Aspects of Neutrino Physics and CP Violation

Aspects of Neutrino Physics and CP Violation
Pedro Miguel Lopes Loreto dos Santos
Dissertação para a obtenção de Grau de Mestre em

Engenharia Fı́sica Tecnológica
Júri
Presidente: Prof. Maria Teresa Haderer de la Peña Stadler

Orientador: Dr. Maria Margarida Nesbitt Rebelo da Silva
Vogal: Dr. David Emmanuel-Costa
Vogal: Dr. Ivo de Medeiros Varzielas
Outubro 2011
Acknowledgements
I dedicate this thesis to my family.
To Neli Maria Pereira Lopes.
To João Alberto Loreto dos Santos.
To Avó Carlota.
To Avó Conceição.
To Avô Chico Lopes.
To Avô Joaquim.
i
Resumo
Este estudo discute as propriedades fı́sicas dos neutrinos e do sector leptónico. Nesse sentido, é analisada não
só a sua natureza e correspondente estrutura da matriz de mistura leptónica, como também são apresentados
diferentes tipos de mecanismos de seesaw associados a um termo de massa de Majorana. Fazemos uma
breve revisão da oscilação de neutrinos e do duplo decaimento beta sem neutrinos, apresentando o estado
observacional e experimental actual, assim como os últimos resultados relativos ao ângulo de mistura θ13
obtidos pelo T2K. São, de seguida, explicadas as condições necessárias e suficientes para a produção de uma
assimetria bariónica, ou mais especificamente as condições de Sakharov, e exploramos uma delas, a violação
CP, no sector leptónico. Finalmente focamos a nossa atenção na criação dessa assimetria pela leptogénese,
mostrando de uma forma qualitativa que esta contém todos os elementos necessários à sua geração.
Palavras-chave: fı́sica de neutrinos, partı́culas de Majorana, violação CP, leptogénese.
iii
Abstract
A concise analysis on the physical properties of neutrinos and the leptonic sector is presented. In this context,
we discuss its nature (Dirac vs. Majorana) and corresponding structure of the leptonic mixing matrix, along
with the different types of seesaw mechanism. We briefly review neutrino oscillation and double-β decay
scheme, presenting the current observational and experimental status, including the novel results from T2K
regarding θ13 . After explaining the theoretical conditions for producing a matter-antimatter asymmetry in
the Universe, or more specifically the Sakharov’s conditions, we explore one of them, CP violation, in the
leptonic sector. Finally, we focus our attention in the creation of that asymmetry through leptogenesis, and
we show that, in this scenario, all the qualitative ingredients are guaranteed once the seesaw mechanism is
assumed to be the source of neutrino masses.
Keywords: neutrino physics, Majorana particles, CP violation, leptogenesis.
v
Contents
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
Resumo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
1 Introduction 1
2 Theoretical background 3
2.1 Majorana Spinor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1.1 Dirac Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1.2 Charge conjugation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.3 Lorentz transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.4 Dirac and Majorana mass terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.5 A few words on Majorana mass term . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Discrete transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 Parity P . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.2 Charge Conjugation C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.3 CP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.4 Photon Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.5 Time Reversal T . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.6 CPT Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3 Experimental situation 15
3.1 Reconstructing mass matrices from experiments . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2 Neutrino oscillations experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4 Minimal extension of the Standard Model of particle physics 23

4.1 Extend Majorana masses to the SM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.2 Additional relevant CP transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.3 Seesaw mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.3.1 Type I seesaw mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.3.2 Type II seesaw mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.3.3 A final word on seesaw mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.4 Leptonic mixing matrix in low energy physics . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.4.1 Parametrisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
vii
4.4.2 Degenerate neutrino mass matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.4.3 Rephasing invariance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.4.4 Non-Unitarity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.5 Neutrino Oscillations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.6 Neutrinoless double beta-decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.7 Baryon Asymmetry of the Universe and Sakharov’s Conditions . . . . . . . . . . . . . . . . . 46
5 CP violation in the leptonic sector 49

5.1 General CP conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.2 CP violating phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.3 CP Parities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.4 CP violating phases in seesaw mechanism type I . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.5 Leptonic unitary triangles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.6 Low energy invariants in leptonic sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.7 Leptogenesis related invariants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6 Leptogenesis 65
6.1 CP violation () . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.1.1 A further look on CP violation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.2 Out-of-equilibrium dynamics (η) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
6.3 Lepton and B + L violation (Csphal ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.4 Baryon asymmetry from leptogenesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
7 Conclusions 77
A Some properties of the matrix C 79
B General mass term 81
C The third low energy invariant 83
D Calculating CP violation in Leptogenesis 85

D.1 The baryon asymmetry as an initial condition . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
D.2 Implication of unitarity and CPT in decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
D.3 Hubble expansion rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
D.4 Decay rate minimum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
D.5 Thermal equilibrium and the third Sakharov condition . . . . . . . . . . . . . . . . . . . . . . 88
viii
List of Tables
3.1 Neutrino parameters summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

2
3.2 sin 2θ13 experimental results summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3 Five main Long-baseline neutrino oscillation experiments. . . . . . . . . . . . . . . . . . . . . 20
5.1 Number of CP violating phases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
D.1 SU (2) × U (1) quantum numbers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
ix
List of Figures
2.1 Difference between massive Dirac and Majorana particles. . . . . . . . . . . . . . . . . . . . . 8
3.1 Neutrino mass patterns as indicated by the current data. . . . . . . . . . . . . . . . . . . . . 16
4.1 Neutrino flavour oscillation in vacuum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.2 Double-β transition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.3 0νββ decay process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.4 Schematic departure from equilibrium of the B violating process X ←→ Y + B. . . . . . . . . 48
6.1 The diagrams contributing to the CP asymmetry αα . . . . . . . . . . . . . . . . . . . . . . . 67

6.2 Sphaleron process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
xi
Chapter 1
Introduction
The historic discovery of neutrino oscillations marks a turning point in particle and nuclear physics and
implies that neutrinos have mass. This possibility was first suggested in late seventies, where theory tried to
explain the smallness of neutrino mass through seesaw mechanism [2–5].
Due to the successfulness of the Standard Model (SM) on explaining most of the high energy experi-
ments, we are induced to extend it in a minimal way so that we describe neutrino oscillation and explore
different possibilities. One of the most interesting relies on the nature of neutrinos - they are either Dirac
particles possessing distinctive antiparticles, or Majorana, neutral particles which are truly identical to their
antiparticles [6]. On the one hand, massive Dirac neutrinos are realised in gauge theories in which the lepton
charges, Le , Lµ and Lτ , are not conserved, but a specific combination of the latter, e.g. L = Le + Lµ + Lτ ,
is conserved; on the other hand, massive Majorana neutrinos arise if no lepton charge is conserved by the
electroweak interactions. Thus, the question of the nature of massive neutrinos is directly related to the
question of the fundamental symmetries of the elementary particle interactions.
It is important to keep in mind that different phenomena arise depending on the nature of neutrinos. When
we take neutrinos to be Dirac particles, oscillation has been our source of experimental results [7,8]. However,
processes such as double-β decay are viable in the case neutrinos are Majorana, and their observation can
not only indeed confirm it, but can also provide constraints on the general mass scale of neutrinos, to which
we only have an upper bound.
From the problem of the nature of massive neutrinos also arise different possibilities on the way CP is
violated. It is unambiguously established that in the quark sector there is one CP violating phase [9]. In
the context of Dirac neutrinos one would possibly have an analogous CP violating phase in the leptonic
sector. There is, however, the possibility that instead of one CP violating phase, we have three CP violating
phases [10] in the decoupling limit, in the case neutrinos are Majorana. This study addresses a rigorous
analysis to this subject.
In the context of CP violation there is the puzzle of the baryon asymmetry of the Universe (BAU). We
may wonder why such asymmetry is not an initial condition. Indeed, it would require not only a very precise
fine-tuning, but we would also have such asymmetry exponentially diluted away by inflation. The Standard
Model contains all the ingredients proposed by Sakharov [11] to dynamically generate the observed BAU, but,
yet, it fails to explain an asymmetry as large as the one observed (see, e.g. [12]), and for that reason additional
sources of CP violation are called for. This new physics must, first, distinguish matter from antimatter in
a more pronounced way than the weak interactions of SM do. Second, it should provide a departure from
thermal equilibrium during the history of the universe, or modify the electroweak phase transition. It that
1
context we present leptogenesis [13], which provides a very elegant explanation regarding this issue.
This study is organized as follows. In chapter 2, we present some theoretical formalism involving neutrinos,
especially regarding the Majorana spinor and the CP transformations we shall apply later on in this study.
In chapter 3 we list the current experimental bounds on the mixing angles and mass-squared differences,
and present a brief description about how different parameters can be (are) experimentally determined. In
chapter 4 we do not only present our framework, including our minimal extension to the Standard Model
and the different seesaw mechanisms, but we also briefly explore the two most relevant phenomena involving
neutrinos: neutrino oscillation and double-β decay. To finalise this general physics part of the study, we also
present the Sakharov’s conditions. In chapter 5 we proceed within our framework with a rigorous analysis
of CP violation in the leptonic sector. Chapter 6 provides us a model for baryogenesis through leptogenesis
which naturally fulfils all Sakharov conditions and also demonstrate its plausibility. Chapter 7 concludes.
2
Chapter 2
Theoretical background
We start our study with a brief description of some theoretical aspects, even before presenting our physics
framework. For this purpose, we shall properly define the Majorana spinor and its properties, so that we can
present a consistent theory in chapter 4. It is important to point out that the nature of neutrinos (‘Majorana’
vs. ‘Dirac’) is not known; however, the majority of neutrino mass models predict the ‘Majorana’ type so it
appears to be more natural this to take place. Another subject we shall study are P, C, T and combinations
of these transformations, which play a crucial role not only in this study, but in general in particle physics.
2.1 Majorana Spinor

