a different theory
present.
This indeed turns out to be true. In fact, it was suggested that there should exist,
similarly to QED, an exchange boson. However, to produce such a scale and range, they
would have to be massive with a mass MW . Assuming a coupling constant g for the
process, the scale would be set by expressions of the type
g 2
GF 2 , (6.2)
MW
indicating a mass scale of about 100 GeV for the process, if g is not too different from one.
This already sets the scale of the weak interactions. Furthermore, the appearance of a
mass-scale indicates that the weak interactions will have only a limited range, of the order
of 1/MW , and would be screened beyond this scale. Its range is therefore much shorter
than that of any other force.
Over the time other weak effects were found, all characterized by the scale GF . In
particular, after some time the postulated interaction bosons have been observed directly.
There are two charged ones, the W , and a neutral one Z 0 , with masses about 80 and 91
GeV, respectively, and where the superscripts indicate their electric charge. In fact, the
weak interaction turns out to have a very similar structure as QED and QCD in the sense
that it is a gauge theory, where additional gauge bosons appear. The doublet structure
already suggests as the underlying gauge group one with two different charges and three
gauge bosons. It finally turns out that the correct gauge group is SU(2). That the W
carry electric charge indicate a relation to QED, which will be discussed in section 6.7.
However, masses are not permitted for gauge bosons, or else gauge invariance would be
broken. This breaking has severe effects, making the theory non-renormalizable. There-
fore, despite the experimental verification of the existence of the weak gauge bosons and
their masses, this would not improve the theoretical description. This was only achieved
after gauge theories have been combined with the Goldstone theorem described in section
5.7.
To understand that this is the right way, consider a photon A coupled to some external
current j in the Lagrangian
1
L = ( A A )( A A ) j A .
4
It will be this current, and its generalization, which will be instrumental in obtaining a
massive gauge boson and in addition a gauge symmetry, and thus a superficially massive
vector boson. However, since these two concepts are usually contradictory, this will lead
to a hiding of the gauge symmetry, similarly as the global symmetry has been hidden in
section 5.7.
The corresponding equation of motion for the photon is
2 A A = j . (6.3)
Furthermore, the field should, somehow, satisfy the condition
( 2 + M 2 )A = 0. (6.4)
To proceed further, it is useful to choose a particular gauge, in this case the Landau-
Lorentz gauge A = 0. Of course, fixing a gauge is a perfectly acceptable way to perform
a calculation, provided the calculation is not violating gauge invariance at any point. Then
the results for gauge-invariant quantities, like the number of physical polarizations, will be
valid. In the present case, doing a calculation without fixing the gauge is at best tedious.
The results can be translated to the general case of an arbitrary gauge, but it should
90 6.4. W and Z bosons
be kept in mind that any relation on gauge-variant quantities, like the gauge field A ,
will only be valid in this particular gauge. However, statements about gauge-invariant
quantities, like, e. g., the field-strength tensor F in this Abelian example, will hold in
any gauge.
After imposing this gauge, the equation of motion (6.3) for the spatial components
becomes
2A~ = ~j. (6.5)
To simplify matters further, restrict to a time-independent situation. Then all time-
derivatives vanish, and the equation (6.4) takes the form
~ = M 2 A.
2A ~ (6.6)
From the equations (6.5) and (6.6) it is clear that the current must satisfy the condition
~
~j = M 2 A
in order that the photon (in this gauge) becomes apparently massive and at the same
time the gauge invariance is not broken. It is left to find some way of producing the
~ is due
appropriate current. Physically, the origin of such a current being proportional to A
to the response of a medium to the acting electromagnetic fields. E. g., this is realized by
the Meissner effect in superconductivity. Therefore, giving a mass to the photon requires
a medium which provides a response such that the photon becomes damped and can
therefore propagate only over a finite distance. In the electroweak case, the role of this
medium will be taken by a condensate of Higgs particles.
This is a screening process as can be seen from the following simple example. Consider
the Maxwell-equation
~ B = ~j = M 2 A.
~
Taking the curl on both sides and using ~ A

~=B
~ yields
~ = M 2 B.
2B ~
In a one-dimensional setting this becomes

d2
B = M 2 B,
dx2
which is immediately solved by
B = exp(Mx).
Thus the magnetic field is damped on a characteristic length 1/M, the screening length.
The inverse of the screening length, being a mass in natural units, is therefore the charac-
teristic energy scale, or mass, of the damping process. However, by this only an effective
mass has been obtained. The other vital ingredient for a massive vector boson is a third
polarization state. Similarly, also this other degree of freedom will be provided effectively
by the medium, as will be discussed below.
To make the preceding discussion applicable to elementary particle physics, it is useful
to write them down covariantly. For this purpose, select the Landau gauge A = 0.
Then the Maxwell equation takes the form
2 A = j .
The necessary requirement that, at least approximately, the current has a component
proportional to A is then directly obtained by considering the current for a charged,
scalar field, which has the form
j = iq(+ ( + )) 2q 2 A ||2 ,
which can be derived as the Noether current from the corresponding Lagrangian
L = D + D
D = + iqA .
Thus, the Maxwell equation for A becomes
2 A = iq(+ ( + )) 2q 2 A ||2.
This will have precisely the required form, if the modulus of the field, || = v can be
arranged to have a non-zero value in the vacuum. However, this will only be possible, if
its expectation value obeys
h0||0i =
6 0 (6.7)
in the selected gauge. In this case, the photon would acquire an effective mass of (2q 2 v 2 )1/2 .
And by this, the photon is screened out, making the effect short-ranged. How to implement
this will be discussed next.
6.5 The Higgs effect and the Higgs boson

Motivated by the previous discussion, and by the considerations in section 5.7 it appears
reasonable to start with a scalar field with self-interactions, call it Higgs field, and couple
it to a gauge field. When coupling the Higgs to a gauge field, symmetry breaking, now
called the Higgs (or more precisely Englert-Brout-Higgs, and several other names) effect
92 6.5. The Higgs effect and the Higgs boson
becomes more complicated, but at the same time also more interesting, than in the case
without gauge fields.
For simplicity, start with an Abelian gauge theory, coupled to a single, complex scalar,
the so-called Abelian Higgs model, before going to the full electroweak and non-Abelian
case. This theory has the Lagrangian
1 1
L = (( + iqA ))+ ( + iqA ) ( A A )( A A )
2 4
1 2 + 1 2 + 2
+ ( ) . (6.8)
2 2 f2
Note that the potential terms are not modified by the presence of the gauge-field. There-
fore, the extrema have still the same form and values as in the previous case, at least
classically. However, it cannot be excluded that the quartic + A A term strongly dis-
torts the potential. This does not appear to be the case in the electroweak interaction,
and this possibility will therefore be ignored.
To make the consequences of the Higgs effect transparent, it is useful to rewrite the
scalar field as3
f
(x) = + (x) exp(i(x)).
2
This is just another reparametrization for the scalar field, compared to and previously.
It is such that at = 0 this field configuration will be a classical minimum of the potential
for any value of the phase . Inserting this parametrization into the Lagrangian (6.8)
yields
2 2 2
q2

1 1 f f f
L = +
+ + qA + + A A +
2 2 2 2 4 2
2 2
4
1 1 f 1 f
( A A )( A A ) + 2 + 2
+
4 2 2 2f 2
This is an interesting structure, where the interaction pattern of the photon with the radial
and angular part are more readily observable.
Now, it is possible to make the deliberate gauge choice
1
A = 2 . (6.9)
q
3
Note that if the space-time manifold is not simply connected and/or contains holes, it becomes
important that is only defined modulo 2. For flat Minkowski (or Euclidean) space, this is of no
importance. However, it can be important, e. g., in finite temperature calculations. It is definitely
important in ordinary quantum mechanics, where, e. g., the Aharanov-Bohm effect and flux quantization
depend on this.
This is always possible. It is implemented by first going to Landau gauge and then perform
the gauge transformation
1
A A +
q
exp(i).
This gauge choice has two consequences. The first is that it makes the scalar field real
everywhere. Therefore, the possibility of selecting the vacuum expectation value of to
be real is a gauge choice. Any other possibilities, e. g. purely imaginary, would be an
equally well justified gauge choice. This also implies that the actual value of the vacuum
expectation value f of is a gauge-dependent quantity, and will vanish when not fixing a
gauge. The Lagrangian then takes the form4
2
q2

1 f
L = + A A +
2 4 2
2 4
1 2

1 1 f f
( A A )( A A ) + 2 + + (6.10)
4 2 2 2 f2 2
i. e., the second term has now exactly the form of a screening term, and yields an effective
mass qf /4 for the photon field. Furthermore, if could be neglected, the Lagrangian
would be just
q2f 2 1
L= A A ( A A )( A A ),
8 4
i. e., the one of a massive gauge field. Together with the gauge condition for A , which links
the longitudinal part of the gauge boson to the now explicitly absent degree of freedom
, this implies that the field A acts now indeed as a massive spin-1 field. Furthermore,
this Lagrangian is no longer gauge-invariant. This is not a problem, as it was obtained by
a gauge choice. Thus, it is said that the gauge symmetry is hidden. Its consequences are
still manifest, e. g., the mass for the gauge boson is not a free parameter, but given by the
other parameters in the theory. By measuring such relations, it is in principle possible to
determine whether a theory has a hidden symmetry or not.
Colloquial, the hiding of a symmetry is also referred to as the breaking of the symmetry
in analogy with the case of a global symmetry. However, a theorem, Elitzurs theorem,
actually forbids this to be literally true.
The Goldstone theorem of section 5.7, actually guarantees that a mass will be provided
to each gauge boson associated with one of the hidden directions at the classical level. This
4
Upon quantization, additional terms are introduced due to the gauge-fixing. At the classically level
studied here, they can be ignored.
94 6.5. The Higgs effect and the Higgs boson
hidden direction is here the phase, which is massless. This masslessness of the Goldstone
boson is actually guaranteed by the Goldstone theorem. Hence, a massless Goldstone
boson is effectively providing a mass to a gauge boson by becoming its third component
via gauge-rotation, and vanishes by this from the spectrum. It is the tri-linear couplings
which provide the explicit mixing terms delivering the additional degree of freedom for
the gauge boson at the level of the Lagrangian. Hence, the number of degrees of freedom
is preserved in the process: In the beginning there were two scalar and two vector degrees
of freedom, now there is just one scalar degree of freedom, but three vector degrees of
freedom. The gauge-transformation made nothing more than to shift one of the dynamic
degrees of freedom from one field to the other. This was possible due to the fact that both
the scalar and the photon are transforming non-trivially under gauge-transformations.
This can, and will be, generalized below for other gauges. In general, it turns out that
for a covariant gauge there are indeed six degrees of freedom, four of the vector field, and
two from the scalars. Only after calculating a process it will turn out that certain degrees
of freedom cancel out, yielding just a system which appears like having a massive vector
particle and a single scalar.
Note that though the original scalar field was charged as a complex field, the radial
excitation as the remaining degree of freedom is actually no longer charged: The coupling
structure appearing in (6.10) is not the one expected for a charged field.
However, the choice (6.9), which is called the unitary gauge, is extremely intransparent
and cumbersome for most actual calculations. The reason is that the quantization of the
the theory in this gauge requires the introduction of an infinite number of further terms
into the Lagrangian, though with fixed coefficients.
A more convenient possibility, though at the cost of having unphysical degrees of
freedom which only cancel at the end, are t Hooft gauges. To define this gauge once more
the decomposition
= (, f + )
for the scalar field is useful. The Lagrangian then takes the form, up to constant terms,
1 (gf )2 1 1
L = F F + A A + 2 2 + gf A (6.11)
4 2 2 2
1 2 2
+2igA ( + ) 2g(gf )AA g 2 (2 + 2 )A A ( + 2 )2 .
2 f2
There are a number of interesting observations. Again, there is an effective mass for A ,
due to the four-field interaction term. Secondly, only the field has a conventional mass
term, the mass-term for the field has canceled with the two-field-two-condensate term
from the quartic piece of the potential. Finally, there will be terms of type f A . This
implies that a photon can change into a while moving, with a strength proportional to
the condensate f , and thus mixing between one of the Higgs degrees of freedom and the
photon occurs. Therefore, the photon and this scalar, the would-be Goldstone boson, will
mix. Note finally that though many more interaction terms have appeared, none of them
has any new free parameter, as a consequence of the now hidden symmetry. The only
quantity which looks like a new quantity is the condensate value f , though it is classically
uniquely determined by the shape of the potential. In a quantum calculation, it can also
be determined, but not in perturbation theory, where it remains a fit parameter.
The gauge condition to be used for quantizing this Lagrangian is then given by
A = qf , (6.12)
the so-called renormalizable or t Hooft gauge, where is an arbitrary parameter, a so-

called gauge parameter. This gauge choice removes at the quantum level the mixing
between the photon and the Higgs. The resulting theory has then the massive Higgs field,
a massless Goldstone field, and a massive photon field. The surplus degrees of freedom
are, as stated, not actually there, but will cancel at the quantum level.
The form of the Lagrangian (6.11) already indicates how the weak gauge bosons will
receive their mass. The necessary Higgs boson has by now indeed been discovered, and its
mass is found to be about 125 GeV.
6.6 Parity violation and fermion masses

So far, the construction seems to reproduce very well and elegantly the observed phe-
nomenology of the electroweak interactions.
However, there is one serious flaw. The Dirac equation for fermions, like leptons and
quarks, has the form

1 5 1 + 5
0 = (i D m) = i D + m+ m .
2 2
The covariant derivative is a vector under a weak isospin gauge transformation, and so
is the spinor (1 5 )/2. However, the spinor (1 + 5 )/2 is a singlet under such a
gauge transformation. Hence, not all terms in the Dirac equation transform covariantly,
and therefore weak isospin cannot be a symmetry for massive fermions. Another way of
observing this is that the mass term for fermions in the Lagrangian can be written as
Lm = m(L R R L ),
and therefore cannot transform as a gauge singlet.