Several physics properties of neutrinos may arise when we consider neutrinos to be described by a Majorana
spinor, instead of the typical Dirac spinor. In the former case, we see that we only need one Weyl spinor to
build a mass term, but we also see that no charges can be conserved. To have a proper description of this
spinor, we should firstly introduce the well-known Dirac equation, as we do in the sequel.
2.1.1 Dirac Equation

The Dirac equation has been very successfully used for the description of elementary spin 1/2 particles, such
as electrons, and it is consistent with both the principles of quantum mechanics and the theory of special
relativity. The equation for the 4-component spinor describing interaction-free Dirac particles reads
(i~γ µ ∂µ − mc)ψ(x, t) = 0. (2.1)
In natural units (with ~ = 1, c = 1) this can be seen as the Euler-Lagrange equation arising from the
Lagrangian
i
L= / − ψ ∂ψ)
(ψ ∂ψ / − mψψ, (2.2)
2
where the Feynman slash notation and the Dirac adjoint denote, respectively, ∂/ ≡ γ µ ∂µ and ψ ≡ ψ † γ 0 . In
these equations γ µ are the well-known Dirac matrices which satisfy the following conditions
{γ µ , γ ν } = 2g µν , (2.3)
γ0 γµ γ0 = γµ† , (2.4)
3
where one can use gµν = diag(1, −1, −1, −1). The first condition is necessary so that the Dirac equation
complies with Klein-Gordon equation, and inherently with the energy momentum relation E 2 = p2 + m2 .
The second relation is needed in order to have an Hermitian Lagrangian (or the observable Hamiltonian).1
It can more explicitly be written as follows
γ0† = γ0 , γi† = −γi , (2.5)
where i = 1, 2, 3. It is also worth mentioning a Pauli theorem which enables changing from one to another
irreducible representations of the Dirac algebra. It states that if γµ and γµ0 are two irreducible representations
of the Dirac algebra, then there is a non-singular matrix S such that
γµ0 = Sγµ S −1 (2.6)
and this matrix is unique, apart for an arbitrary multiplicative factor. Moreover, the spinor transforms as
follows
ψ 0 = Sψ. (2.7)
2.1.2 Charge conjugation

To proceed with the definition of the Majorana spinor, one should firstly introduce the charge conjugation
of a Dirac spinor, ψ c . For this purpose, one considers the Dirac equation for a charged particle, e.g. the
electron, which is defined by
(i∂/ − eA
/ − m)ψ = 0. (2.8)
and, additionally, one also considers the Dirac equation applied to its antiparticle, which is
/ − m)ψ c = 0,
(i∂/ + eA (2.9)
where ψ c is the wave function of the antiparticle. In order to relate both, one should notice that the relative
sign of i∂/ and eA
/ is the opposite. For this reason, we consider the complex conjugate of (2.8)
∗ ∗
(−iγ µ ∂µ − eγ µ Aµ − m)ψ ∗ = 0, (2.10)
∗
and compare both equations. When we introduce the Dirac adjoint definition γ 0T ψ ∗ = ψ̄ T and γ 0T γ µ γ 0T =
γ µT from (2.4), the above equation yields
µT T
−γ (i∂µ + eAµ ) − m ψ = 0, (2.11)
and then it is straightforward to observe that one can define the invertible matrix C such that the following
condition holds
Cγ µT C −1 = −γ µ , (2.12)
−
→
1 It / = (iψ † γ 0 γ µ ∂ µ ψ)† =
is straightforward to check Hermiticity requires this condition. For instance, in the term iψ ∂ψ
←− † −
→ µ† 0 † †
−iψ † ∂ µ γ µ† γ 0 ψ † 0 µ
= iψ ∂ µ γ γ ψ, we must consider γ γ = γ γ . µ† 0
4
in order to identify (2.9) with (2.11). As a result, we also define the charge conjugate of a spinor ψ to be
ψ c ≡ C ψ̄ T . (2.13)
This is equivalent to ψ c ≡ Cγ0T ψ ∗ . Having in mind the properties of the matrix C, defined in the appendix
A, one can derive ψ c = (ψ c )† γ 0 = ψ T γ 0∗ C † γ 0 = ψ T (C −1 )T = −ψ T C −1 , and use in the course of the work
this identity
ψ c ≡ −ψ T C −1 . (2.14)
As we will show in the next section, ψ c is a spinor since it Lorentz-transforms as a Dirac spinor.
2.1.3 Lorentz transformation

As mentioned before, ψ c must Lorentz-transform appropriately in order to be considered a spinor. One shall
consider infinitesimal Lorentz transformations (rotations and boost) where one takes the coordinate of a
spacetime point from
xµ → x0µ = xµ + ω µν xν , (2.15)
which is equivalent to
xµ → x0µ = Λµν xν , (2.16)
with Λµν = δ µν + ω µν . A fermion field transforms as follows

0 0 i µν
ψ (x ) = exp − ω σµν ψ(x), (2.17)
4
where
i
σµν = [γµ , γν ] , (2.18)
2

and the Lorentz transformation is defined by S(Λ) = exp − 4i ω µν σµν , resulting from the following condition
obtained considering the Lorentz covariance of the Dirac equation (2.1)
S(Λ)−1 γ µ S(Λ) = Λµν γ ν . (2.19)
By taking the charge conjugate of (2.17), the following result can be written
c ∗
[ψ 0 (x0 )] = Cγ0T [ψ 0 (x0 )]
= Cγ0T S ∗ ψ ∗ (x)

i µν ∗
= Cγ0T exp ω σµν (Cγ0T )−1 [ψ(x)]c . (2.20)
4
Using, successively (2.4) to take γ 0T γ µ∗ γ 0T = γ µT , and (2.12), then the following identity holds
Cγ0T γµ∗ (Cγ0T )−1 = −γµ . (2.21)
With above equality one obtains
∗ i
Cγ0T σµν (Cγ0T )−1 = − [γµ , γν ] = −σµν . (2.22)
2
5
Accordingly, one can simplify (2.20) and finally write

0 0 c i µν ∗
[ψ (x )] = exp − ω σµν [ψ(x)]c . (2.23)
4
It is clear, apart from the spinor charge conjugation that this equation is the same as (2.17). In other words,
this equation tells us that [ψ(x)]c transforms exactly the same way that the fermion field does under proper
Lorentz transformations, therefore it is Lorentz covariant.
2.1.4 Dirac and Majorana mass terms

Consider the case of free fields without interactions, eq. (2.2). The mass term is given in terms of its Weyl
spinors as
Lm = mψψ = m(ψL + ψR )(ψL + ψR ) (2.24)

= m(ψL ψR + ψR ψL ), (2.25)
where the Dirac adjoint ψ = ψ † γ 0 already introduced guarantees the Lorentz-invariance of the term. We
considered a real m for simplicity. Note that ψR/L ≡ 21 (1 ± γ5 )ψ.
From the above we see that a mass term connects two fields with different chiralities, since terms like
ψL φL = ψR φR = 0. In the context of a Dirac mass term, we consider two independent L and R spinors, and,
consequently, the mass eigenstate is
ψ = ψL + ψR . (2.26)
Regarding physics in Standard Model, since we only have left-handed neutrinos, neutrinos remain massless.
One should now derive the Majorana mass term. For that purpose, it is important to establish notation
and notice that the charge conjugate spinor ψ c has the opposite chirality of ψ, i.e.
c 1
ψL ≡ (ψL )c = (1 + γ5 )ψ c = (ψ c )R (2.27)
2
With this spinor, we can build the Majorana mass term, firstly introduced in 1937 by Ettore Majorana [6],
c
by considering that the right-handed field in (2.24) is, in fact, ψL . This leads to the following mass eigenstate
c
χ = ψL + ψL . (2.28)
And to the usual definition of a Majorana spinor2
χc = χ (2.30)
2 We will keep this definition for notation simplicity. In fact, as one shall explore in section 5.3, in the most general definition
of a Majorana spinor one can have a relative phase in the components of φ1 and φ2 which manifest itself in the self conjugate
condition. For instance, χ is the following
c
χ = ψL + exp(iξ)ψL , (2.29)
which leads to a mass term in the Lagrangian that reads
c )(ψ + exp(iξ)ψ c )
mL χ1 χ1 = mL (ψL + exp(−iξ)ψL L L
c cψ
= mL exp(iξ)ψL ψL + mL exp(−iξ)ψL L
0c ψ 0 + ψ 0 ψ 0c )
= mL (ψL L L L
0 = exp(iθ/2)ψ .
where the last equation holds just by doing a simple phase redefinition ψL → ψL L
6
It is now clear that with only one independent Weyl spinor we can obtain a mass term, while in the usual
Dirac framework we always need two independent Weyl spinors.
Considering now the second independent spinor, ψR , we can obtain the second mass eigenstate
c
ω = ψR + ψR , ω c = ω, (2.31)
and we now redefine eq. (2.24) as Majorana-type mass terms that connect the L and R components of
conjugate fields, as follows
c ψ + ψ ψ c ) = m χχ,
LL = mL (ψL (2.32)
L L L L
cψ + c
LR = mR (ψR R ψR ψR ) = mR ωω, (2.33)
where both χ and ω are Majorana spinors.

It is important to clarify that sometimes when one refers, for instance, to left-handed Majorana spinor
(e.g. left-handed light neutrinos), this is an abuse of language. Indeed, such spinor (2.28) is not left-handed
- the charge conjugate of a left- (right-) handed spinor is a right- (left-) handed one - but has, instead, both
chiralities, although the denomination may be, in fact, misleading.
The combination of both Dirac and Majorana mass terms can lead us to the most general mass term (the
Dirac-Majorana mass term)3
c ψ c ) + m∗ ψ ψ c + M ψ c ψ + h.c.
2LDM = mD (ψL ψR + ψR L L L L R R R
! !
∗
mL mD c
c
ψL
= ψL ψR + h.c. (2.34)
mD MR ψR
= ΨL MΨR + ΨR M∗ ΨL ,
c c
where one can see, from (2.27), that ψL is a right-handed field, and the opposite stands for ψR . In the last
c c T T
step were also introduced the following definitions ΨL ≡ (ψL , ψR ) , ΨR ≡ (ψL , ψR ) and
!
m∗L mD
M≡ . (2.35)
mD MR
2.1.5 A few words on Majorana mass term

Analyzing the above definitions, it is clear that:
i) Majorana fermions have no conserved charge. For instance, considering a U (1) transformation, ψ →
eiα ψ, it is clear that the Dirac mass is invariant under this transformation. This symmetry has an
associated conserved quantum number, which can be, for example, electric charge or lepton number.
However, the Majorana mass term is not invariant under this U (1) transformation. For this reason,
Majorana mass term can only be considered in electrically neutral particles, since there is no evidence
of electrical charge violation. Conversely, if neutrinos are Majorana particles, the total lepton charge
is not conserved. This is a very practical prediction, and if neutrinos are Dirac particles, we shall find
that processes like neutrinoless double-β decay are forbidden;
ii) Majorana fermions are more ‘economic’, since we only need one independent Weyl spinor to build a
3 h.c. through this work signifies Hermitian conjugate.
7
mass term. This is due to the self-conjugacy of these fermions, that allows to build a mass term with
only one chiral spinor.4 When one considers a Dirac mass term, one needs two independent Weyl
spinors, corresponding to left- and right handedness;
iii) As it is shown in appendix B, the most general case with two independent Weyl spinors describes two
Majorana particles with different masses. However, when one sets mL = MR = 0, one obtains the
usual Dirac fermion (describing particle and antiparticle);
iv) In the sequence of the above, we see that in the case of Dirac fermions, a Lorentz boost can change
particle’s handedness, whereas CPT can transform a particle into an antiparticle, or vice-versa (see
fig. 2.1 - for instance, a Lorentz boost transforms νL in νR and CPT transforms a left-handed νL in
a right-handed antiparticle ν̄R ). This gives a total of four possible neutrino states. However, since a
Majorana particle is its own antiparticle, i.e. ν̄R is identical to νR , and ν̄L is identical to νL , we see
that CPT is equivalent to a Lorentz boost - a Majorana particle has only two distinct states;
Lorentz boost
Dirac
νL νR νL νR
CPT CPT
Lorentz boost
Majorana
νL νR
CPT
Figure 2.1: Schematic drawing of the difference between massive Dirac and Majorana particles.
v) Majorana mass terms are n symmetric by definition.

o In fact, having in mind the anticommutative prop-
† 3
erties of fermionic fields, ψα (~x, t), ψβ (~y , t) = δ (~x − ~y )δαβ and the antissymmetry of C (shown in
T T
(A.6)), it follows that ψi Cψj = ψj Cψi . In other words, the Majorana mass term is only meaningful
in anticommuting fields.
2.2 Discrete transformations

Since it is a crucial part of this work, in this section one should define how quantum fields behave under the
discrete space-time transformations of P and T, as well as their behavior under charge conjugation C which
physically corresponds to reversing the sign of all charges. In this context, the combined transformation of
CP plays an important role in physics as we shall see in chapter 4, and is subject to a rigorous treatment in
the literature (see [1]).
4 For instance, if one considers a left-handed spinor ψ , then ψ c is right-handed, therefore it is possible to build a mass term,
L L
as we previously checked.
8
It is extremely important to study these transformations since they can hide symmetries of Nature. For
this reason one should be more specific about what means to have a certain law of physics which is symmetric
under certain transformations. Consider, for instance, a motion that satisfies the laws of classical mechanics.
Then, reflect the motion into a mirror and check if that motion, or more precisely, all the motions that
satisfy the laws of classical mechanics also satisfy them after being reflected in the mirror. If they do, then
classical mechanics is said to be symmetric under mirror inversion. In general, suppose one applies a certain
transformation to a motion that follows a certain law of physics. If the resulting motion satisfies the same
laws, and if such is the case for all motion that satisfies the law, then the law of physics is said to be symmetric
under the given transformation.
In the context of quantum mechanics, the above criterion for the symmetry of physical law can formally
be stated as follows: for the state vectors |ii and |f i,respectively representing certain initial and final states,
there exist |i0 i = U|ii and |j 0 i = U|ji which represent, in the case of mirror inversion, the corresponding
states reflected into a mirror. This transformation is induced by a Unitary operator in the Hilbert space U.
Furthermore, if the laws of physics are symmetric under this transformation, the transition probability is the
same before and after the transformation, i.e.
|hf 0 |S|i0 i|2 = |hf |S|ii|2 . (2.36)
where the same operator S is used for the transformed states.

Since it is now clear how a transformation and its respective symmetry can be defined, one shall begin
with parity.
2.2.1 Parity P
Dirac Field
The parity transformation P is the operation to change the signs of the three space coordinates. Therefore,
this is equivalent to a finding of a Lorentz transformation defined by Λµν = diag(1, −1, −1, −1). Formally,
this transformation is induced by a Unitary operator P . The action of this operator in the spinor ψ(x, t)
shall take it to ψ(−x, t).
In order to find this transformation it is clear, comparing with (2.17), that one should find some unitary
matrix P subject to
Pψ(x, t)P † = P ψ(x, t). (2.37)
From (2.4) one can take
γ0 γi γ0 = −γi . (2.38)
When one compares this equation with (2.19), it is clear that one should apply a S = γ0 to the Dirac equation
in with the purpose of flipping the sign of the spacial coordinates. In fact, since we are considering a unitary
operator, one takes S(Λ) = exp(iβp )γ 0 ,5 i.e.
Pψ(t, ~x)P † = exp(iβp )γ 0 ψ(t, −~x). (2.39)

5 Please note that, although S(Λ) is unitary in a parity transformation, that is not, in general, the case.
9
Note that we also have
Pψ(t, ~x)P † = Pψ † (t, ~x)P † γ 0

= exp(−iβp )ψ(t, −~x)γ 0 . (2.40)
and
P 2 ψ(t, ~x)P †2 = exp(2iβp )ψ(t, ~x). (2.41)
The P is an operator in the Hilbert space and is shown to be unitary
P † P = 1. (2.42)
Photon Field
The photon is a vector field. This means that when one parity transforms it one should flip the sign of its
spacial dependencies, as we usually do with a vector. Therefore, one should have
PAµ (t, ~x)P † = gµν Aν (t, −~x) (2.43)
Considering the electromagnetic current LA µ

int = eψγ Aµ ψ introduced before, and since we know that deriva-
tives also flip the sign if its spacial coordinates, ∂ µ → g µν ∂ν , then it is clear that the the electromagnetic
interaction is invariant under a parity transformation.
2.2.2 Charge Conjugation C

Dirac Field
As introduced in section 2.1.2, we define the charge conjugation operation C to be the one that exchanges
particle and antiparticle without changing its momentum and spin. Taking in account that we have already
derived the charge conjugation spinor in eq. (2.13), then the action for the charge conjugation operator is
the following
Cψ(t, ~x)C † = exp(iβc )ψ c (t, ~x). (2.44)
With the above result, one can also derive
Cψ(t, ~x)C † = exp(−iβc )ψ c (t, ~x). (2.45)
Considering the C transformations above and the definitions for the charge conjugate spinor and its Dirac
adjoint, given respectively in (2.13) and (2.14), one finds, as expected, that
C 2 ψC †2 = 1. (2.46)
Finally, as in the case of the P operator, the C is also unitary
C † C = 1. (2.47)
One should note that in the case of charge conjugation, all the space-time variables are the same on the
10
both sides of the equality.
Photon Field
To define the charge conjugation in the photon field, one must consider that the electromagnetic interaction
is invariant under a C transformation. Moreover, we know that the charged current eψγ µ ψ ≡ j µ = (ρ, ~j)
must change sign under C, since it is charge e dependent, representing, respectively, the charge density and
density of current charge. Hence, since one has Cj µ C † = −j µ , then in order to have the mentioned invariance
of electromagnetic interaction one must take
CAµ (t, ~x)C † = −Aµ (t, ~x). (2.48)
2.2.3 CP
Fermion Field
It is a key goal in this work to explore CP transformations in the leptonic sector. Hence it is crucial to define
such transformations. Considering the weak interaction, we see that this interaction creates left-handed
fermions or right-handed antifermions. Under particle-antiparticle exchange, the statement becomes ‘the
weak interaction creates left-handed antifermions and right-handed fermions’ which contradicts the original
statement. However, if one further transforms the above statement by parity, it becomes ‘the weak interaction
creates right-handed antifermions and left-handed fermions’ which is the same as the first statement. Thus,
one would expect that this interaction is invariant under P followed by C, i.e. CP. It should not matter
which is applied first.
In fact this is not a valid assumption, since the quark-W coupling in the Standard Model is in general not
invariant under CP. Therefore, since CP is not a symmetry of the Lagrangian, we have no reason whatsoever
to suppose it is conserved in the leptonic sector. Furthermore, there is the evidence that the known CP
violation is insufficient to explain the observed Baryon Asymmetry of the Universe (BAU) in the context of
electroweak baryogenesis, which means new physics is called for.
Let us proceed by defining CP. Considering both (2.39) and (2.44), one can define
(CP)ψ(t, ~x)(CP)† = exp(iβψ )γ 0 ψ c (t, −~x) (2.49)