96 6.7. Weak isospin, hypercharge, and electroweak unification
However, massless fermions can be accommodated in the theory. Since the observed
quarks and (at least almost) all leptons have a mass, it is therefore necessary to find a
different mechanism which provides the fermions with a mass without spoiling the isoweak
gauge invariance.
A possibility to do so is by invoking the Higgs-effect also for the fermions and not only
for the weak gauge bosons. By adding an interaction
+
k 1 5 k 1 + 5 k 1 5 k 1 + 5
Lh = gf k i ij + ij j + gf k i ij + ij j ,
2 2 2 2
this is possible. The constant matrices and have to be chosen such that the terms
become gauge-invariant. Calculating their precise form is tedious, but straightforward. If,
in this interaction, the Higgs field acquires a vacuum expectation value, = f +quantum
fluctuations, this term becomes an effective mass term for the fermions, and it is triv-
ially gauge-invariant. Alongside with it comes then an interaction of Yukawa-type of the
fermions with the Higgs particle. However, the interaction strength is not a free parameter
of the theory, since the coupling constants are uniquely related to the tree-level mass mf
of the fermions by
2mf
gf = .
f
But the 12 coupling constants for the three generations of quarks and leptons are not
further constrained by the theory, introducing a large number of additional parameters in
the theory. Though phenomenologically successful, this is the reason why many feel this
type of description of the electroweak sector is insufficient. However, even if it would be
incorrect after all, it is an acceptable description at energies accessible so far, and thus
has become part of the standard model.
6.7 Weak isospin, hypercharge, and electroweak uni-

fication
It is now clear that a theoretical description of the weak interactions requires a gauge
theory, as well as some matter field(s) to hide it and provide a mass to the weak gauge
bosons. The fact that there are three distinct gauge bosons, two Ws and one Z indicates
that the gauge theory has to be more complex than just the U(1) gauge theory of QED. It
will thus be a non-Abelian gauge theory. Furthermore, since two of them are charged the
connection to the electromagnetic interactions will not be trivial, and there will be some
kind of mixing. All of these aspects will be taken into account in the following.
6.7.1 Constructing the gauge group

Phenomenological, as discussed in section 6.3, the weak interactions provides transitions of
two types. One is a charge-changing reaction, which acts between two gauge eigenstates.
In case of the leptons, these charge eigenstates are almost (up to a violation of the order
of the neutrino masses) exactly a doublet - e. g., the electron and its associated electron
neutrino. Therefore, the gauge group of the weak interaction should provide a doublet
representation. In case of the quarks this is less obvious, but also they furnish a doublet
structure. Hence, an appropriate gauge group for the weak interaction will contain at
least SU(2). Since only the left-handed particles are affected, this group is often denoted
as SU(2)L , but this index will be dropped here. Therefore, there are three doublets,
generations, of leptons and quarks, respectively in the standard model. However, the
mass eigenstates mix all three generations, as will be discussed in detail below.
Since the electric charge of the members of the doublets differ by one unit, the off-
diagonal gauge bosons, the W, must carry a charge of one unit. Furthermore, such a
gauge group has three generators. The third must therefore be uncharged, as it mediates
interactions without exchanging members of a doublet.
The quantum number distinguishing the two eigenstates of a doublet is called the third
3
component of the weak isospin t, and will be denoted by t3 or IW . Therefore, the gauge
group of the weak interactions is called the weak isospin group.
However, the weak gauge bosons are charged. Therefore, ordinary electromagnetic
interactions have to be included somehow. Ordinary electromagnetism has as gauge group
the Abelian U(1). The natural ansatz for the gauge group of the electroweak interactions
is thus the gauge group SU(2)U(1)5 . With this second factor-group comes a further
quantum number, which is called the hypercharge y. The ordinary electromagnetic charge
is then given by
y
eQ = e t3 + . (6.13)
2
Thus, the ordinary electromagnetic interaction must be somehow mediated by a mixture
of the neutral weak gauge boson and the gauge boson of the U(1). This is dictated by
observation: It is not possible to adjust otherwise the quantum numbers of the particles
such that experiments are reproduced. The hypercharge of all left-handed leptons is 1,
while the one of left-handed quarks is y = +1/3.
Right-handed particles are neutral under the weak interaction. In contrast to the
t = 1/2 doublets of the left-handed particle, they belong to a singlet, t = 0. All in all,
the following assignment of quantum numbers for charge, not mass, eigenstates will be
5
Actually, the correct choice is SU(2)/Z2 U(1), although this difference is very rarely of relevance.
necessary to reproduce the experimental findings:
Left-handed neutrinos: t = 1/2, t3 = 1/2, y = 1 (Q = 0)
Left-handed leptons: t = 1/2, t3 = 1/2, y = 1 (Q = 1)
Right-handed neutrinos: t = 0, t3 = 0, y = 0 (Q = 0)
Right-handed leptons: t = 0, t3 = 0, y = 2 (Q = 1)
Left-handed up-type (u, c, t) quarks: t = 1/2, t3 = 1/2, y = 1/3 (Q = 2/3)
Left-handed down-type (d, s, b) quarks: t = 1/2, t3 = 1/2, y = 1/3 (Q = 1/3)
Right-handed up-type quarks: t = 0, t3 = 0, y = 4/3 (Q = 2/3)
Right-handed down-type quarks: t = 0, t3 = 0, y = 2/3 (Q = 1/3)
W+ : t = 1, t3 = 1, y = 0 (Q = 1)
W : t = 1, t3 = 1, y = 0 (Q = 1)
Z: t = 1, t3 = 0, y = 0 (Q = 0)
: t = 0, t3 = 0, y = 0 (Q = 0)
Gluon: t = 0, t3 = 0, y = 0 (Q = 0)
Higgs: a complex doublet, t = 1/2 with weak hypercharge y = 1. This implies zero
charge for the t3 = 1/2 component, and positive charge for the t3 = 1/2 component
and negative charge for its complex conjugate
This concludes the list of charge assignments for the standard model particles. The Higgs
case will be special, and will be detailed in great length below.
Since at the present time the photon field and the Z boson are not yet readily identified,
it is necessary to keep the gauge boson fields for the SU(2) and U(1) group differently, and
these will be denoted by W and B respectively. The corresponding pure gauge part of the
electroweak Lagrangian will therefore be
1 1
Lg = Ga G
a F F

4 4
F = B B
Ga = Wa Wa + gf abc Wb Wc ,
where g is the weak isospin gauge coupling. f abc are the structure constants of the weak
isospin gauge group, which is just the SU(2) gauge group.
Coupling matter fields to these gauge fields proceeds using the ordinary covariant
derivative, which takes the form
ig ig y
D = + a Wa + B ,
2 2
where g is the hypercharge coupling constant, which is modified by the empirical factor y.
Note that y is not constrained by the gauge symmetries, and its value is purely determined
by experiment. Thus, why it takes the rational values it has is an unresolved question
to date. However, this fact would come about naturally, if the weak gauge group would
originate from a different gauge group at higher energies, say SU(2)U(1)SU(3), which
is hidden to the extent that all other fields charged under this larger gauge group are
effectively so heavy that they cannot be observed with current experiments.
The matrices a are determined by the representation of SU(2) in which the matter
fields are in. For a doublet, these will be the Pauli-matrices. For the adjoint representation,
these would be given by the structure constants, bca = f abc , and so on. For fermions, of
course, this covariant derivative is contracted with the Dirac matrices . Precisely, to
couple only to the left-handed spinors, it will be contracted with (1 5 )/2 for the Wa
term and with for the kinetic and hypercharge term. By this, the phenomenological
couplings are recovered in the low-energy limit, as the exchange of a massive gauge boson
then becomes proportional to 1/M 2 , thus recovering the Fermi-coupling g 2 /M 2 . How this
mass disappears in the case of the Abelian gauge group will be discussed next.
6.7.2 Hiding the electroweak symmetry

To have a viable theory of the electroweak sector it is necessary to hide the symmetry
such that three gauge bosons become massive, and one uncharged one remains massless.
Though this can be of course arranged in any gauge, it is most simple to perform this in
the unitary gauge. Since three fields have to be massive, this will require three pseudo-
Goldstone bosons. Also, since empirically two of them have to be charged, as the W
bosons are charged, the simplest realization is by coupling a complex doublet scalar field,
the Higgs field, to the electroweak gauge theory
Lh = Lg + (D )+ D + V (+ )
!
1 1 + i2
= (6.14)
2 3 + i4
where the field is thus in the fundamental representation of the gauge group. Its hy-
percharge will be 1, an assignment which will be necessary below to obtain a massless
photon. The potential V can only depend on the gauge-invariant combination + , and
thus can, to be renormalizable, only contain a mass-term and a quartic self-interaction.
The mass-term must be again of the wrong sign (imaginary mass), such that there exists
a possibility for the field to acquire a (gauge-dependent) vacuum-expectation value.
To work in unitary gauge it is best to rewrite the Higgs field in the form
!
i a a 0
=e 2 .

There are now the three a fields and the field. Performing a gauge transformation such
that the phase becomes exactly canceled, and setting = f + with f constant makes
the situation similar to the one in the Abelian Higgs model. Note that by a global gauge
transformation the component with non-vanishing expectation value can be selected still
at will.
The mass will be made evident by investigating the equation of motion for the gauge
bosons
2 Wa ( Wa ) = ja + jaW
2 B ( B ) = jy ,
where j W contains contributions which involve only the W s, while j contains all remaining
terms. Only the latter will be relevant below, and therfeore for it the subscript will
be dropped. Thus, it is necessary to determine the weak isospin current ja and the
hypercharge current jy . In contrast to the Abelian case this current will now also contain
contributions from the self-interactions of the W gauge bosons. However, at tree-level
these can be ignored.
The contribution of the Higgs field to the currents6 are
ig + a
ja = ( D (D )+ a )
2
ig +
jy = ( D (D )+ ).
2
The appearance of a makes manifest that the current ja is a weak isovector current, while
their absence signifies that the hypercharge current jy is a weak isoscalar current.
6
Of course, all fields with non-zero weak isospin will contribute to this current. But only the Higgs-field
will do so classically.
Expanding the covariant derivatives yields7

ig + a g2 + gg y + a
= ja + a
( ( ) ) W a
B (6.15)
2 2 2
ig + gg y + (g y)2 +
jy = ( ( )+ ) a Wa B .
2 2 2
It should be noted that the two equations are not only coupled by the Higgs field, but by
both, the Wa and B gauge fields. Selecting now the Higgs field to be of the form
!
0
= f + quantum fluctuations,

2
where the quantum fluctuations vanish at tree-level will provide the vacuum expectation
value !
0
h0||0i = f .