0 T
= exp(iβψ )γ Cψ (t, −~x) (2.50)
with βψ ≡ βp + βc . Then
(CP)ψ(t, ~x)(CP)† = exp(−iβψ )ψ c (t, −~x)γ 0 (2.51)

= − exp(−iβψ )ψ T (t, −~x)C −1 γ 0 , (2.52)
where, in the last identity, we considered a few properties of the matrix C, defined in appendix A, to take
ψ c = (ψ c )† γ 0 = ψ T γ 0∗ C † γ 0 = ψ T (C −1 )T = −ψ T C −1 .
By using the above identities, one can also define the following very useful CP transformations, regarding
the charge conjugate spinors
(CP)ψ c (t, ~x)(CP)† = − exp(−iβψ )γ 0 ψ (2.53)

(CP)ψ c (t, ~x)(CP)† = − exp(iβψ )ψγ . 0
(2.54)
11
2.2.4 Photon Field
By both (2.43) and (2.48), one can define
(CP)Aµ (t, ~x)(CP)† = −gµν Aν (t, −~x). (2.55)
Then it comes, as a consequence of the invariance under P and C, that electromagnetic interaction is invariant
under a CP transformation.
Other relevant CP transformations we will use in the course of this work are presented in section 4.2.
2.2.5 Time Reversal T

Dirac Field
In general, a law of physics is said to be invariant under time reversal if one reverses the time of the
phenomenon that satisfies a certain law of physics, and the resulting phenomenon still satisfies the same law
of physics. This is the case of classical mechanics or electromagnetism.
In quantum mechanics, the time-reversal invariance is stated as the invariance of transition probability
when everything is time-reversed; namely, the initial and final states are interchanged where all momenta
and spins are flipped while keeping particle types and energies the same
|hi0 |S|f 0 i|2 = |hf |S|ii|2 , (2.56)
where |i0 i and |f 0 i represent the states |ii and |f i under time reversal, namely with momenta and spins
flipped.
We will consider now a very brief discussion on time reversal. If one considers a T transformation that
interchanges the initial and final states, then one gets hT i|T f i = hj|ii = hi|ji∗ . It is therefore clear that time
reversal transformation can be achieved by an operator
T = T0 K0 (2.57)
where K0 complex conjugates and T0 is to be determined. When one considers the time reversal, the sign of
the spacial components of the potential vector A and the momentum p are flipped. Therefore, considering
Dirac equation (2.8), it is clear that the operator T must obey to
T γi T −1 = −γi , T γ0 T −1 = γ0 . (2.58)
Taking into account the effect of K0 in (2.57), then the above relation stands as
T0 γi∗ T0 −1 = −γi , T0 γ0∗ T0 −1 = γ0 . (2.59)
Considering the Dirac representation, where γ0 , γ1 and γ3 are real, and γ2 imaginary, it is valid to take
T = i exp(iβt )γ1 γ3 K0 6 . Thus, one can now introduce the action under a Time reversal operator, which is
T
T ψ(t, ~x)T −1 = i exp(iβt )γ1 γ3 ψ ∗ (−t, ~x) = exp(iβt )γ 2 γ5 ψ (−t, ~x) (2.60)
−1 T 0 T 2
T ψ(t, ~x)T = i exp(−iβt )ψ (−t, ~x)γ γ3 γ1 = exp(−iβt )ψ (−t, ~x)γ5 γ (2.61)
6 The factor i is arbitrary
12
Photon field
Having in mind the definition of j µ = (ρ, ~j), by physical intuition one knows that the charge density should
hold, whereas the density of current should flip the sign. Hence, in order to keep electromagnetism invariant
under T , one must have
T Aµ (t, ~x)T † = gµν Aν (−t, ~x). (2.62)
2.2.6 CPT Theorem

CPT is a fundamental theorem which states that any local quantum field theory with an Hermitian Lorentz
invariant Lagrangian and satisfying the spin-statistics theorem - namely commutators for bosons and anti-
commutators for fermions - is invariant under the compound operation CPT , where the operators can be
placed in any order.
One consequence of this theorem, proved by G.Lüders, W.Pauli and J.Bell is related to the fact that if
one of the discrete symmetries does not hold, then another one must be violated. This is important to fulfill
the Sakharov conditions, which we shall analyze in section 4.7. A proof of this theorem is beyond the scope
of this study and can be found, for instance, in [14].
13
Chapter 3
Experimental situation
When the Standard Model was conceived, neutrinos were thought to be massless and different lepton numbers
were believed to be conserved (which would not be the case if right-handed neutrinos were present). This
was a reason for not introducing right-handed neutrinos decades ago. However, the phenomenon of neutrino
oscillations is established today, and is in fact our main source of experimental results, demonstrating that
neutrino flavour states (νe , νµ , ντ ) are indeed quantum superpositions of states (ν1 , ν2 , ν3 ) with definite
masses mi . These oscillations manifest themselves in experiments with solar, atmospheric, accelerator and
reactor neutrinos.
As one shall study in part 4, the simplest form of the lepton mixing matrix, or alternatively called
Pontecorvo-Maki-Nakagawa-Sakata (PMNS) matrix, which relates flavour and mass neutrino eigenstates, is
given in terms of a unitary matrix with three mixing angles and three CP violating phases. In standard
parametrisation [15], which is now presented, these parameters are, respectively θ12 , θ13 , θ23 and δ, α, β, so
that
  
c12 c13 s12 c13 s13 e−iδ 1 0 0
  
UP M N S = −s12 c23 − c12 s23 s13 eiδ c12 c23 − s12 s23 s13 eiδ s23 c13  0 eiα 0 (3.1)
iδ iδ
s12 s23 − c12 c23 s13 e −c12 s23 − s12 c23 s13 e c23 c13 0 0 eiβ
with cij ≡ cos θij and sij ≡ sin θij . The angles θ12 ≡ θ and θ23 ≡ θA are very well determined by oscillation
data. Additionally, the results obtained for the mass splitting parameters ∆m221 ≡ ∆m2 ≡ m22 − m21 and
|∆m231 | ≡ |∆m2A | ≡ |m23 − m21 | are also rather accurate. This contrasts with θ13 , where accounting with 2σ
only upper bounds can be placed, and with the phases, which as we will analyse in section 3.1, are rather
hard to retrieve from experiments.
Considering the data regarding the mass splitting, based on the experimental results summarized in table
3.1, one can arrange the different mass levels as sketched in fig. 3.1
p
i) Normal hierarchy, i.e. m1 m2 < m3 . In this case we can deduce the value of m3 ≈ ∆m232 ∼
0.04 eV, where ∆m232 ≡ m23 − m22 > 0. The mass of the lightest neutrino is not know, but if we take
p
m1 m2 , then we get the value of m2 ≈ ∆m221 ∼ 0.009 eV;
p
ii) Inverted hierarchy, i.e. m1 ' m2 m3 . In this case m2,3 ≈ ∆m232 ∼ 0.04 eV, where ∆m232 ≡
m23 − m22 < 0. We have no information about m3 except that its value is much less than the other two
masses;
iii) (Almost) Degenerate neutrinos, i.e. m1 ' m2 ' m3 . This is the case when all mi & 0.1 eV. In
15
such case the mass difference is negligible when comparing to the absolute mass of neutrinos, and thus
one can treat them as almost degenerate.
m2 m2
νe
νµ
ντ
m32 m22
solar~7×10−5eV2
m12
atmospheric
~2×10−3eV2
atmospheric
m2 2 ~2×10−3eV2
solar~7×10−5eV2
m12 m32
? ?
0 0
Figure 3.1: Neutrino

FIG. mass patterns masses
3: Neutrino as indicated by the current
and mixings data (figure
as indicated extracted
by the currentfrom
data.[16]). The left
hand diagram shows normal spectrum and the right hand side, an inverse spectrum.
2 2 2
We stillhave
do not 23 ≡ which
∆mknow m3 − of
m2these
< 0.patterns
We have no information
is correct. about
Indeed, based onm3 except that
observations, we its
havevalue
only
placed upper boundsless
is much on than
neutrino
themasses,
other meaning that we still do not know its overall mass scale. It this
two masses.
context, we should note that the methods presented in this chapter are not very valuable for the determi-
nation of lightest
(iii) neutrino
Degenerate mass in the
neutrinos, i.e.cases
m1 of
≃ amnormal or inverted mass-hierarchy (its mass would be too
2 ≃ m3 .
small), but can be useful for determining the absolute masses in the case of degenerate neutrinos, i.e. when
Oscillation
all mi & 0.1 eV. experiments do not tell us about the overall scale of masses. It is therefore
important to explore to what extent the absolute values of the masses can be determined.
Let us present the current status of neutrino mixing parameters considered in the standard parametri-
While
sation, eq. discussing
(3.1). The the question
analysis of absolute
assumed masses,
that i) there it isthree
are only good to keep
mixed, in neutrinos;
active mind thatii)none
CPTof
is
conserved, so thatdiscussed
the methods masses and mixing
below canangles in the
provide anyneutrino and antineutrino
information about the channel
lightestcoincide;
neutrinoand iii)
mass
neutrino masses and mixings have pure ‘vacuum origin’, or equivalently, that these parameters are due to
the in the caseswith
interaction of aHiggs
normal or that
field(s) inverted mass-hierarchy.
develop a VEV at a scaleThey
much are most
larger thanuseful formass.
neutrino determining
absolute masses
In table 3.1 in one
we omit the of
case
theof degenerate
mixing angles, neutrinos, i.e., when
θ13 . This mixing angle all mi ≥ 0.1toeV.
is supposed be small, and
experiments could only place reliable upper bounds to it. However, in June 2011, T2K (Tokai to Kamioka)
One can directly search for the kinematical effect of nonzero neutrino masses in beta-
experiment released results [8] which show an evidence that θ13 although very small, is not zero. It was
decayinbythis
detected, looking for structure
experiment, near
six electron the end
neutrinos point
created outofofthe electron
muon energy
neutrinos, whenspectrum. This
in a three flavour
oscillation
search scenario with ∆m
is sensitive
2
to 32
neutrino 10−3 eV2 ,regardless
= 2.4 × masses sin2 θ23 = 0.5 and sin2 θ13
of whether = neutrinos
the 0, the expected
are number of
Dirac or
pP
such events is 1.5 ± 0.3(syst.). Under this hypothesis, they claim that the probability to observe six or more
2 2
Majorana
candidate eventsparticles.
is 7 × 10−3One is corresponds
, which sensitive toto the quantity
a 2.5σ β ≡ the θ i |U
deviationmfrom ei | mi . The Troitsk
13 null-hypothesis.
and Mainz
If such experiments
evidence place the present
becomes measurement, it couldupper limit
lead to evenon ≤ 2.2 eV.measurements,
mβ interesting
more The proposednamely
KA-
related to CP violating, considering that the possibility of CP violation dependent on a Dirac phase δ (which
TRIN experiment is projected to be sensitive to mβ > 0.2 eV, which will have important
means, for instance, different conversion probabilities for neutrinos and antineutrinos, i.e. P (νi → νj ) 6=
P (νimplications
i → νj )) relies for thenon-vanishing
on the theory of neutrino
of all themasses. For instance,
mixing angles (θ12 , θ13 , θif the
23 ). result one
However, is positive, it
should note
thatwill
since eiδ always
imply appears with
a degenerate a factor on
spectrum; sin θthe
13 , ‘Dirac’ CP violation
other hand effectsresult
a negative must will
be small.
be a very useful
constraint.
16
If neutrinos are Majorana particles, the rate for ββ0ν decay Majorana mass for the
Parameter best fit ±1σ 2σ 3σ
∆m221 [10−5 eV 2 ] 7.59+0.20

−0.18 7.24 − 7.99 7.09 − 8.19
2.45 ± 0.09 2.28 − 2.64 2.18 − 2.73

∆m231 [10−3 eV 2 ]
−(2.34+0.10
−0.09 ) −(2.17 − 2.54) −(2.08 − 2.64)
sin2 θ12 0.312+0.017

−0.015 0.28 − 0.35 0.27 − 0.36
0.51 ± 0.06 0.41 − 0.61

sin2 θ23 0.39 − 0.64
0.52 ± 0.06 0.42 − 0.61
Table 3.1: Neutrino parameters summary. For ∆m232 , sin2 θ23 and sin2 θ13 , the upper (lower) row corresponds
to normal (inverted) neutrino mass hierarchy. Dirac phase δ = 0. These results are presented in [7].
It is clear that θ13 6= 0 is not an highly accurate measurement so far, but it may get better in the future.
It is important to stress that these results have been viewed as the first reasonably robust experimental
evidence that this angle is nonzero. One should also mention that data are consistent with lower bounds
of 0.03 and 0.04 for sin2 2θ13 , considering respectively a normal and inverted hierarchy. All the results are
summarized in table 3.2.
sin2 2θ 13 experimental results

Prior to T2K [17]
(it accounts with solar and atmospheric neutrinos + KamLAND + Chooz + Palo Verde +
+ short-baseline experiments (Bugey4, ROVNO, Bugey3, Krasnoyarsk, ILL Gösgen)
0.040+0.035
−0.024 ≤ 0.105 ≤ 0.135
(best fit ±1σ) (2σ) (3σ)
0.051+0.035
−0.027 ≤ 0.120 ≤ 0.150
T2K [8] (see also [18] for a global neutrino analisys regarding θ13 )
0.03 < sin2 2θ13 < 0.28

at 90% Confidence level (C.L.)
0.04 < sin2 2θ13 < 0.34
expected/observed number of νµ → νe events - 1.5 ± 0.3 / 6 (2.5σ significance)
Table 3.2: sin2 2θ13 experimental results summary. The results Prior to T2K take into account the values
presented previously in table 3.1. In T2K were considered sin2 θ23 = 0.5, |∆232 | = 2.4 × 10−3 eV2 and δ = 0.
the upper (lower) row corresponds to normal (inverted) neutrino mass hierarchy.
In the following sections we will present a more detailed information on the reconstruction of the mass
matrix from experimental results, namely on the determination of the mixing parameters and masses, as well
as results arising from other processes such as neutrinoless double β-decay.
3.1 Reconstructing mass matrices from experiments

In this section one shall analyse which parameters we can retrieve from experiments. As we will show
in chapter 4, for neutrinos in a Majorana framework one can fully reconstruct the mass matrix from three
17
mixing angles, three phases and three neutrino masses. Note that here we are only considering the decoupling
limit, as explained in section 4.4. There are four known methods with sufficient accuracy to measure these
parameters
1. Neutrino oscillations;
2. Neutrinoless double β-decay (0νββ), which can only occur if neutrinos are Majorana (see section 4.6
for a further analysis on this phenomenon);
3. Electron neutrino mass can be extracted from β-decay spectra;
4. The sum of neutrino masses can be inferred from cosmology.
The results presented in the beginning of the chapter had neutrino oscillation as its main source. Indeed,
one can determine, through the neutrino oscillation experiments, all three mixing angles θ12 , θ23 and θ13 up
to a good precision. The presented results show clearly that the former two are already well determined,
although their error still can decrease, when one considers all the experiments running at the moment.
Moreover, regarding θ13 , as mentioned before, many efforts are being placed on its determination, and one
expects to have a better precision, especially when T2K restarts its beam. Moreover, there is also a faint
chance for the ‘Dirac’ phase δ to be measured via long-baseline νµ → νe oscillation searches, provided a
not too small θ13 . These experiments are also expected to improve the neutrino mass-squared difference
bounds and determine ∆231 sign [19]. Note, however, that a precise and unambiguous measurement of δ
would probably need higher level experiments than the ones we deal with today, namely a neutrino factory.
There are three more parameters one should obtain to fully determine neutrino mass matrix - the overall
scale of neutrino masses and Majorana phases. To obtain the former, one considers the methods 2-4, which
measure different but complementary quantities. As will be more evident in section 4.6, double β-decay rate
would scale with the absolute square of the so called effective neutrino mass, which takes into account the
neutrino mixing matrix U as the following shows
X
2
Γ0νββ ∝ Uei mi ≡ |mee | (3.2)
Considering the standard parametrisation, eq. (3.1), it comes out clear that we may have Majorana masses,
CP violation and at the same time a vanishing mee . One can easily check that by taking the case of inverted
hierarchy, where m1 ' m2 , and by recalling that Ue3 ' 0, we can see that different contributions may
interfere destructively in the 0νββ amplitude. Nevertheless, it important to keep in mind that the above
relation gives us rather good constraints on neutrino masses for most of the parameter space.
In the case of the investigation of a β-decay spectrum, usually the ‘average electron neutrino mass’ m(νe )
is determined
X
m2 (νe ) ≡ |Uei |2 mi 2 , (3.3)
which is a coherent sum and therefore not sensitive to phases. One must also mention that cosmology can
provide us an indirect measurement of neutrino masses. Indeed, cosmology is sensitive to the energy densities
of the various components of the Universe, and specifically to the sum of the physical mass of neutrinos
n
X
mi . (3.4)
i=1
18
Measurements of this result rely on the improvement of both observational techniques and larger data sets.
They obviously also depend on the model, but when these results are obtained, conservative models, such as
the Standard Model of Cosmology, must be considered.
All these methods are able to provide three upper bounds regarding neutrino masses, but no positive
detection of these mass combinations was performed, being the respective upper bounds the following [20–22]
|mee | . (0.21 − 0.53) eV (90% C.L.), (3.5)

2
m (νe ) < 2.3 eV (95% C.L.), (3.6)
X
mi . 0.5eV. (3.7)
Note that, as mentioned before, the limits set by experiments are expected to drop in the future. For instance,
the bounds set by 0νββ are expected to be lowered in the future, unless the nuclear physics of this process
spoils this intent. Clearly the limit set by experiments involving β-decay date from late 90’s and are still
high, but it is expected that the KATRIN experiment can be sensitive to neutrino masses down to 0.2eV [23].
The cosmological probes are also expected to lower the limits, testing neutrino masses down to 0.1eV.
The other two remaining observables are the ‘Majorana’ phases, which are measurable only in processes
that violate lepton number. At present time, the only such process that seems viable experimentally is
neutrinoless double β-decay and it is sensitive to the particular combination of ‘Majorana’ phases. However,
it is important to stress that the relation between the rate for neutrinoless double β-decay and the ‘Majorana’
phases, eq. (3.2), is obtained under the assumption that neutrino masses are the only source of lepton number
violation in this process. Moreover, for given oscillation parameters, the ‘Majorana’ phases should not be
too close to 0 or π/2 (which as we shall show in section 5.2 are CP conserving cases) and the uncertainty
on both the experimental and theoretical side (i.e., the nuclear matrix elements) should be small [24]. Also
note that this process only helps us determining one phase, while for the remaining it seems that there is no
realistic experiment able to disclosure it [19].
3.2 Neutrino oscillations experiments