2
This is then as in the Abelian case, and indeed sufficient to provide a mass for the gauge
bosons.
Ignore for a moment the hypercharge contribution. Then the term linear in Wa on the
right-hand side of (6.15) provides a term
2
2 a gf
M W = Wa ,
2
and thus a mass MW = gf /2 is provided. Thus, the equation of motion for the Wa field
takes the form
( 2 + MW
2
)Wa ( Wa ) = ja (f, , Wa, ...),
where the mass-term has been removed from the current. Thus, formally this is the
equation of motion for a massive gauge boson, interacting with itself and other fields due
to the current ja . As has been seen in the Abelian case, the additional degree of freedom
has been selected by the choice of gauge, and is provided by the phase of the Higgs field.
Next it is necessary to check what happens when including also the B part once
more. Since the t3 = 1/2 component of the Higgs field delivers the mass, this component
should not be electrically charged. According to the relation (6.13) thus the assignment
of the hypercharge y = 1 for the Higgs field is a-posterior justified. In the full current
of the Higgs field, besides the contribution + Wa also the contribution gg y/2B+ a
appears. Since the matrices 1 and 2 are off-diagonal it follows that
!
0
(0, f ) 1,2 = 0.
f
7
Using { a , b } = 2 ab .
Thus only in the third component of the current a contribution due to the vacuum ex-
pectation value of the Higgs field appears. As a consequence, despite the appearance of
the additional B gauge boson the result for the mass (at tree-level) for the off-diagonal
W1,2 bosons is not changed. However, this is not the case for the third component. The
relevant part of the equation of motion reads
2 2 2
2 3 3 g f gf g f
W ( W ) = W3 + B .
2 2 2
Consequently, also the relevant part of the equation of motion for the hypercharge gauge
boson is of similar structure
2
2 gf gf g f 3
B ( B ) = B + W .
2 2 2
Hence, even in the vacuum the equations are coupled, and thus both gauge bosons mix.
These equations can be decoupled by changing variables as
A = W3 sin W + B cos W
Z = W3 cos W B sin W ,
which are the fields given the name of the photon A and the Z boson Z . The mixing
parameter W is the (Glashow-)Weinberg angle W , and is given entirely in terms of the
coupling constants g and g as
g
tan W =
g
g
cos W = p
g2 + g2
g
sin W = p .
g2 + g2
Using the inverse transformations
A cos W + Z sin W
W3 =
cos W sin W
A sin W Z cos W
B =
cos W sin W
it is possible to recast the equations of motion. The equation of motion for the photon
takes the form
2 A ( A ) = sin W j3 .
Here, j3 contains all contributions which are not providing a mass to the gauge bosons.
Thus, the photon is effectively massless. The price paid is that it now has also direct
interactions with the weak gauge bosons. On the other hand, the equation for the Z
boson takes the form

2 f 2 (g 2 + g 2 )
Z Z = Z + cos W j3 .
4
Thus, the Z boson acquires a mass larger than the W 1,2 boson. In particular the relation
is
MZ cos W = MW .
According to the best measurements this effect is, however, only of order 15%. At the same
time the self-interaction of the Z boson with the other weak bosons is reduced compare
to the one of the original W3 boson.
Of course, this changes also the form of the coupling to the neutral fields for matter. In
particular, the symmetry of SU(2) is no longer manifest, and the W 1,2 cannot be treated
on the same footing as the neutral bosons. E. g., the neutral part of the coupling in the
covariant derivative now takes the form
y g y
D(N ) = + ig sin W A t3 + +i Z t3 cos2 W sin2 W .
2 cos W 2
Using the relation (6.13), it is possible to identify the conventional electric charge as
e = g sin W ,
i. e., the observed electric charge is smaller than the hypercharge. It should be noted that
this also modifies the character of the interaction. While the interaction with the photon
is purely vectorial, and the one with the W bosons remains left-handed (axial-vector),
the interaction with the Z boson is now a mixture of both, and the mixing is parametrized
by the Weinberg angle.
Note that the masslessness of the photon is directly related to the fact that the corre-
sponding component of the Higgs-field has no vacuum expectation value,
y 3

+ h0||0i = 0,
2 2
and thus the vacuum is invariant under a gauge transformation involving a gauge trans-
formation of A ,
h0||0i = h0| exp(i(y/2 + 3 ))|0i .
Thus, the original SU(2)U(1) gauge group is hidden, and only a particular combination
of the subgroup U(1) of SU(2) and the factor U(1) is not hidden, but a manifest gauge
symmetry of the system, and thus this U(1) subgroup is the stability group of the elec-
troweak gauge group SU(2)U(1). It is said, by an abuse of language, that the gauge
104 6.8. CP violation and the CKM matrix
group SU(2)U(1) has been broken down to U(1). Since this gauge symmetry is mani-
fest, the associated gauge boson, the photon A , must be massless. If, instead, one would
calculate without the change of basis, none of the gauge symmetries would be manifest.
However, the mixing of the B and the W3 would ensure that at the end of the calculation
everything would come out as expected from a manifest electromagnetic symmetry.
Also, this analysis is specific to the unitary gauge. In other gauges the situation may be
significantly different formally. Only when determining gauge-invariant observables, like
scattering cross-sections or the masses of gauge invariant bound-states, like positronium,
everything will be the same once more.
6.8 CP violation and the CKM matrix

In any strong or electromagnetic process the quark (and lepton) flavor is conserved. E. g.,
the strangeness content is not changing. This is not true for weak processes. It is found
that they violate the flavor number of both, quarks and leptons. In case of the leptons,
this effect is suppressed by the small neutrinos masses involved, but in case of quarks this
is a significant effect.
Considering the weak decays of a neutron compared to that of a strange , it is found
that the relative strengths can be expressed.as
gS=0 = g cos C
gS=1 = g sin C
where g is a universal strength parameter for the weak interactions, its coupling constant.
The angle parametrizing the decay is called the Cabibbo angle. A similar relation also
holds in the leptonic sector for the muon quantum number
g=0 = g cos CL
g=1 = g sin CL ,
where, however, sin CL is almost one, while in the quark sector sin C is about 0.22. Cor-
responding observations are also made for other flavors.
This result implies that the mass eigenstates of the matter particles are not at the
same time also weak eigenstates, but they mix. Hence, on top of the P-violating and
C-violating factors of (1 5 )/2, it is necessary to include something into the interaction
which provides this mixing. This can be done by introducing a flavor-dependent unitary
coupling matrix
(L)

0 cos C 0
g
G = cos C(L) 0
(L)
sin C .
2 (L)
0 sin C 0
This is equivalent to just use a doublet

u d cos C + s sin C ,
e. g., in the quark sector. Such a doublet structure can be associated with (weak) charges
Qu = 1/2 and Qds = 1/2. This is just the weak isospin.
Hence, the flavor (and mass) eigenstates of the quarks are effectively rotated by a
unitary matrix. For two generations, this matrix, the Cabibbo matrix is uniquely given
by !
cos C sin C
VC = ,
sin C cos C
with again the Cabibbo angle C , with a value of about sin C 0.22. For three gener-
ations, there exist no unique parametrization of the mixing matrix, which is called the
Cabibbo-Kobayashi-Maskawa (CKM) matrix. The interaction is thus parametrized as

dL
T +
(uL , cL , tL ) W VCKM sL + h.c.

bL

Vud Vus Vub
u d
VCKM = VL VL = Vcd Vcs Vcb .

Vtd Vts Vtb

The absolute values of the entries have been measured independently, and are given by
|Vud | = 0.9743(3) |Vus | = 0.2252(9) |Vub | = 4.2(5) 103
|Vcd | = 0.23(2) |Vcs | = 1.01(3) |Vcb| = 41(1) 103 .
|Vtd | = 8.4(6) 103 |Vts | = 43(3) 103 |Vtb | = 0.89(7)
Thus, the CKM matrix is strongly diagonal-dominant, especially towards larger quark
masses, and within errors compatible with a unitary matrix.
To make the unitarity more explicit, it is common to recast the CKM matrix into the
following form

c12 c13 s12 c13 ei13 s13
VCKM = s12 c23 ei13 c12 s23 s13 c12 c23 ei13 s12 s23 s13 s23 c13 . (6.16)

s12 s23 ei13 c12 c23 s13 c12 s23 ei13 s12 c23 s13 c23 c13
cij = cos ij
sij = sin ij
106 6.9. Neutrino oscillations and the PMNS matrix
To have only 4 free parameters (12 , 13 , 23 , and 13 ) requires not only unitarity, but
also to exploit some freedom in redefining the absolute phases of the quark fields. Testing
whether this matrix is indeed unitary by measuring the nine components individual is
currently recognized as a rather sensitive test for physics beyond the standard model. The
condition for unitarity can be recast into a form that a sum of three functions of the angles
has to be , forming a triangle. Thus, testing for the unitarity, and thus standard-model
compatibility of CP violations, is often termed measuring the unitarity triangle.
The presence of this matrix also gives rise to the possibility that not only C and P are
violated separately, but that also the compound symmetry CP is violated (and therefore
also T by virtue of the CPT-theorem). That occurs to lowest order in perturbation theory
by a box-diagram exchanging the quark flavor of two quarks by the exchange of two W
bosons.
That such a process violates CP can be seen as follows. The process just described
is equivalent to the oscillation of, e. g., a ds bound state into a sd bound state, i. e., a
neutral kaon K 0 into its anti-particle K 0 . The C and P quantum numbers of both particles
are P= 1, C= 1 and P= 1, C= 1, respectively, and thus CP= 1 and CP= 1. Thus,
any such transition violates CP. Performing the calculation of the corresponding diagram
yields that it is proportional to the quantity
= sin 12 sin 23 sin 13 cos 12 cos 23 cos2 13 sin 13

(m2t m2c )(m2t m2u )(m2c m2u )(m2b m2s )(m2b m2d )(m2s m2d ).
Thus, such a process, and thus CP violation is possible if there is mixing at all (all ij
non-zero) with a non-trivial phase 13 , and the masses of the quarks with fixed charge are
all different. They may be degenerate with ones of different charge, however. Since such
oscillations are experimentally observed, this already implies the existence of a non-trivial
quark-mixing matrix. The value of 13 is hard to determine, but is roughly 1.2.
Within the standard model, there is no explanation of this mixing, however, and thus
these are only parameters.
6.9 Neutrino oscillations and the PMNS matrix

A consequence of the mixing by the CKM matrix is that it is possible to start at some
point with a particle of any flavor, and after a while it has transformed into a particle of a
different flavor. The reason for this is a quantum effect, and proceeds essentially through
emission and reabsorption of a virtual W . In quantum mechanical terms for a two-body
system, this can be easily deduced. Take a Hamiltonian H as

!
H0
H= .
H0
This leads to a time-evolution operator

!
cos(t) i sin(t)
U(t) = eiH0 t .
i sin(t) cos(t)
Under time evolution an initial pure state (1, 0) will therefore acquire a lower component,
if is non-zero. On the other hand, if the composition of a pure state after an elapsed
time is measured, it is possible to obtain the size of . The probability P to find a particle
in the state (0, 1) after a time t is given by
P = sin2 (t). (6.17)
In the standard model, the corresponding expression for the transition probability involves
the mass difference between the two states. To lowest order it is given for the two-flavor
case by 2
2 m L
P = sin sin2 (2),
4E
where m is the mass difference between both states, E is the energy of the original
particle, L is the distance traveled, and is the corresponding Cabbibo angle. If the
probability, the energy and the distance is known for several cases, both the Cabbibo
angle and the mass difference can be obtained. Of course, both states have to have the
same conserved quantum numbers, like electrical charge.
The same calculation can be performed in the more relevant three-flavor case. The
obtained transition probability from a flavor fi to a flavor fj is more lengthy, and given
by
m2kl
X X
2 2
P (fi fj , L, E) = |Vjk | |Vik | + 2 |Vjk Vik Vil Vjl | cos L arg(Vjk Vik Vil Vjl )
2E
k k>l
This sum shows that the process is sensitive to all matrix elements, including the CP-
violating phase, of the CKM matrix, while it cannot determine the sign of the mass
differences.
For the quarks, the first observation of CKM effects were due to decays. Such oscil-
lations have also been found later, and studied to determine the matrix elements more
precisely. E. g., it is possible with a certain probability that a particle oscillates from a
108 6.9. Neutrino oscillations and the PMNS matrix
short-lived state to a long-lived state. This is the case for the so-called K-short KS and
K-long KL kaons, mixed bound states of an (anti-)strange and an (anti-)down quark. This
has been experimentally observed, but the distance L is of laboratory size, about 15 m for
KL and 15 cm for KS , giving a m of about 3.5 1012 MeV. However, in this case the
effect is rather used for a precision determination of the mixing angle, since the mass can
be accurately determined using other means.
In the lepton sector, these oscillations were the original hint for non-zero neutrino
masses. As it is visible, such oscillations only occurs if the involved particles have a mass.
Thus, only with massive neutrinos there was chance to see the effect. But there is no direct
determination of their mass, and the best results so far is an upper limit on the order of 2
eV from the -decay of tritium. Observing these oscillations then indicated the presence
of neutrino masses, and that there is a CKM-like matrix also in the lepton sector, as also
otherwise there would be no mixing - for a unit matrix the above given formula reduces
to the first term only.
With this, it is possible to determine the mass difference of neutrinos. It is found
that |m12 | = 0.0087 eV and |m23 | = 0.049 eV. As only the squares can be determined,
it is so far no possible to establish which neutrino is the heaviest, and if one of them is
massless. Still, the mass difference of 0.05 eV indicates that with an increase in sensi-
tivity by a factor of 40 it can be determined in decay experiments whether the electron
neutrino is the heaviest. If it is, the mass hierarchy is called inverse, as the first flavor,
corresponding to the electron, is lightest. Otherwise, it would be called a normal hierar-
chy. As a side remark, these tiny mass differences imply that the oscillation lengths L are
typically macroscopically, and of the order of several hundred kilometers and more for one
oscillation.
Aside with these mass differences, it was also possible to determine some of the elements
of the CKM matrix in the lepton sector, which is called Pontecorvo-Maki-Nakagawa-
Sakata, or PMNS, matrix. In terms of the parametrization (6.16), the values of the known
three angels are
sin2 12 = 0.31(2)
sin2 23 = 0.4(1)
sin2 13 = 0.025(4)
while the value of 13 was not yet measured, and it is thus not clear whether there is also
CP violation in the lepton sector. However, the values of the other angles imply that with
the sensitivity of current experiments, it should be possible to get a reasonable estimate
of 13 until the mid 2020s.
The values of the three real angles are quite astonishing. They imply that, irrespective
of the value of 13 , the PMNS matrix is not strongly diagonal dominant. Thus, the lepton
flavors mix much stronger than the corresponding quark flavors, and only the very small
values of the neutrino masses reduce the oscillation probabilities so strongly that the effect
is essentially not seen for electrons, muons, and tauons, and even requires the very weakly
interacting neutrinos, which can move over long distances, to observe it at all.
Again, at the current time there is no understanding for why this is the case, nor why
the quark sector and the lepton sector are so extremely different also in this respect, and
not only for the hierarchy of masses.
6.10 The weak interactions as a gauge theory

Combining everything, the electroweak sector of the standard model, the Glashow-Salam-
Weinberg theory is of the following type: It is based on a gauge theory with a gauge group
G, which is SU(2)U(1), weak isospin and hypercharge. There are left-handed and right-
handed fermions included, which are differently charged under the gauge group G. These
are doublets for the left-handed quarks and leptons, and singlets for the right-handed
quarks and leptons for the weak isospin. There is also the scalar Higgs field which is again
a doublet.
To formulate the Lagrangian of the electroweak sector of the standard model requires
hence a number of steps. First of all, the vacuum expectation value for the Higgs field

is chosen to be f / 2. Its direction is chosen such that it is manifest electrically neutral.
That is provided by the requirement
! !
y 1 0 1
3
Q = 0 = + = = 0.
2 2 0 0 2
The field is then split as

!
+
= ,
1 (f + + i)
2
with = (+ )+ , i. e., its hermitiean conjugate. The fields + and carry integer
electric charge plus and minus, while the fields and are neutral. Since a non-vanishing
value of f leaves only a U(1) symmetry manifest, and are the would-be Goldstone
bosons. This leads to the properties of the Higgs fields and vector bosons as discussed
previously.
110 6.10. The weak interactions as a gauge theory
The fermions appear as left-handed doublets in three generations

!
L
i
LLi =
liL
!
L
u i
QLi = ,
dLi
where i counts the generations, l are the leptons e, and , the corresponding neutrinos
e , , , u the up-type quarks u, c, t, and d the down-type quarks d, s, b. Correspondingly
exist the right-handed singlet fields liR , iR , uR R
i , and di . Using this basis the Yukawa
interaction part reads
LY = LLi Glr R r L r R r L ur R r L dr R r
ij lj + Li Gij j Qi Gij uj + Qi Gij dj + h.c..
The matrices G are connected to mass matrices by
1 1 1 1
Mijl = Glij f Mij = Gij f Miju = Guij f Mijd = Gdij f.
2 2 2 2
It is then possible to transform the fermion fields8 f into eigenstates of these mass-matrices,
and thus mass eigenstates, by a unitary transformation
fif L = Uik
fL L
fk (6.18)
fif R = Uik
fR R
fk ,
for left-handed and right-handed fermions respectively, and f numbers the fermion species
l, , u, and d and i the generation. The fermion masses are therefore
1 X fL f
mf i = Uik Gkm (U f R )+
mi f.
2 km
In this basis the fermions are no longer charge eigenstates of the weak interaction, and
thus the matrices U correspond to the CKM matrices. In fact, in neutral interactions
which are not changing flavors always combinations of type U f L (U f L )+ appear, and thus
they are not affected. For flavor-changing (non-neutral) currents the matrices
V q = U uL (U dL )+ (6.19)
V l = U L (U lL )+
8
The vacuum expectation value f of the Higgs field, carrying no indices, should not be confused with
the fermion fields fif , which carry various indices, f denoting the fermion class.
remain, providing the flavor mixing. Finally, the electric charge is given by

e= 4 = g sin W = g cos W ,
with the standard value 1/137.
Putting everything together, the lengthy Lagrangian for the electroweak standard
model emerges:
L = firs (i mf )fi eQr firs firs A (6.20)

e
+ (I 3 frL firL sin2 W Qr firs firs )Z
sin W cos W W r i
e
+ (firL Vijr fjrL W+ + firL (V r )+ rL
ij fj W )

2 sin W
1
| A A ie(W W+ W W+ )|2
4
2
1 cos W + +

Z Z + ie (W W W W )
4 sin W
2
1 + + + + cos W + +

W W ie(W A W A ) + ie (W Z W Z )
2 sin W
2
1 e + ie
+ ( + i) i W + iM Z Z + Z ( + i)
2 sin W 2 cos W sin W
cos2 W sin2 W ie
+ | + + ieA + ie Z + iMW W+ W + ( + i)|2
2 cos W sin W 2 sin W
ef 2 1
f 22 ( + + | + i|2 )
sin W MW 2
2 2
e f 1
2 2
( + + | + i|2 )2
4 sin W MW 2
emri
(frs fi 2IW 3 rs rs
r ifi 5 fi )
2 sin W MW i
e mri rR r rL + rL r + rR
+ (fi Vij fj + fi (V )ij fj )
2 sin W MW
e mri rL r rR + rR r + rL
+ (fi Vij fj + fi (V )ij fj ),
2 sin W MW
3
where a sum over fermion species r is understood and IW is the corresponding weak isospin
quantum number.
112 6.10. The weak interactions as a gauge theory
Note that this Lagrangian is invariant under the infinitesimal gauge transformation
A A + A + ie(W+ W + ) (6.21)
cos W
Z Z + Z ie (W+ W + )
sin W
e
W W + i (W (sin W A cos W Z ) (sin W u cos W Z ) )
sin W
e e
+ Z + i (+ + )
2 sin W cos W 2 sin W
e e
(v + )Z + (+ + + )
2 sin W cos W 2 sin W
sin2 W cos2 W Z

A ie
ie + (f + i)
2 cos W sin W 2 sin W

f L f L A sin W f 1 ie
fi fi ie Qi + Qi 2 Z fiL + Vij fjL
cos W 2 sin W 2 sin W

sin W Z
fif R ieQfi A + fif R . (6.22)
cos W
The index for the left-handed fermion fields counts the isospin directions. The in-
finitesimal gauge functions are determined from the underlying weak isospin i and
hypercharge Y gauge transformations by
1 1
= ( i2 )
g
1 1
A = cos W Y
sin W 3
g g
1 1
Z = cos W 3 + sin W Y .
g g
It is now straightforward to upgrade the Lagrangian of the electroweak sector of the
standard model (6.20) to the full Lagrangian of the standard model by adding the one for
the strong interactions
1 rs a rs a
Ls = Ga G a g fi fi G
4
Ga = Ga Ga + g habc Gb Gc ,
where the generators a and the structure constants habc belong to the gauge group of
the strong interactions, the so-called color group SU(3)9 , and g is the corresponding
coupling constants. Ga are the gauge fields of the gluons, and the fermions now have also
an (implicit) vector structure in the strong-space, making them three-dimensional color
vectors or singlets, for quarks and leptons, respectively.
9
Actually SU(3)/Z3 , similar to the case of the weak isospin group.
Note that the number of Higgs fields is larger than the minimal number required: Only
two (real) degrees of Higgs are necessary to write down a consistent theory of the weak
interactions. However, in this case only one of the weak gauge bosons would become mas-
sive. Hence, experiments requires four degrees of freedom. This introduces an additional
global symmetry between the four Higgs degrees of freedom, similar to the flavor symme-
tries of the fermions. It is called a custodial symmetry, as in absence of other interactions
it would lead to mass-degenerate W and Z bosons. However, QED breaks this symmetry
explicitly, and leads, as has been shown above, to the slight mass difference between W
and Z bosons.
6.11 Anomalies
As noted in section 4.9, it is possible that a symmetry is broken by the quantization.
Such a breaking is called an anomalous breaking, mainly for the reason that this was
not expected to occur in the beginning. Today, it is clear that such a breaking is deeply
linked into the quantization process, and is in most cases also related to the necessity for
renormalization.
It is now possible to consider an anomaly for a global or a local symmetry. Anomalies
for global symmetries do not pose any inherent problems. Indeed, the axial part UA (1)
of the chiral symmetry described in section 4.12 is broken anomalously in the standard
model. This breaking is directly observable consequences. In that particular case, it leads
to a much larger partial decay width of the 0 into two photons, as it would be without
this anomaly. Also the large mass of the meson, as indicated in section 5.11 can be
shown to originate from the same anomaly.
The situation is much more severe for an anomalous breaking of a local symmetry. If
this occurs, this implies that while the classical theory is gauge-invariant, the quantum
version would not be so. Hence, the quantum version would depend on the gauge chosen
classically, and observables would become gauge-dependent. This would not yield a con-
sistent theory, and would therefore be discarded as a theory of nature. Thus, only theories
without local anomalies are considered for particle physics.
Considering theories of the type the standard model, i. e. a number of gauge fields with
fermions and scalars, it turns out that the possibility of gauge anomalies is tied to the
presence of fermions. Theories with only gauge fields and scalars do not develop anomalies.
Furthermore, it can be shown that only theories with parity violations can be affected by
local anomalies. Finally, anomalies can only occur for certain gauge groups10 , and not for
10
In total, the Lie groups U(1), SU(N > 2), Sp(N ), and O(2 < N < 6) are affected.
114 6.11. Anomalies
arbitrary ones, and only for particular assignments of charges to the fermions. However,
this affected combination of charge assignments and gauge groups includes the one of the
standard model. Thus, in principle, the standard model would be anomalous.
Fortunately, there is an escape route left. Though indeed such anomalies occur, it is
possible that if there are more than one anomaly, the consequences of these anomalies
cancel each other. If such an anomaly cancellation occurs, the theory becomes anomaly-
free, and is again well-defined. Given the gauge groups and charge assignments of the
standard model, the condition that all anomalies are canceled and the standard model is
consistent is

X 2 1 Nc
Qf = Ng (0 1) + Nc = Ng 1 + = 0,
f
3 3 3
where the sum is over all fermion flavors, quarks and leptons alike, ng is the number of
generations, Qf is the electric charge of each flavor f , and Nc is the number of colors. As
can be seen, the anomalies indeed cancel, and they do so for each generation individually.
However, the cancellation requires a particular ratio of the quark electric charges and the
lepton electric charges as well as the number of colors. Since the anomalies originate from
the parity-violating interaction, all three forces are involved in guaranteeing the anomaly-
freedom of the standard model.
This fact has led to many speculations. The most favored one is that this indicates an
underlying structure, which provides a common origin to all three force of the standard
model. Such theories are known as grand-unifying theories (GUT). Such theories have
one underlying gauge group, which would be spontaneously broken at high energies, and
therefore appear like three separate forces at low energies. This reasoning is supported by
the fact that the running gauge couplings of all three forces all approach a similar size at
about 1015 GeV, the GUT scale, which suggests a common origin at this energy scale.
A second explanation is that only a theory like the standard model, which such a fine
balance, has all the features necessary to permit sentient life to arise to recognize this
feature. In a universe with different laws of nature, where this kind of balance is not
met nobody would be there to observe it. This is called the anthropomorphic principle.
Though it cannot be dispelled easily, it is not as compelling an argument, as it does not
explain why such a universe should exist at all. This problem is often circumvented by
the requirement that actually many universes with all kinds of laws of nature exist, and
we just happen to be in one where we could exist.
The third possibility is, of course, that all of this is just coincidence, and nature is just
this way.
At any rate, future experiments will decided, which of these three options is realized. In
fact, experiments have already ruled out the simplest candidates for grand-unified theories,
though many remain.
6.12 Landau-poles, triviality, and naturalness