The results we presented in the beginning of the chapter regarding θ12 , θ13 , θ23 , ∆m221 , |∆m231 | is due to
the spectacular developments in neutrino oscillation experiments in the last few years. In this context, we
present the big picture of how these experiments have been evolving, how we retrieve the parameters from
them, and we will give a special focus on θ13 and the possibility of measuring δ.
The first neutrino experiment started in the late 1960’s and was performed by Davis and collaborators at
Homestake. In that experiment, νe is absorbed in the reaction νe +37 Cl →37 Cl+e− . From this very beginning
of the solar-neutrino observation, it was recognized that the observed flux was significantly smaller than the
one predicted by the Solar Standard Model (SSM). This deficit has been called ‘the solar-neutrino problem’.
Since then, further studies have been performed regarding solar neutrinos, such as the Gallium experiments
71
Ga (GALLEX and GNO at Gran Sasso in Italy and SAGE at Baksan in Russia), Borexino, and Sudbury
Neutrino Observatory (SNO), which combined with Kamiokande and later on with Super-Kamiokande’s
(SK) high statistics νe elastic scattering in a large water-Cherenkov detector, provided direct evidence for
flavour conversion of solar neutrinos [25, 26]. Furthermore, SNO and SK-I also gave the compelling evidence
for muon neutrino disappearance which is consistent with two-neutrino oscillation νµ ↔ ντ . These two
detectors, together with other experiments, including the reactor results from KamLAND or Chooz (to be
non-exhaustive), provided very good determinations on the neutrino parameters, as summarised in table 3.1.
19
It is clear that despite the valuable results obtained from all these experiments, they are limited in terms
of possibilities. Indeed, we are entering the precision era and we expect having a better control of the
systematic uncertainties and beam flux. For that reason, we started running long-baseline experiments. In
such experiments, we basically produce a neutrino beam, with a certain energy E, and then we detect it a
few hundred kilometers away, where oscillations become apparent.
So let us understand how these variables, the energy E of the beam and the L distance of the detector,
are defined. For that purpose consider the transition probability, which we will derive in section 4.5, written
in the following way
X ∆m2αβ L
∗ ∗
Pνi →νj (L, E) = δij − 4 Re Uiα Ujα Uiβ Ujβ sin2 (3.8)
4E
α>β
X ∆m2αβ L
∗ ∗
+2 Im Uiα Ujα Uiβ Ujβ sin .
2E
α>β
In two flavour limit, i.e. when ∆212 ∆223 ≈ ∆213 , the neutrino oscillation maximum yields at
∆m2αβ [eV2 ]L[km] π

2.54 = (3.9)
4E[GeV] 2
which means that, for ∼ 1GeV neutrinos, ∆m212 ∼ 10−4 would require a baseline of order the Earth’s
diameters, while for atmospheric neutrinos we only need a baseline of a few hundred kilometres1 (note that
we are neglecting matter effects.). These two variables are set in such a way to match certain physics goals.
Consider now the beam production. The basic idea behind these experiments is to obtain a rather pure
νµ beam. To obtain such beam, firstly a proton beam is created and smashed into fixed targets, producing a
large set of hadrons. In the sequence, there is a magnetic focusing of positive pions and kaons. After focusing
those particles, we set a ‘decay tunnel’, so that the decays π + → µ+ νµ and K + → µ+ νµ occur, and we get
a collimated beam of νµ . In this process we see that the neutrino energy spectrum is naturally determined
by decay kinematic and magnetic focussing optics and that the beam is rather pure [27]. Table 3.3 presents
the main long-baseline oscillation experiments.
Experiment Run Peak Eν Baseline Detector
1st Generation K2K 1999-2004 1 GeV 250 km Water Č
NuMI/MINOS 2005-2011+ 3 GeV 735 km Iron/Scint

2nd Generation
CNGS/Opera 2008- 17 GeV 735 km Emulsion
T2K 2010- 0.7 GeV 295 km Water Č

3rd Generation
NOνA (?) 2013(?)- 1.8 GeV 810 km Liq. Scint.
Table 3.3: Five main Long-baseline neutrino oscillation experiments.
These experiments were conceived regarding the following main experimental goals
i) K2K: confirm atmospheric neutrino oscillations;

1 Most of these experiments sample unoscillated neutrino beam close to production to help in the modulation of the beam
20
ii) MINOS: precise measurement of |∆32 |2 (and θ23 ), and, as long as not too small, try a shot at θ13 (to
which only achieved a reliable upper bound);
iii) Opera: observe tau appearance in νµ ↔ ντ oscillations;
iv) T2K: observe νµ ↔ νe oscillations and measurement of θ13 ;
v) NOνA: observe νµ ↔ νe oscillations at a longer baseline to measure θ13 (similarly to T2K). Due to
the longer distance, it has some sensitivity to the mass hierarchy. Combining T2K and NOνA would
improve the results.
In these experiments, and specially in the latter two in which we shall focus on, we are mostly interested
in θ23 and θ13 . These parameters are related, respectively, to two different phenomena: νµ disappearance
and νe appearance. To understand such statement, consider the survival probability of νµ , obtained from
(3.8), as
Pνµ →νµ = 1 − 4|Uµ2 |2 |Uµ1 |2 sin2 (1.27∆m221 L/E)

− 4|Uµ3 |2 |Uµ1 |2 sin2 (1.27∆m231 L/E)
− 4|Uµ3 |2 |Uµ2 |2 sin2 (1.27∆m232 L/E) (3.10)
Neglecting, since we are considering a baseline of a few kilometres, the contribution of the term dependent
on ∆m21 , taking ∆m232 ' ∆m231 and using three-generation unitarity of PMNS matrix (|Uµ1 |2 + |Uµ1 |2 =
1 − |Uµ3 |2 ), we obtain
Pνµ →νµ ' 1 − 4|Uµ3 |2 (1 − |Uµ3 |2 ) sin2 (1.27∆m232 L/E)

= 1 − cos4 θ13 sin2 2θ23 1 − sin2 2θ13 sin2 (1.27∆m232 L/E). (3.11)
By neglecting the second term inside parentheses, due to its dependence on sin2 2θ13 , the νµ disappearance
probability yields
Pνµ →νx ≈ cos4 θ13 sin2 2θ23 sin2 (1.27∆m232 L/E). (3.12)
Analogously we can obtain the dependence on θ13 , which is obtained through νe appearance. Its probability
is the following
Pνµ →νe ≈ sin2 θ23 sin2 2θ13 sin2 (1.27∆m232 L/E). (3.13)
Recall that θ13 are the main goals of T2K and NOνA. Note that in the above analysis we did not take into
account the effect of matter, which can be relevant. Finally, it is also important to understand whether we
can find an evidence of (Dirac) CP violation in these experiments. A measure of CP violation would be
the difference of oscillation probabilities between neutrinos and anti-neutrinos, Pνi →νj and Pνi →νj , which is
given by [28] (for oscillation in matter see [29])
∆Pννij ≡ Pνi →νj − Pνi →νj = ±16J sin ∆21 sin ∆31 sin ∆32 , (3.14)
where ∆αβ ≡ ∆m2αβ L/4E and J is a measure of CP violation, and is given in the standard parametrisation
21
by
J = s12 c12 s23 c23 s13 c213 sin δ. (3.15)
To obtain this result we cannot neglect the solar term, otherwise we would not obtain any CP violation.
Furthermore, we see that CP violation is maximum when ∆32 = (2n + 1)π/2 and grows with n since sin ∆21
grows with n. Notice also that for CP violating terms to be non-zero, the kinematical phases ∆αβ (namely
∆32 ) cannot be nπ. To keep things simple we shall not proceed with an exhaustive analysis, specially because
for an adequate description of neutrino propagation in the Earth it is very important to take into account
the effect of matter, since it can induce a fake CP violating effect even if the CP phase is zero or π - such
analysis can be found in [29] and we shall summarise it now. From eq. (3.15) it is clear that if θ13 is too
small, no leptonic CP violation can be observed. Moreover, mass hierarchy plays a role on determining
leptonic CP violation. Indeed, NOνA alone may be able to determine the hierarchy if sign(∆m231 )sin δ is
close to one. Otherwise a combination of T2K and NOνA is our best option for determining hierarchy. And
even if hierarchy is not determined, there is still a substantial region of (θ13 , δ) space which still allows the
observation of leptonic CP violation, mostly from the combination of T2K and NOνA which can be a very
powerful tool to observe CP violation, as long as θ13 is not too small.
22
Chapter 4
Minimal extension of the Standard Model of

particle physics
Neutrino mass is the basis for many of the properties that the neutrino can acquire, namely mixing, oscil-
lations, decays, magnetic moments, possibility of CP violation in the leptonic sector, etc.. However, in the
Standard Model (SM) neutrinos are strictly massless. By considering neutrinos massless, the SM made them
essentially different from other fermions such as the charged leptons (e, µ, τ ) and the quarks (u, d, c, s, t,
b), which are known to have masses.
Contrarily to these particles, in the SM only one helicity state per each neutrino is present. Thus,
neutrinos cannot have a Dirac mass term, which requires both helicity states to hold. Neutrinos could,
in principle, have a Majorana mass term, which requires only one helicity state of a particle and uses the
opposite helicity state of the antiparticle. However it is important to stress that since νL is part of the SU (2)
doublet, the Majorana mass term transforms as a SU (2)L triplet: it is not gauge invariant. As one shall see
in seesaw mechanism type II, section 4.2, to obtain such a mass term one must introduce another field, a
scalar triplet, in order to keep the symmetry SU (2) × U (1) of the Lagrangian.
This type of mass also breaks lepton number by two units. When one inspects the SM Lagrangian, one
has a exact lepton number symmetry even after symmetry breaking, and therefore those terms can never
arise in perturbation theory. This means lepton number conservation is a fortuitous ‘accident’ in the SM, and
that neutrinos are massless to all orders in perturbation theory. One could also ask whether non-perturbative
effects could induce a neutrino mass in SM. Anomalies play a role here, and when one considers the lepton
number L, such current is broken. However, it also happens that when one considers the baryon number
B current, which is also accidentally conserved to all orders in perturbation theory, its non-perturbative
violation has also an identical form, so that B − L current is conserved to all orders in gauge couplings.
Having in mind that the neutrino mass operator violates B − L, then neutrino masses cannot acquire any
mass even in the presence of non-perturbative effects.1
Contrast this issue with the masslessness of the photon, which is massless because of a conserved gauge
symmetry which, in turn, governs the electromagnetic interaction. Neutrinos have no such symmetry principle
in the SM, so things were arranged to produce such result. This was a consequence of the experimental data
from the past, even though it was against the expectations of many theorists.
The situation, however, is now drastically altered. Experiments provide an evidence that neutrinos have
mass, and as mentioned before, new physics arises from this fact. An usual extension to the SM is, for
instance, the consideration of a set of right-handed neutrino fields together with the seesaw mechanism type
1 In chapter 6 we shall explore the possibility of a non-perturbative violation of B + L, in the context of Leptogenesis.
23
I, to help us understand the smallness of neutrino masses. With the introduction of these specific fields, we
can also discuss whether neutrinos are Dirac particles or Majorana particles. They can of course be Dirac
particles, like all charged fermions have to be. But there is also the possibility that neutrinos are their own
antiparticles since they do not appear to have electric charge. In fact, as explored in chapter 2, for a spinor
with no conserved charges, as neutrinos seem to, the Majorana mass term is perfectly valid.
With the introduction of a mass term and mixing matrix, some physical phases arise, and those can
be an evidence of CP violation, which is crucial to understand the existence of matter in the Universe. In
fact, one of the conditions to have a matter-antimatter asymmetry, provided symmetric initial conditions, is
the non-existence of CP invariance in the Hamiltonian. Therefore, to study this violation we will properly
describe the leptonic mixing matrix in this chapter of the study. We shall also describe neutrino oscillation,
neutrinoless double β-decay and finish this chapter by presenting the Sakharov conditions for baryogenesis.
Among those we find CP violation, and we will find that as an argument to address our efforts to an extended
analysis of CP violation later in chapter 5. Such violation, and specifically the mixing and mass matrix, is
naturally different when we consider neutrinos to be Dirac or Majorana, whose definitions and algebra we
already introduced in chapter 2, and will apply in the next section.
4.1 Extend Majorana masses to the SM