As has been noted, the three gauge couplings of the standard model depend on energy.
The strong and weak coupling diminish at high energies, and both interactions are asymp-
totically free. However, both have very different energy behaviors: The strong interaction
becomes stronger towards the infrared, while the weak interaction reaches a maximum,
and then diminishes also towards small energies, mostly due to the screening by the Higgs
effect. The electromagnetic coupling, however, rises towards large energies.
Assume for a moment that the three coupling unify at some point, as discussed in
the previous section 6.11, and that the resulting GUT is asymptotically free. Hence, the
high-energy behavior of the gauge interactions becomes well-defined. Then there are still
13 non-gauge interactions in the standard model left: The four-Higgs self-interaction, and
the 12 Yukawa couplings between the Higgs and the fermions11 . As all these interactions
are not of a gauge type, they are all not asymptotically free. Hence, all of them rise with
energy, though even the largest one, the top-Higgs-Yukawa coupling, very slowly. As a
consequence, perturbation theory fails at some scale due to the emergence of Landau poles
in these couplings12 .
This rise is strongly dependent on the mass of the Higgs. If the Higgs mass increases,
the couplings start to rise quicker. For a Higgs mass of around 1 TeV, perturbation theory
will already fail at 1 TeV, as then the four-Higgs coupling develops the Landau pole at the
same energy. At the same time a Higgs of this mass will have a decay width of the same
order as its mass. This shows that the physics in the Higgs sector is very sensitive to the
values of the parameters, in fact quadratically. This is in contrast to the remainder of the
standard model. Both the fermion and gauge properties are only logarithmically sensitive
to the parameters.
To achieve that the Higgs mass is of the same order as the W and Z mass requires
that the parameters in the standard model are very finely tuned, to some ten orders of
magnitude. This is the so-called fine-tuning problem. In fact, if random values for the
11
It is assumed for simplicity here that all neutrinos are massive, though the experimental data are still
compatible with one of them being massless.
12
Occasionally one finds in the literature the argument of perturbativity, i. e. that any decent theory
describing high energies must have weak coupling, and be perturbatively treatable. However, there is no
reason for this assumption except technical ones. Hence, such statements express a hope rather than any
reasonable physics.
116 6.13. Baryon-number violation
couplings would be taken, the Higgs mass was would usually end up at masses orders of
magnitudes larger than the W and Z mass. Why this is not the case is called the hierarchy
problem: Why is there no large hierarchy between the electroweak scale and the Higgs
mass scale? This is sometimes rephrased as the naturalness problem: Without any prior
information, it would naively be expected that all parameters in a theory are of the same
order of magnitude. This is not the case in the standard model. Why do the values deviate
so strongly from their natural value?
There is another problem connected with these questions. If only the Higgs sector is
regarded, the theory is called a 4 theory, as it has the same structure as the Lagrangian
(5.5), except that there are four scalar fields. It is found that in such a theory quantum
effects always drive the self-interaction term in the potential to zero, i. e. the quantum
theory is non-interacting, even if the classical theory is interacting. Theories in which
this occurs are called trivial. Such a triviality can be removed, if the theory is equipped
with an intrinsic energy cutoff, which introduces an additional parameter, but restricts the
validity of the theory to below a certain energy level. It is then just a low-energy-effective
theory of some underlying theory.
It is not clear whether this triviality problem persists when the Higgs sector is coupled
to the rest of the standard model. If it does, the necessary cut-off for the standard model
will again depend sensitively on the mass of the Higgs boson. Practically, this is not a
serious issue, as with the present mass of the Higgs of roughly 125 GeV this triviality
cut-off can be easily as large as 1019 GeV, far beyond the current reach of experiments
and observations. Still, it remains as a doubt on how fundamental the standard model,
and especially the Higgs particle, really is, in accordance with the problems introduced by
the necessity of renormalization as described in section 3.7.
6.13 Baryon-number violation

Our current understanding of the origin of the universe indicates that it emerged from a big
bang, a space-time singularity where everything started as pure energy. Why is then not an
equal amount of matter and anti-matter present today, but there is a preference for matter?
CP violation explains that there is indeed a preference for matter over anti-matter, but
the apparent conservation of lepton and baryon number seems to indicate that this is only
true for mesons and other states which do not carry either these quantum numbers. This
impression is wrong, as, in fact, there is a process violating baryon (and lepton) number
conservation in the standard model. Unfortunately, both this process and CP violation
turn out to be quantitatively too weak to explain with our current understanding of the
early evolution of the universe the quantitative level of the asymmetry between matter
and anti-matter. Thus, the current belief is that so far undiscovered physics is responsible
for the missing (large) amount.
It is nonetheless instructive to understand how baryon number violation comes about
in the standard model. Lepton number violation proceeds in the same way, but is even
more suppressed, due to the much smaller masses.
The basic ingredient is once more a classical field configuration, this time of Yang-Mills
theory. Define the field strength tensors F = a Fa
, with a the Pauli matrices - here the
weak interactions with the gauge group SU(2) is the interesting one. To proceed further,
it is useful to make the formal replacement it t, which can be undone at the end. This
is an analytic continuation from Minkowski space-time to Euclidean space-time, as now
all components of the metric have the same sign.
The Jacobi identity (5.13) together with the non-Abelian version of the homogeneous
Maxwell equation of QED,
F + F + F = 0,
yield the Bianchi identity
D F + D F + D F = 0,
D F = Dij jk
a a
F .
Together, they imply

1
0 = D F = D F
2
1
F = F ,
2
where F is called the dual field-strength tensor. The correctness of these statements
follow from the anti-symmetry of the field-strength tensor and the Jacobi identity (5.13).
However, the (inhomogeneous) Maxwell equation
D F = 0, (6.23)
are therefore trivially solved by (anti-)self-dual solutions
F = F , (6.24)
as these convert the equation (6.23) into the trivial Bianchi identity. The self-duality
equations (6.24) have the advantage of being only first-order differential equations instead
of second-order differential equations, and are therefore easier to solve.
118 6.13. Baryon-number violation
Furthermore, classical solutions have to have a finite amount of energy, and therefore
their behavior at large distances is constrained. Especially, since the Lagrangian can be
written as the sum of the squares of the electric and magnetic field strength, both these
fields must vanish. This can only occur if the potential becomes at large distances gauge-
equivalent to the vacuum, i. e. it has the form
Aa a = A = ig(x) g 1(x),
where g = g a a is an arbitrary function. Since this is the second part of the gauge
transformation (5.17) in matrix notation, this is just a version of the vacuum A = 0.
Since all choices are gauge-equivalent, any choice will do. One possibility which turns out
to be technically convenient is
x
g(x) =
|x|
= (1, i a ). (6.25)
The simplest extension of this is a multiplication with a function f (x2 ) which becomes 1
at large distances,
x
Aa a = if (x2 )g(x) g 1(x) = 2f (x2 )
x2
1
= ( )
4i
= (1, i a ).
The matrices are called t Hooft symbols. Thus, this ansatz mixes non-trivially the
weak isospin and space-time.
Plugging this in into the self-duality equation (6.24) yields a first-order differential
equation for f (x2 ),
df
x2 f (1 f )) = 0.
dx
The solution to this equation, which can be obtained by separation of variables, is
x2
f (x2 ) = ,
x2 + 2
where is an integration constant. The function indeed goes to one at large distances, as
required. The structure described by this field configuration is now localized in space-time
at the origin, and extended over a range of size . Such a localized event in space and time
is called an instanton. Solving with the equation with the other sign in the self-duality
equation (6.24) yields a similar result, though with some small differences, and is called
an anti-instanton.
Going back to Minkowski space-time, the field configuration will have a singularity
at x2 = 2 . This is called a sphaleron, and hence a violent event in space-time. Note
that the gauge-field strength does not explicitly appear in the calculation. This result
can therefore not be obtained perturbatively, and the presence of instantons is a non-
perturbative effect.
Seeing now that these indeed create baryon number violation is unfortunately tech-
nically very complicated, so here only the most important steps will be sketched. One
highly non-trivial, but very fundamental, insight needed is that instantons turn out to be
very much connected with the anomalies of section 6.11. In fact, any instanton induces
via the global anomaly interactions between fermions. Especially, it can connect fermions
of different types, as long as all are charged and all are affected by the anomaly. Espe-
cially, this yields that instantons effectively create an interaction which involves, besides
the gauge fields also three quarks and one lepton, and thus permits baryon number viola-
tion. However, this integration still conserves fermion number, and as a consequence the
change in the baryon number must be offset by the same change in lepton number. Still,
this implies that reducing the baryon number by one can be offset by a change in lepton
number by one, which implies that a proton can be converted, with the involvement of
gauge fields, into a positron. Hence, baryon (and lepton number) are not conserved in the
standard model.
The effect is, however, small, and suppressed exponentially by exp(c/W ), where c
is a number, and W is the weak isospin fine-structure constant. Since the latter is small,
the suppression is huge, and the life-time of the proton in the standard model exceeds the
current upper experimental upper limit for the decay in any channel of 2.1 1028 years by
many orders of magnitude. Hence, the baryon number violation in the standard model is
not able to explain the fact that so much more baryons than anti-baryons exist. However,
this is a statement about the current state of the universe, and especially its temperature,
and it may change in earlier times. Especially, it can be shown that the effect becomes
exponentially enhanced with temperature. Still, the temepratures necessary to make this a
sucfficiently effective process have been available for too short times in the early universe.
6.14 The early universe

The weak interactions, and particularily the Higgs effect, play also another important role
in the early universe. Just as the magnetization in a magnet can be removed by heating it
up, so can the Higgs condensate melt at high temperatures, making the symmetry manifest
once more. This process is different in nature from the effectively manifest symmetry at
120 6.14. The early universe
large energies, which is a quantitative effect as all masses become negligible. However,
there is not necessarily a phase transition associated with the melting of the condensate.
In fact, it is in general not a symmetry restoration, as in case of a global symmetry to
be discussed here as well. This is due to the fact that the symmetry is just hidden, not
broken. As such, there is no local gauge-invariant order parameter associated with it.
Only gauge-dependent order parameters can be local, but in their case the temperature
where the symmetry becomes manifest once more is in general gauge-dependent. Only a
transition which would be indicated by a non-local gauge-invariant order parameter could
in principle mark a true phase transition.
Again, it is quite useful to first study the case of a global symmetry, then of an Abelian
local symmetry before going to the electroweak theory.
6.14.1 Global symmetry

A useful starting point is given by the Lagrangian (5.8)
1 1 1 1
L = (6f 2 2 ) 2 + (2f 2 2 )2
2 2 2 2
1
2f ( 2 + 2 ) ( 2 + 2 )2 2 f 2 + f 4 . (6.26)
4
In this case the explicit zero-energy contribution is kept for reasons that will become
apparent shortly, but will be essentially the same as when treating non-relativistic Bose-
Einstein condensation. Only terms linear in the fields have been dropped, as they will not
contribute in the following. The situation is similar as before, but now the condensate f
has not been specified by the minimization of the classical potential, but is kept as a free
quantity, which will take its value dynamically.
To investigate the thermodynamic behavior it is useful to analyze the thermodynamic
potential in analogy to the non-relativistic case as
Z
(T, f ) = P (T, f ) = T ln ,
V
where P is the pressure, T the temperature, and V the volume. Z is the partition sum.
For the following purposes, it is sufficient to use the so-called mean-field approximation. In
this case, the interaction terms are first expanded around f , and higher-order terms which
involve more than two fields are neglected. In the Lagrangian (6.26), this has already been
done.
Without going into the technical details, the thermodynamic potential can be evaluated
directly. It reads
(T, f ) = 2 f 2 + f 4 (6.27)
3
2 2

dp 1 + 2
Z
1 2
+ 3
+ T ln 1 e T + ln 1 e T
(2) 2
p q
1 = 6f + p = m2 + p2
2 2 2
p q
2 = 2f 2 2 + p2 = m2 + p2 .
The frequencies consists of the momenta and the masses of the particles after hiding
the symmetry, which is dependent on the value of the condensate f . There are three
contributions. The first outside the integral is the classical contribution. The second
are the first two terms inside the integral. They are the contributions from quantum
fluctuations. The third term represents thermal fluctuations.
To recover the results from section 5.7, the second term must be neglected and the
zero-temperature limit taken. This yields
(0, f ) = 2 f 2 + f 4 .
As in section 5.7, this potential has a minimum at non-zero f , f = 2 /(2). Inserting this
into the Lagrangian (6.26) makes it equivalent to (5.8), with a massive Higgs boson and a
massless Goldstone boson.
Something new happens at finite temperature. At small temperature, little other
happens than that it is possible to excite Higgs bosons or Goldstone bosons, which then
form a thermal bath of non-interacting bosons, and the total pressure is just the sum of
their respective pressures. However, the value of f will become temperature-dependent:
At each temperature it will take the value which minimizes the thermodynamic potential.
When going to higher temperatures, it is useful to make a high-temperature expansion
for the thermodynamic potential. High temperature requires here T to be larger than the
scale of the zero-temperature case, which is given by the condensate, which is of order

/ . In this case, it is possible to obtain an expansion for . The leading terms up to
order O(1) are given by
2 2 T 2

4 1 2
(T, f ) = f + T f 2 T 4
2
. (6.28)
3 45 12
Note that at very large temperatures only the term 2 T 4 /45 is relevant, which is pre-
cisely the one of a free non-interacting gas of two boson species, a Stefan-Boltzmann-like
behavior.
This results exhibits one interesting feature. The term of order f 2 has a temperature-
dependent coefficient, which changes sign at13 Tc2 = 32 /. As a consequence, the shape of
the thermodynamic potential as a function of f changes. Below Tc , it has a minimum away
from zero, as at zero temperature. With increasing temperature, this minimum moves to
smaller and smaller values, and arrives at zero at Tc . Hence, at Tc , the value of f changes
from a non-zero to a zero value, and the symmetry becomes manifest once more. Above
Tc , the minimum stays at zero, and for all higher temperatures the symmetry is manifest.
Replacing f with its temperature-dependent value in (6.28) yields the expressions
2 2
2
T =T 2 2
T <Tc = 12T 45 + 36
T4 = c 2
5
2 2
4 2 2 2 T =T
T >Tc = 4 45 T 4 12T = c
2,
5
which coincide at Tc . Also their first derivatives with respect to the temperature are equal
at Tc
r
2
dT <Tc T T =T 8 + 5 2
= (8 2 T 2 + 10T 2 152) 90 c
=
dT 300
r
dT >Tc T T =Tc 8 2 + 5 2
= (8 2 T 2 + 152) 90 = ,
dT 300
but their second derivatives are not
d2 T <Tc 2 2 T 2 T =Tc (25 + 24 2 ) 2
= 6
(4 + 5) 15
=
dT 2 30
2
d T >Tc 8 2 T 2 +52 T =Tc (5 + 24 2 ) 2
= 30
= .
dT 2 30
Thus, a phase transition of second order occurs at Tc .
As stressed previously repeatedly, it is possible that quantum effects could modify the
pattern considerably or even melt the condensate. It is therefore instructive to investigate
the leading quantum corrections to the previous discussion.
This is also necessary for another reason. If the symmetry becomes manifest once more
at large temperatures, the mass of Higgs-like excitations become tachyonic, indicating a
flaw of the theory. That can be seen directly by reading off the condensate-dependent
masses of the excitations,
m2 = 6f 2 2 = 2 (T Tc ) + (22 T 2 )(Tc T )
T 2
m2 = 2f 2 2 = 2 (T Tc ) (Tc T ).
3
13
Note that strictly speaking using the high-temperature expansion at this temperature is doubtful. For
the purpose here it will be kept since it makes the mechanisms more evident then the rather technical
calculations necessary beyond the high-temperature expansion. The qualitative outcome, however, is not
altered, at least within the first few orders of perturbation theory.
Furthermore, also the Goldstone theorem is violated, as the mass of the Goldstone boson
is no longer zero14 . Both problems are fixed by quantum corrections, demonstrating the
importance of quantum fluctuations even in the high-temperature phase.
In the expression for the free energy (6.27) the zero-point energy, and thus the quantum
fluctuations have been neglected. Including them yields the result
3
2 2 T 2 (m3 + m3 )T 4 8 2 T 2 e2+ 2
(T, f ) = T 4 + ln
45 12 12 32 2 2
8 2 T 2 e2+1 T 2

2 f 2 1 + + 2 ln
4 2 32
8 2 T 2 e2+1

5
+f 4 1 + + 2 ln .
8 2
Herein is the Euler number. The term is a quantum correction, which can be fixed, e.
g. by measuring Tc or by other experimental input, as can be seen from the new expression
for Tc where f vanishes,
32 24 2 e2+1

2
Tc = 1+ + 2 ln .
4
Here, only the qualitative result is interesting, and its precise numerical value of no im-
portance.
To obtain the corrections for the masses, it is necessary to calculate the corresponding
quantum corrections. Without going into the details, the complete masses to this order
are
T 2 32

2 2 1 2
m = 2 1 (Tc T ) + T (T Tc )
32 3
32

2 2
m = T (T Tc ).
3
These results yield a number of interesting observation. First, since Tc is larger15 than
32 /, the mass of the Higgs is always positive, stabilizing the system. Secondly, in this
case the mass of the Goldstone boson is always zero below the phase transition tempera-
ture, in agreement with the Goldstone theorem. Above the phase transition, the masses of
both particle degenerate, and the symmetry is manifest once more also in the spectrum.
14
In a full quantum treatment, the role of the Goldstone boson could be played at finite temperature
by some composite excitation instead. However, at the mean-field level no such excitations are available,
and thus the Goldstone theorem is violated.
15
It is not obvious that cannot be negative and large, thus making the improved estimate for Tc
smaller than before. However, it turns out not to be the case at this order for any renormalization
prescription.
These properties are generic for symmetries hiding by a condensate which thaws with in-
creasing temperature. Also that the mean-field approximation is in general insufficient is
a lesson which should be kept duly in mind. Of course, at the present time much more
sophisticated methods are available to treat this problem, though they are in general very
complicated.
6.14.2 Abelian case

As was visible in the previous case of the global symmetry, quantum fluctuations are
important to obtain a consistent result. Again going through the mean-field calcula-
tions (introducing only a mean-field for the Higgs fields condensate, but neither for the
pseudo-Goldstone boson nor for the photon), it is possible to obtain once more a high-
temperature expansion for the free energy. The full expression at mean-field level then
reads
e2 2 T 2 2 2 4

4
(f, T ) = f + + T f2
2 2
T ,
3 4 12 45
where e is now the electromagnetic coupling constant. First, the Stefan-Boltzmann contri-
bution proportional to T 4 is now twice as large, as the photon also contributes two scalar
degrees of freedom. Furthermore, the leading high-temperature behavior is not altered
otherwise by the presence of the interaction. At the level of the mean-field approximation
there is still only a non-interacting gas of two scalars and two photon polarizations left.
However, the term crucial for the phase transition, the one of order T 2 , is affected by
the interactions. Thus, the interactions have an influence on the phase transition. At
mean-field level they just shift the phase transition temperature.
Beyond mean-field, their impact is more relevant. At one-loop level, the phase structure
becomes dependent on the relative size of and e.
As long as is larger16 than e, the situation is found to be in agreement as when
the photon field would be absent. In particular, the condensate melts at a (now also e-
dependent) critical temperature, and above the phase transition both the scalars and the
photon become massless again.
If is smaller than e, the critical temperature becomes lower than the effective mass
of the photon. As a consequence, the second-order phase transition may in fact become
first order. Even more drastic, if is below 3e4 /(32 2 ), the phase transition temperature
decreases to zero, and spontaneous breaking is not occurring at all. In that case, the
stronger photon-scalar interactions stabilize the vacuum, and no condensate can form.
16
If e and are approximately equal, even higher order corrections can have a qualitative impact.
An important insight is found in the case of = 0. At first sight no condensation is

possible. However, the interactions mediated by the photon may still be sufficiently strong
and attractive enough that the scalars condense, yielding the same physics as if would
not be zero, and would be sufficiently large. In this case, the condensation is a genuine
quantum effect, as only due to (one-)loop quantum corrections spontaneous condensation
of the Higgs occurs.
6.14.3 The electroweak case

Also in the electroweak case it is useful to use the Lagrangian in the t Hooft gauge,
given by (6.20). Though straight-forward, the calculation in the present case is much
more cumbersome than in the previous cases. For the present purpose, the masses of
the fermions (and thus the CKM matrices) can be neglected. The imprecision of this is
comparable to the effects of going to the next order. The final result at mean-field level is

427 2 4 f 2 T2 3g 2 g 2 f 4 4 2 T 2

2
(f, T ) = T 2 + + + + .
360 2 4 4 4 4 4 6
Extremalizing this expression with respect to f yields the critical temperature for the
electroweak standard model in this approximation as
42
Tc2 = .
3g 2 g 2
2 + 4
+ 4
The temperature dependence of the condensate f and the pressure then read, respectively,
2 T2

2
f = 1 2 (Tc T )
Tc
2
4272 T 4 2 T2 2 2 2 4
360
+ 4 1 T 2 + f6 T 4 , T Tc
P = 2T 4
c
2T 2 4
.
427

360
+ 6
, 4
T T c
The corresponding phase transition is at mean-field level thus of second order, as the
second derivative of the pressure exhibits a discontinuity. For a Higgs mass of 100 GeV
the critical temperature is about 200 GeV. For the actual value of the Higgs mass of 126
GeV, it is only slightly higher. Thus the transition temperature is of the same order as
the Higgs condensate, and about three orders of magnitude larger than the corresponding
temperature in QCD.
Note that all problems with consistency of the mean-field approach pertain also to the
full electroweak standard model. Therefore, for a consistent treatment at least leading or-
der corrections have to be included. As previously, they do not change the phase transition
temperature, but may change its order.
The relevance of such a temperature is only given in the early universe, or perhaps
during a collapse to a black hole. Here, the more certain case of the early universe will be
treated.
To assess the relevance for the early universe it is necessary to add equations which
describe its development. For the present purpose simplified versions of the Einstein
equations are sufficient. Adding energy conservation gives
2
dR 8G 2
= R (6.29)
dt 3
d(R3 )
= 3P R2 , (6.30)
dR
where G is Newtons constant, is the energy density, R is the scale factor, essentially
given by the Ricci curvature scalar, t the proper time, and P is the pressure. To close the
equations, an equation of state is necessary, which is given as a function of the pressure
by thermodynamic relations as
P
= P + T .
T
Since the sicknesses of the mean-field approximation are not too problematic for this
estimate, it is sufficient to use it for obtaining the corresponding energy density as
( 2 2
1281 2 4 32
360
+ 4T 4 T + 1 T 2 6T ,
4
T Tc
= 2
c
2T 2 4
c .
1281 4
360
T + 6
+ 4
, T Tc
Rewriting equation (6.30) in terms of the temperature yields the ordinary differential
equation
d dR
R +3 = 3P.
dT dT
Imposing as a boundary condition that R should be one at the phase transition yields
(
Tc Tc2 b2
3 T T 2 b2
, T Tc
R = Tc Tc2 +a2
T T 2 +a2
, T Tc
302
a2 =
427 2
a2 Tc2 (1 r)
b2 =
a2 + rTc2
4
r = 2 2 .
6 + 9g4 + 3g4
For a Higgs mass about 100 GeV the characteristic parameters are r = 0.22, a = 6 GeV and
b = 10 GeV. Thus, the dominant behavior is that R behaves like Tc3 /T 3 , up to some small
modifications close to the phase transition, and thus drops essentially in the electroweak
domain.
An interesting consequence is obtained if R3 is multiplied by the entropy
P
s= T 3.
T
An elementary calculations yields thus that sR3 is constant. Since s is in units inverse
length cubed, this is just the statement that entropy is conserved since R only describes
the expansion of a unit length over time. Hence, the electroweak interactions at mean-
field level conserve entropy. In particular, this is a consequence of the second order phase
transition, which is not permitting latent heat or supercooling. Finally, inserting the
numbers shows that R increases somewhat slower around the phase transition. Thus, the
expansion of the universe slows down during the electroweak phase transition.
Of course, all of this is just an estimate. That it can never be fully correct is seen by
the fact that R diverges at the finite temperature T = b, much above the QCD phase tran-
sition (and nowadays) temperature. This is an artifact of the high-temperature expansion
involved. To obtain the correct behavior down to the QCD phase transition would require
more detailed calculations.
Chapter 7
Beyond the standard model
At the end of 2009 the largest particle physics experiment so far has been started, the
LHC at CERN. With proton-proton collisions at a center-of-mass energy of up to 14 TeV,
possibly upgraded to 33 TeV by the end of the 2020ies, there are two major objectives.
One was to complete the current picture of the standard model of particle physics by
finding the Higgs boson. This has been accomplished, even though its properties remain
still to be established with satisfactory precision, to make sure that it is indeed the Higgs
boson predicted by the standard model.
The second objective is to search for new physics beyond the standard model. For var-
ious reasons it is believed that there will be new phenomena appearing in particle physics
at a scale of 1 TeV, and thus within reach of the LHC. Though this is not guaranteed,
there is motivation for it.
In fact, there are a number of reasons to believe that there exists physics beyond the
standard model. These reasons can be categorized as being from within the standard
model, by the existence of gravity, and by observations which do not fit into the standard
model. Essentially all of the latter category are from astronomy, and there are currently
essentially no observations in terrestrial experiments which are reproducible and do not
satisfactorily agree with the standard model, and none which disagree with any reasonable
statistical accuracy.
Of course, it should always be kept in mind that the standard model has never been
completely solved. Though what has been solved, in particular using perturbation theory,
agrees excellently with measurements, it is a highly non-linear theory. It cannot a-priori
be excluded that some of the reasons to be listed here are actually completely within the
standard model, once it is possible to solve it exactly.
Many of the observations to be listed can be explained easily, but not necessarily, by
new physics at a scale of 1 TeV. However, it cannot be exclude that there is no new
128
Chapter 7. Beyond the standard model 129
phenomena necessary for any of them up to a scale of 1015 GeV, or possibly up to the
Planck scale of 1019 GeV, in which case the energy domain between the standard model
and this scale is known as the great dessert.
7.1 Inconsistencies of the standard model