In this section, one shall particularize the mass term in (2.34) of section 2.1.4 to neutrinos and introduce
an extension of the SM. For this purpose, we identify the different Weyl spinors to neutrinos: ψL ≡ νL0 ,
T T
c 0 c 0 , ψ ≡ ν 0 , ψ c ≡ (ν 0 )c ≡ Cν 0 . Note that ν 0 are neutrinos that take part in electroweak
ψR ≡ (νR ) ≡ CνR R R L L L L
0
interactions and for that reason they are usually called active neutrinos, while νR will, in principle, have no
interaction besides gravitation in our effective theory. Considering this notation, the Dirac-Majorana mass
term, eq. (2.34), becomes
0
2Lm = mD (νL0 νR 0 )c (ν 0 )c ) + m∗ ν 0 (ν 0 )c + M (ν 0 )c ν 0 + h.c.
+ (νR L L L L R R R
! !
m∗ m 0 c
(νL )
L D
= νL0 (νR0 )c
0
+ h.c.. (4.1)
mD MR νR
We are now in position to advance to an extension of the SM by simply adding to the standard spectrum
a few heavy particles in a seesaw framework (for more details see section 4.3). In seesaw mechanism type
0
I we introduce, at least, one νR per light massive neutrino existent (taking in consideration the two mass
differences presented, we must have, at least, two right-handed neutrinos) from which we have MR 6= 0,
whereas in type II we introduce a scalar triplet, which induces mL 6= 0. Thus, considering to be natural
having three heavy neutrinos and taking in consideration eq. (4.1), the following is the most generic leptonic
mass term after spontaneous gauge symmetry breaking
1
Lm = −[lL ml lR + νL mD νR + νL Cm∗L νL T − νR
T −1
C MR νR ] + h.c.
2
1
= −nL CMnL T + nTL C −1 M∗ nL − lL ml lR − lR m†l lL , (4.2)
2
where mD , mL , MR and ml denote the neutrino Dirac mass matrix, the left- and the right-handed neutrino
c T
Majorana mass matrix and the charged lepton mass matrix, respectively, with nL ≡ (νL , νR ) . Since the
right-handed neutrinos are SU (2) × U (1) invariant, its Majorana mass term can have a value much above
the scale v of the electroweak symmetry breaking, leading through the seesaw mechanism, to naturally
24
small neutrino masses, as one shall see in section 4.3. Moreover, it is also clear why we could not have the
left-handed neutrino Majorana mass term in the SM. In fact, left-handed leptons are arranged in SU (2)
doublets
!
νL
LL = , (4.3)
lL
and have hypercharge YW ≡ 2(Q − T3 ) = −1, where Q stands for electric charge and T3 is the weak isospin of
the particle. Analysing the mass term νLT C −1 νL , it has a weak isospin T3 = 1, therefore a weak hypercharge
YW ≡ 2(Q − T3 ) = −2. This means we would need another Higgs field (not present in SM), this time with
T
YW = 2. Such a Higgs field could be, for example, part of a scalar triplet ∆ ≡ ∆++ , ∆+ , ∆0 , as it is
explored in the type II seesaw mechanism.
In addition, one should now define the relevant Lagrangian regarding weak interactions involving neutri-
nos. Having in mind the SM’s covariant derivative
g
Dµ = ∂µ + ieAµ Q − ig(Wµ+ T+ + Wµ− T− ) − i Zµ (T3 − Qs2w ), (4.4)
cw
one can derive the charged and neutral-current interactions of a fermion multiplet f from the general gauge-
kinetic Lagrangian
LGK = if γ µ Dµ f. (4.5)
Particularizing the above for the leptonic sector, the relevant charged-current interaction is the following
g
0 γ µ ν 0 W − + h.c..
LW = √ νL0 γ µ lL
0
Wµ+ + lL L µ (4.6)
2
Finally, one should also introduce the scalar sector of the SM, which consists of only one doublet, φ, with
Y = 1/2, and its SU (2) conjugate doublet, φ̃ (Y = −1/2), which are
!
ϕ+
φ= , (4.7)
ϕ0
!
∗ ϕ0†
φ̃ = iτ2 φ = . (4.8)
−ϕ−
In the above equation ϕ± are the Goldstone bosons to be absorbed in the longitudinal components of W ± ,
while from the complex field ϕ0 arises both the physical Higgs particle and the Goldstone boson absorbed in
the longitudinal component of Z, through the Higgs mechanism, which is explained in almost any Quantum
Field Theory book.
Following the introduction of these fields, we will now define their CP transformations. After that, we
shall introduce the general framework we will consider in the course of the work, starting from the seesaw
mechanism.
25
4.2 Additional relevant CP transformations
In the course of this work there are other relevant CP transformations in addition to the ones we mentioned
in section 2.2. This is, for instance, the case of the W bosons introduced in the previous section involving
the interaction Lagrangian, eq. (4.6). In order to find these CP transformations, one must introduce the P
and C of a scalar field φ, subject the Klein-Gordon equation ( + m2 )φ = 0. This field transformation reads
Pφ(t, ~x)P † = exp(iαp )φ(t, −~x),

Pφ† (t, ~x)P † = exp(−iαp )φ† (t, −~x),
(4.9)
Cφ(t, ~x)C † = exp(iαc )φ† (t, ~x),
Cφ† (t, ~x)C † = exp(−iαc )φ(t, ~x).
One should now recall that both a pure gauge Lagrangian [30] and the scalar potential of the SM with
only one doublet, must conserve CP. This means CP violation can only arise from the simultaneous presence
of Yukawa interactions and gauge interactions. As a consequence, if one considers the Klein-Gordon fields
ϕ+ and ϕ− combined in SU (2) doublets (4.7), which transform as follows
CPϕ+ (t, ~x)CP † = exp(iηW )ϕ− (t, −~x),

(4.10)
CPϕ− (t, ~x)CP † = exp(−iηW )ϕ+ (t, −~x),
and analysing the interactions in the kinetic term of the Higgs field (Dφ)† (Dφ) (the covariant derivative D
is defined in (4.4)), one must postulate the following CP transformations to assure the CP invariance of that
part of the Lagrangian
CPW +µ (t, ~x)CP † = − exp(iηW )Wµ− (t, −~x), (4.11)

CPW −µ (t, ~x)CP † = − exp(−iηW )Wµ+ (t, −~x), (4.12)
0 † 0
CPϕ (t, ~x)CP = ϕ (t, −~x), (4.13)
Furthermore, one should also define the relevant transformations for the lepton fields organized in families
in our theory. Since those fields are represented by spinors, one should consider both eqs. (2.50) and (2.52)
to obtain these transformations. Thus, considering the charged current interaction Lagrangian in (4.6), the
most general CP transformation one can have that leaves the gauge interaction invariant is the following
T T
0
(CP)lL (CP)† = U 0 γ 0 ClL
0 0
(CP)lR (CP)† = V 0 γ 0 ClR
0
T T
(4.14)
(CP)νL0 (CP)† = U 0 γ 0 CνL0 0
(CP)νR (CP)† = W 0 γ 0 CνR
0 ,
and
0 (CP)† = −l0T C −1 γ 0 U 0†
(CP)lL 0 (CP)† = −l0T C −1 γ 0 V 0†
(CP)lR
L R
(4.15)
(CP)νL0 (CP)† = −νL0T C −1 γ 0 U 0† 0 (CP)† = −ν 0T C −1 γ 0 W 0† ,
(CP)νR R
where U 0 , V 0 and W 0 are unitary matrices containing the CP phases and they act in flavour space. This
condition means families can mix under a CP transform, while unitarity must be required in order to preserve
all fermion Lagrangian terms. Taking this into account, and since we required the charged current term
νL γ µ lL to stand in the CP transformation, the unitary matrix U 0 is the same acting on lL and νL . This also
means that in the right-handed spinors sector, V 0 and W 0 are not required to be the same.
26
These transformations regarding the gauge bosons and the lepton fields will be extremely useful in chapter
5, where we shall define conditions for CP to hold.
4.3 Seesaw mechanism