There are a number of factual and perceived flaws of the standard model, which make it
likely that it cannot be the ultimate theory.
The one most striking reason is the need of the renormalization, as described in section
3.7: It is not possible to determine within the standard model processes at arbitrary
high energies. The corresponding calculations break down eventually, and yield infinities.
Though we have learned how to absorb this lack of knowledge in a few parameters, the
renormalization constants, it is clear that there are things the theory cannot describe.
Thus it seems plausible that at some energy scale these infinities are resolved by new
processes, which are unknown so far. In this sense, the standard model is often referred
to as an low-energy effective theory of the corresponding high-energy theory.
This theory can in fact also not be a (conventional) quantum field theory, as this flaw
is a characteristic of such theories. Though theories exist which reduce the severity of
the problem, supersymmetry at the forefront of them, it appears that it is not possible
to completely resolve it, though this cannot be excluded. Thus, it is commonly believed
that the high-energy theory is structurally different from the standard model, like a string
theory. This is also the case when it comes to gravity, as discussed in the next section.
There are a number of aesthetic flaws of the standard model as well. First, there
are about thirty different free parameters of the theory, varying by at least ten orders
of magnitude and some of them are absurdly fine-tuned without any internal necessity.
There is no possibility to understand their size or nature within the standard model, and
this is unsatisfactory. Even if their origin alone could be understood, their relative size is
a mystery as well. This is particularly true in case of the Higgs and the electroweak sector
in general. There is no reason for the Higgs to have a mass which is small compared to the
scale of the theory from which the standard model emerges. In particular, no symmetry
protects the Higgs mass from large radiative corrections due to the underlying theory,
which could make it much more massive, and therefore inconsistent with experimental
data, than all the other standard model particles. Why this is not so is called the hierarchy
problem, despite the fact that it could just be accidentally so, and not a flaw of the theory.
Even if this scale should be of the order of a few tens of TeV, there is still a factor of possibly
100 involved, which is not as dramatic as if the scale would be, say, 1015 GeV. Therefore,
130 7.1. Inconsistencies of the standard model
this case is also called the little hierarchy problem.

There is another strikingly odd thing with these parameters. The charges of the leptons
and quarks need not be the same just because of QED - in contrast to the weak or strong
charge, actually. They could differ, and in particular do not need to have the ratio of small
integer numbers as they do - 1 to 2/3 or 1/3. This is due to the fact that the gauge group
of QED is Abelian. However, if they would not match within the experimental precision
of more than ten orders of magnitude, and also the number of charged particles under the
strong and weak force were not perfectly balanced in each generation, then actually the
standard model would not work, and neither would physics with neutral atoms. This is
due to the development of a quantum anomaly, i. e., an effect solely related to quantizing
the theory which would make it inconsistent, see section 6.11. Only with the generation
structure of quarks and leptons with the assigned charges of the standard model, this
can be avoided. This is ultimately mysterious, and no explanation exists for this in the
standard model, except that it is necessary for it to work, which is unsatisfactory.
There is also absolutely no reason inside the standard model, why there should be
more than one family, since the aforementioned cancellation works within each family
independently and is complete. However, at least three families are necessary to have inside
the standard model CP violating processes, i. e. processes which favor matter over anti-
matter, see section 6.8. As discussed later, such processes are necessary for the observation
that the world around us is made from matter. But there is no reason, why there should
only be three families, and not four, five, or more. And if there should be a fourth family
around, why would its neutrino be so heavy, more than 100 GeV, compared to the other
ones, as can already be inferred from existing experimental data.
Another particle, however, could still be beyond the standard model: The Higgs boson.
As the properties of the observed candidate are not yet beyond doubt, there is no certainty,
however expected and likely, that it really is the Higgs. If it is not, the whole conceptual
structure of the electroweak sector of the standard model immediately collapses, and is
necessarily replaced by something else. Therefore, despite the fact that a relatively light
Higgs boson is in perfect agreement with all available data, this is not necessarily the case
in nature.
As an aside, there is also a very fundamental question concerning the Higgs sector. At
the current time, it is not yet clear whether there can exist, even in the limited sense of
a renormalizable quantum field theory, a meaningful theory of an interacting scalar field.
This is the so-called triviality problem. So far, it is only essentially clear that the only
consistent four-dimensional theory describing only a spin zero boson is one without any
interactions. Whether this can be changed by adding additional fields, as in the standard
model, is an open question. However, since this problem can be postponed to energy
scales as high as 1015 GeV, or possibly even higher, in particular for a Higgs as light as
the observed one, this question is not necessarily of practical relevance.
Finally, when extrapolating the running gauge couplings for a not-to-massive Higgs to
an energy scale of about 1015 GeV, their values almost meet, suggesting that at this scale
a kind of unification would be possible. However, they do not meet exactly, and this is
somewhat puzzling as well. Why should this seem to appear almost, but not perfectly so?
7.2 Gravity
7.2.1 Problems with quantization
One obviously, and suspiciously, lacking element of the standard model is gravity. Up to
now, no fully consistent quantization of gravity has been obtained. Usually the problem is
that a canonical quantized theory of gravity is not perturbatively renormalizable. This is
visible when writing down the Einstein-Hilbert-Lagrangian of gravity, based on the metric
g , LEH ,

1 1
LEH = R + LM (7.1)
2

= g

R = +
1
= g ( g + g g )
2
From this, it can be inferred that the coupling constant involved, or equivalently New-
tons constant, is dimensionful, as is the cosmological constant . Any matter is described
by the matter Lagrangian LM . This also yields the Euler-Lagrange equations of gravity,
the Einstein equations
1
R g R + g = T (7.2)
2
LM
T = LM + 2 (g = ) (7.3)
g
Superficial (perturbative) calculations for (7.1) show that the theory is perturbatively non-
renormalizable. As a consequence, an infinite hierarchy of independent parameters, all to
be fixed by experiment, would be necessary to perform perturbative calculations, spoiling
any predictivity of the theory. In pure gravity, these problems occur at NNLO, for matter
coupled to gravity already at the leading order of radiative corrections. In particular, this
132 7.2. Gravity

implies that the theory is perturbatively not reliable beyond the scale . Though this
may be an artifact of perturbation theory, this has led to developments like supergravity
based on local supersymmetry or loop quantum gravity.
Irrespective of the details, the lack of gravity is an obvious flaw of the standard model.
Along with this lack comes also a riddle. The natural quantum scale of gravity is given
by the Planck scale
1
MP = 1.22 1019 GeV.
GN
This is 17 orders of magnitude larger than the natural scale of the electroweak interactions,
and 19 orders of magnitude larger than the one of QCD. The origin of this mismatch is
yet unsolved.
One of the more popular proposals how to explain this, discussed in section ??, is that
this is only an apparent mismatch: The scales of gravity and the standard model are the
same, but gravity is able to propagate also in additional dimensions not accessible by the
standard model. The mismatch comes from the ratio of the total volumes, the bulk, and
the apparent four dimensional volume, which is thus only a boundary, a so-called brane.
7.2.2 Asymptotic safety

Reiterating, the problem with the renormalizability of quantum gravity is a purely per-
turbative statement, since only perturbative arguments have been used to establish it.
Thus, the possibility remains that the theory is not having such a problem, it is said to
be asymptotically safe, and the problem is a mere artifact of perturbation theory. In this
case, when performing a proper, non-perturbative calculation, no such problems would be
involved. In fact, this includes the possibility that imposes just an intrinsic cutoff of
physics, and that this is simply the highest attainable energy, similarly as the speed of
light is the maximum velocity. As a consequence, the divergences encountered in particle
physics then only results from taking the improper limit .
This or a similar concept of asymptotic safety can be illustrated by the use of a running
coupling, this time the one of quantum gravity. The naive perturbative picture implies
that the running gravitational coupling increases without bounds if the energy is increased,
similarly to the case of QCD if the energy is decreased. Since the theory is non-linearly
coupled, an increasing coupling will back-couple to itself, and therefore may limit its own
growth, leading to a saturation at large energies, and thus becomes finite. This makes the
theory then perfectly stable and well-behaved. However, such a non-linear back-coupling
cannot be captured naturally by perturbation theory, which is a small-field expansion, and
thus linear in nature. It thus fails in the same way as it fails at small energies for QCD.
Non-perturbative methods, like renormalization-group methods or numerical simulations,

have provided indications that indeed such a thing may happen in quantum gravity, though
this requires confirmation.
As an aside, it has also been proposed that a similar solution may resolve both the
hierarchy problem and the triviality problem of the Higgs sector of the standard model,
when applied to the combination of Higgs self-coupling and the Yukawa couplings, and
possibly the gauge couplings.
7.3 Observations from particle physics experiments

There are two generic types of particle physics experiments to search for physics beyond
the standard model, both based on the direct interaction of elementary particles. One are
those at very high energies, where the sheer violence of the interactions are expected to
produce new particles, which can then be measured. The others are very precise low-energy
measurements, where very small deviations from the standard model are attempted to be
detected. Neither of these methods has provided so far any statistically and systematically
robust observation of a deviation from the standard model. Indeed, it happened quite often
that a promising effect vanishes when the statistical accuracy is increased. Also, it has
happened that certain effects have only been observed in some, but not all of conceptually
similar experiments. In these cases, it can again be a statistical effect, or there is always
the possibilities that some, at first glance, minor difference between the experiments can
fake such an effect at one experiment, or can shadow it at another. So far, the experience
was mostly that in such situation a signal was faked, but this then usually involves a
very tedious and long search for the cause. Also, the calculation of the theoretically
expected values is often involved, and it has happened that by improving the calculations,
a previously observed mismatch is actually just an underestimation of the theoretical error.
At the time of writing, while new results are coming in frequently, there are very few
remarkable results which should be mentioned, and which await further scrutiny. On
the other hand, several hundreds of measurements have not yet indicated any deviations
from known physics. One should then be especially wary: If enough measurements are
performed, there is always a statistical probability that some of them show deviations, the
so-called look-elsewhere effect, and a number of such deviations are expected.
One observation from precision experiments is that the magnetic moment of the muon
is not quite the expected one, but shows a very small deviation. However, the theoretical
computations are difficult, as virtual corrections from QCD enter, and it is therefore not
clear whether this problem is just a theoretical one. Nonetheless, new experiments will be
134 7.4. Astronomical observations
performed to at least obtain a better experimental value, while the theoretical calculations
are continously improved.
The next is that two experiments have in agreement reported results that cannot be
explained with just the three ordinary neutrinos. However, these results are rather indirect,
and at least another independent experiment will be required for confirmation.
All of these observations are currently investigated further. Given the amount of
statistics needed, it may take quite some time for a final answer about the reality of
any of these deviations.
7.4 Astronomical observations

During the recent decades a number of cosmological observations have been made, which
cannot be reconciled with the standard model. These will be discussed here.
7.4.1 Direct observations

Some come from space-borne experiments, like the FERMI-Lat satellite and the AMS-2
spectrometer aboard the ISS. Both show an excess of anti-positrons over the expected
yield in cosmic rays. If true, both would hint at an indirect detection of some new matter
particles, perhaps dark matter, discussed in the next section. However, it can not yet be
excluded that the rise in anti-positrons maybe due to not yet known or poorly understood
conventional astrophysical sources.
Also, there are some hints from white dwarf cooling rates which also may support
indirect effects from additional matter particles, as they cool too efficiently as possible with
just the available particles. But once again the effect may still be purely conventional.
7.4.2 Dark matter

One of the most striking observations is that the movement of galaxies, in particular how
matter rotates around the center of galaxies, cannot be described just by the luminous
matter seen in them and general relativity. That is actually a quite old problem, and
known since the early 1930s. Also gravitational lensing, the properties of hot intergalactic
clouds in galaxy clusters, the evolution of galaxy clusters and the properties of the large-
scale structures in the universe all support this finding. In fact, most of the mass must be
in the form of invisible dark matter. This matter is concentrated in the halo of galaxies,
as analyses of the rotation curves and colliding galaxies show. This matter cannot just be
due to non-self-luminous objects like planets, brown dwarfs, cold matter clouds, or black
holes, as the necessary density of such objects would turn up in a cloaking of extragalactic
light and of light from globular clusters. This matter is therefore not made out of any
conventional objects, in particular, it is non-baryonic. Furthermore, it is gravitational
but not electromagnetically active. It also shows different fluid dynamics (observed in the
case of colliding galaxies) as ordinary luminous matter. Also, the dark matter cannot be
strongly interacting, as it otherwise would turn up as bound in nuclei.
Thus this matter has to have particular properties. The only particle in the standard
model which could have provided it would have been a massive neutrino. However, though
the neutrinos do have mass, the upper limits on their mass is so low, and the flux of cosmic
neutrinos too small, to make up even a considerable fraction of the dark matter. This can
be seen by a simple estimate. If the neutrinos have mass and would fill the galaxy up to
the maximum possible by Fermi-statistics, their density would be
p3F
n = (7.4)
2
with the Fermi momentum pF in the non-relativistic case given by m v . Since neutrinos
have to be bound gravitationally to the galaxy, their speed is linked via the Virial theorem
to their potential energy
GN Mgalaxy
v2 = , (7.5)
R
with Newtons constant GN and R the radius of the galaxy. Putting in the known numbers,
and using furthermore that the observational results imply that n , the total number of
neutrinos approximated to be inside a sphere of the size of the galaxy, must give a total
mass larger than the one of the galaxy leads to the bound
14 12
0.001c 1 kpc
m > 100eV ,
3v R
yielding even for a neutrino at the speed of light a lower bound for the mass of about 3
eV, which is excluded by direct measurements in tritium decays.
Therefore, a different type of particles is necessary to fill this gap. In fact, many can-
didate theories for physics beyond the standard model offer candidates for such particles.
But none has been detected so far, despite several dedicated experimental searches for
dark matter. These experiments are enormously complicated by the problem of distin-
guishing the signal from background, in particular natural radioactivity and cosmic rays.
The source of this matter stays therefore mysterious.
But not only the existence of dark matter, also its properties are surprising. The
observations are best explained by dark matter which is in thermal equilibrium. But how
this should be achieved if it is really so weakly interacting is unclear.
On the other hand, the idea of gravitational bound dark matter is also problematic. In
particular, there is no reason why it should neither form celestial dark bodies, which should
be observable by passing in front of luminous matter by gravitational lensing, or why it
should not be bound partly in the planets of our solar system. Only if its temperature
is so high that binding is prohibited this would be in agreement, but then the question
remains why it is so hot, and what is the origin of the enormous amount of energy stored
in the dark matter.
It should be noted that there are also attempts to explain these observations by a de-
parture of gravity from its classical behavior also at long distances. Though parametriza-
tions exist of such a modification which are compatible with observational data, no clean
explanation or necessity for such a modification in classical general relativity has been
established. This proposal is also challenged by recent observations of colliding galaxies
which show that the center-of-mass of the total matter and the center of luminous matter
move differently, which precludes any simple modification of the laws of gravity, and is
much more in-line with the existence of dark matter.
Another alternative is that the problem of asymptotic infrared safeness of quantum
gravity may be related to the apparent existence of dark matter.
7.4.3 Inflation
A second problem is the apparent smoothness of the universe around us, while having
at the same time small highly non-smooth patches, like galaxies, clusters, super clusters,
walls and voids. In the standard model of cosmological evolution this can only be obtained
by a rapid phase of expansion (by a factor e60 ) of the early universe, at temperatures
much larger than the standard model scale, but much less than the gravity scale. None of
the standard model physics can explain this, nor act as an agitator for it. In particular, it
is also very complicated to find a model which at the same time explains the appearance
of inflation and also its end.
However, the predictions of inflation have been very well confirmed by the investigation
of the cosmic microwave background radiation, including non-trivial features and up to
rather high precision.
7.4.4 Curvature, cosmic expansion, and dark energy

Another problem is the apparent flatness of the universe. Over large scales, the angle
sum of a triangle is observed to be indeed . This is obtained from the cosmic microwave
background radiation, in particular the position of the quadrupole moment1 , but also that
the large-scale structure in the universe could not have been formed in the observed way
otherwise. For a universe, which is governed by Einsteins equation of general relativity,
this can only occur if there is a certain amount of energy inside it. Even including the
unknown dark matter, the amount of registered mass can provide at best about 30% of
the required amount to be in agreement with this observation. The other contribution,
amounting to about 70%, of what origin it may ever be, is called dark energy.
A second part of the puzzle is that the cosmic expansion is found to be accelerating.
This is found from distant supernova data, which are only consistent if the universe ex-
pands accelerated today. In particular, other explanations are very hard to reconcile with
the data, as it behaves non-monotonous with distance, in contrast to any kind of light-
screening from any known physical process. Furthermore, the large-scale structures of the
universe indicate this expansion, but also that the universe would be too young (about
10.000.000.000 years) for its oldest stars (about 12-13.000.000.000 years) if this would not
be the case. For such a flat universe such an acceleration within the framework of gen-
eral relativity requires a non-zero cosmological constant , which appears in the Einstein
equations (7.2). This constant could also provide the remaining 70% of the mass to close
the universe, and is in fact a vacuum energy. Such a constant is covariantly conserved,
since both T and the first two terms together are independently in general relativity,
and thus indeed constant. However, the known (quantum) effects contributing to such a
constant provide a much too large value for , about 1040 times too large. These include
quantities like the chiral and gluon condensates2 . These are of order GeV, and in addition
would have the wrong sign. What leads to the necessary enormous suppression is unclear.
7.4.5 Matter-antimatter asymmetry

In the standard model, matter and antimatter are not perfectly symmetric. Due to the CP
violations of the electroweak force, matter is preferred over antimatter, i. e., decays produce
more matter than antimatter, and also baryon and lepton number are not conserved
quantities. However, this process is dominantly non-perturbative. The most striking
evidence that this is a very weak effect is the half-life of the proton, which is larger than
1034 years. Indeed, only at very high-temperature can the effect become relevant.
1
The homogeneity of the universe leads to a vanishing of the monopole moment and the dipole moment
originates from the observers relative speed to the background.
2
There are some subtleties involved here with renormalization in a quantum theory, and the true source
of the problem is that the energy scale of gravity and the standard model are so vastly different. This is
ignored here for the sake of the argument.
After the big-bang, the produced very hot and dense matter was formed essentially from
a system of rapidly decaying and recombining particles. When the system cooled down, the
stable bound states remained in this process, leading first to stable nucleons and leptons in
the baryogenesis, and afterwards to stable nuclei and atoms in the nucleosynthesis. Only
over this time matter could have become dominant over antimatter, leading to the stable
universe observed today. But the electroweak effects would not have been strong enough
for the available time to produce the almost perfect asymmetry of matter vs. antimatter
observed today, by a factor of about 1019 . Thus, a further mechanism must exist which
provides matter dominance today.
There is a profound connection to inflation. It can be shown that inflation would
not have been efficient enough, if the number of baryons would have been conserved in
the process. In particular, the almost-baryon-number conserving electroweak interactions
would have permitted only an inflationary growth of e45 instead of e60 .
The possibility that this violation is sufficient to create pockets of matter at least as
large as our horizon, but not on larger scales has been tested, and found to yield only
pockets of matter much smaller than our horizon.
A further obstacle to a standard-model-conform breaking of matter-antimatter symme-
try is the necessity for a first order phase transition. This is required since in a equilibrium
(or almost equilibrium like at a higher-order transition), the equilibration of matter vs.
anti-matter counters the necessary breaking. However, with a Higgs of about 125 GeV
mass, this will not occur within the standard model.

Introduction To Particle Physics

Uploaded by

Introduction To Particle Physics

Uploaded by

Introduction to Particle Physics

Lecture in WS 2013/14 at the FSU Jena

3 A role model: Quantum electrodynamics 19

3.8 Fixing the inputs and running quantities . . . . . . . . . . . . . . . . . . . 27

4 Invariances and quantum numbers 31

5.16 Yang-Mills theory and QCD . . . . . . . . . . . . . . . . . . . . . . . . . . 74

7 Beyond the standard model 128

1.2 Natural units

1 eV = 1.60 1019 J (1.1)

and thus energies are then measured in inverse fm.

1.3 Elementary particles

1.4 Fermions and bosons

1.5 Particles and anti-particles

However, mass is not a conserved quantity. As a consequence of the famous Einstein

2.1.1 Elastic scattering

Important here is the absence of any potential energy. In scattering experiments, it is

2.1.2 Cross section

the other are differential cross sections

annihilation of particles, a conversion of energy to mass and back, is ubiquitous. Only

2.2.2 Inelastic scattering

This implies especially

the particle which has decayed.

2.4 Cross-sections and particles

2.5 Feynman diagrams

A role model: Quantum

3.1 Electrons and photons

Classically, this potential does not appear in the Maxwell equations

3.2 Formulation as a field theory

3.3 Indistinguishable particles

3.4 Gauge symmetry

3.5 Bound states and the positronium

3.6 Muons and Taus

3.7 Radiative corrections and renormalization

3.8 Fixing the inputs and running quantities

both an electric an a magnetic field. Quantum-mechanically, this is not so obvious. As

powerful tool when performing actual calculations in quantum field theory.

3.9 Landau poles

Invariances and quantum numbers

4.1 Symmetries, groups, and algebras

If it is not commutative, the symmetry group is called non-Abelian.

tion. Any unitary transformation can be written as1

4.2 Noethers theorem

where K is a total divergence, which is neither classically nor quantum-field theoreti-

The explicit form of the variation (4.1) is then

is conserved and thus time-independent,

4.3 Electric charge

4.4 Implications of space-time symmetries

4.6 Time reversal

4.7 Charge parity

4.9 Symmetry breaking

4.10 Fermion number

4.12 Chiral symmetry

5.1 Nuclei and the nuclear force

for the wave-function . Investigating a static situation yields the solution

5.3 Nucleons, isospin, and baryons

5.4 The quark model

for the 0 , i. e. the state of the 0 is

where is a mixing angle. This mixing angle can be experimentally or theoretically

5.5 Color and gluons

5.6 Chiral symmetry breaking

5.7 The Goldstone theorem

vacuum expectation value, i. e. replace

(x) v + (x) + i(x), (5.7)

v + (x) + i(x) ei (v + (x) + i(x)), (5.9)

5.8 Confinement and asymptotic freedom

5.9 Hadronic resonances

5.10 Glueballs, hybrids, tetraquarks, and pentaquarks

5.11 Flavor symmetry and strangeness

5.12 Charm, bottom and quarkonia

5.14 Partons and scattering experiments

5.14.1 Fragmentation functions and jets

5.14.2 Parton distribution functions

5.14.3 Monte-Carlo generators

As is visible, the experimental description of collisions is usually performed in a three step

5.15 Some algebra and group theory

algebra must therefore be N-dimensional.