The most elegant way to implement the required neutrino masses into the SM is via the seesaw mechanism.
There are several realizations of this mechanism which allows neutrinos to acquire very small masses. In
exchange, very heavy particles are added to the particle spectrum. For instance, the original seesaw [2–5],
today called type I, includes heavy neutrinos in the particle content, which, being singlets under the SM gauge
symmetry, can acquire a heavy Majorana mass without a SSB. The attractive feature of this mechanism is
that, from this only assumption, light masses for neutrinos are generated.
Considering all its realizations, the seesaw mechanism allows us to obtain a Majorana mass term of the
form
mν νLT C −1 νL (4.16)
where mν is
v2
mν ∝ cν . (4.17)
M
In this generic neutrino mass matrix we have the coupling of neutrinos, cν , the vacuum expectation value
of the Higgs doublet, v, and a mass large mass M v - for instance, MR in type I and m0 /m2∆ in type
II [10, 31–34] - which automatically induce neutrinos to be lighter than the charged fermions. Clearly, one
would be able to lower mν even considering M ' v (or even M v) just by taking yν 1, however this is
not natural, having in mind all the Yukawa coupling of other fermions.
One should also note that with this mechanism we can also overcome the 105 factor difference between
neutrinos mass (around 1 eV or lighter) and the next lighter particle, the electron (me = 511 keV). However,
it is also true that this mechanism is unable to completely fix the overall scale of the light neutrino masses,
since heavy particle’s mass scale is supposed to be high, but it is not precisely known. Moreover, the lepton
mixing angles remain arbitrary: one can easily obtain any desired value, just by manipulating its parameters.
The seesaw mechanism may suffer some criticism due to this, but the fact is that neutrino data is really hard
to obtain and this mechanism provides the most natural and attractive explanation of the smallness of the
neutrino masses.
4.3.1 Type I seesaw mechanism

The Standard Model, which is based on the gauge group SU (3)c × SU (2)L × U (1)Y , does not include singlet
right-handed neutrinos in its spectrum. In the scope of seesaw mechanism type I, which we will consider,
for instance, in leptogenesis, we will add 3 right-handed fields to the particle spectrum. In SM, the quarks
and leptons transform as QL (3, 2, 13 ), uR (3, 1, 34 ), dR (3, 1, − 23 ), L(1, 2, −1), eR (1, 1, −2), and the Higgs boson
H, responsible for electroweak symmetry breaking, transforms as (1, 2, 1). Analysing SM fermions list, we
can easily conclude that neutrinos do not get mass as a result of the Higgs mechanism due to the fact that
right-handed fields νR are not in its spectrum; as a result, there is no hν L̄φνR coupling that could have given
mass to the neutrinos after symmetry breaking.
Without any further explanation, with the introduction of right-handed neutrinos one would expect the
27
neutrino masses arising from the hν L̄φνR Yukawa coupling to be of the same order as the quark and charged
lepton masses. This is, of course, experimentally excluded, since electron mass, for example, is known to be
at least 105 larger than neutrino masses. In this context, the seesaw mechanism provides a natural reason not
only for the smallness of neutrino masses, but also for this huge factor. It exploits the fact that only neutrinos
can have Majorana mass, as we have seen before. Ironically, this model provides a better understanding to
the lightness of the neutrinos than SM or any other theory can provide to the generational hierarchy factor,
namely the 106 factor between electron and top quark masses.
The number of heavy neutrinos depends on the model. It must be, at least, equal to the number of massive
but light neutrinos. Since, from the current experimental data, we know that two mass differences between
light neutrinos are non-zero (∆221 , |∆232 | =
6 0), there must be, under this mechanism, at least 2 right-handed
neutrino fields introduced. In this study most of our results will take in consideration 3 heavy neutrinos. It
may not be correct, but it surely is plausible, in analogy with the quark sector.
It is also interesting to find that, in the scope of this extension, the introduction of right-handed fields
not only allows us to have the Dirac mass term, but we have to consider the Majorana masses for heavy
T −1
neutrinos. Indeed, to avoid the presence of such a mass term (1/2νR C MR νR as defined in (4.2)), one is
required to impose an extra symmetry to the Lagrangian (for example, lepton number), to overcome the fact
that this term cannot be trivially eliminated, since its hypercharge is YW = 0. Moreover, we see that the
introduction of a νR , which we need for this mechanism, is very appealing from the structural quark - lepton
symmetry.
For definiteness we should firstly consider one left- and one right-handed field. The mass term must
include, as previously stated in (4.2), both Majorana and Dirac mass terms. In this notation we should note
√
that the Dirac mass is given by mD = hν v/ 2, since it results from the Yukawa coupling constant hν and
the scalar field VEV v. MR does not result from any spontaneous symmetry breaking. Thus, the total mass
matrix is √ !
0 hν v/ 2
M= √ . (4.18)
hν v/ 2 MR
Considering the analysis in the appendix B, we see that if MR mD , neutrinos can be approximated as
Dirac particles. For mD MR the Majorana nature of the particle plays an important role, and this is the
case we are going to consider, since the gauge invariant scale MR is expected to be above MW . In this limit,
one can straightforwardly compute M eigenvalues, which are
(hν v)2
− (4.19)
2MR
and MR , and to check that the respective approximate eigenvectors are νL and νR . By this mechanism, one
has a neutrino mass much smaller that the typical fermion masses (which are of order hν v), as long as the
√
limit MR hν v/ 2 stands.
We should now consider that there are m right-handed neutrinos and n left-handed neutrinos, where m
is, at least, as much as the number of massive left-handed neutrinos, otherwise the mass matrix determinant
is zero, and we have zero mass neutrino(s). Taking this into consideration, the most general (n+m)×(n+m)
Majorana mass matrix with mL = 0 that mixes neutrinos is given by
!
0 mD
M= (4.20)
mD T MR
28
√
with (mD )ij = hij v/ 2 (we dropped the ν subscript for notation simplicity). The Lagrangian involving
neutrinos, as we introduced in chapter 4.1, has diagonal charged gauge currents, eq. (4.6), and a non-trivial
symmetric complex mass matrix (eq. (4.2)). To diagonalize it, we can use the property that any complex
symmetric matrix can be diagonalized by one unitary matrix and its transpose. Therefore, when one considers
the following unitary transformation ! !
νL0 νiL
=V , (4.21)
νL0 0 NjL
T
where νL0 0 ≡ νR
0 and νiL and NjL are the physical basis of neutrinos, in an extension of the SM using m
right-handed neutrinos the neutrino mass matrix is diagonalized by
V T M∗ V = D, (4.22)
where D = diag(mν1 , ..., mνn , MN1 , ..., MNm ), with mνi and MNi denoting the physical masses of the light
and heavy Majorana neutrinos, respectively. It is convenient to write V and D in the following form
!
K R
V = ; (4.23)
S T
!
d 0
D = . (4.24)
0 D
Inserting both in eq. (4.22) one obtains
S † mD T K ∗ + K † mD S ∗ + S † MR S ∗ = d (4.25)
† T ∗ † ∗ † ∗
S mD R + K mD T + S MR T = 0 (4.26)
T † mD T R∗ + R† mD T ∗ + T † MR T ∗ = D (4.27)
From eqs. (4.25-4.27), we see that the mixing between light and heavy neutrinos, eq. (4.21), is small, as both
R and S are of order mD /MR . With this relation we can obtain from (4.26) up to an excellent approximation
S † ≈ −K † mD MR −1 . (4.28)
We can use (4.28) and its transpose in eq.(4.25) to obtain the usual seesaw formula
d ≈ −K † mD MR −1 mD T K ∗ , (4.29)
and define the effective mass matrix for light neutrinos as
mIef f ≈ −mD MR −1 mD T , (4.30)
which is an very good approximation and completely demonstrates the way this mechanism works. On the
one hand, it shows that we can naturally have light neutrinos whose mass can be as small as we want, as
long as the Majorana mass MR from right-handed neutrinos is large enough. On the other hand
D ≈ T † MR T ∗ (4.31)
29
shows that we have m neutrinos whose mass matrix is of order MR , which means these neutrinos are very
massive.
However, it is important to stress one of the main disadvantages of this mechanism, clear from eq. (4.29).
√
This equation shows that, if one considers light neutrinos in the 0.1eV scale, with mD = hν v/ 2 ∼ 100GeV
in the electroweak scale, then heavy neutrino mass is around MR ∼ 1014 GeV . This means there cannot be
a direct evidence of heavy neutrinos, which is, of course, a very unattractive feature in a theory.
4.3.2 Type II seesaw mechanism

The most general possibility regarding the Majorana mass matrix must contemplate the matrix mL as a
generic symmetric matrix, instead of having the zero entry in (4.20) as considered in seesaw mechanism type
I. On the contrary, in the seesaw mechanism type II the term νLT C −1 m∗L νL in the Lagrangian (4.2) holds,
being a direct ‘left-left’ contribution.
Hence, we will take Type II as a very attractive version of seesaw mechanism with mass terms both from
the ‘left’ and the ‘right’ sector. Strictly, one would not need a ‘right’ sector, singlet under SU (2) × U (1), and
most calculations, as we shall see, are independent of it, but in this study we will hold the right-handed terms,
so we can obtain a more generic mass matrix. Furthermore, considering both mL , MR 6= 0 can be a very
interesting feature in many theories, specifically in some Grand Unification Theories schemes where there is
some parity symmetry, e.g. SU (2)L × SU (2)R × U (1)B−L . It is important to stress that in our theory we
need an extra Higgs multiplet, since we should preserve our effective Lagrangian symmetry SU (2)L × U (1)Y ,
having in mind that the left mass term stated above has hypercharge YW = −2. This means our Higgs field
coupling to νLT C −1 νL must have YW = 2.
In the following we present the most straightforward case, where a single Higgs triplet is added to the
particle spectrum. Consider the following gauge triplet of SU (2)
 
∆++
 
∆ =  ∆+  . (4.32)
∆0
T
One should now also consider two SU (2) doublets with left-handed fields, (νLi , lLi ) . In order to get
the Yukawa Lagrangian, we must get a singlet (SU (2) invariant) from these fields. To meet this purpose we
know, from group theory, that 3 ⊗ 2 ⊗ 2 = 3 ⊗ (3 ⊕ 1) = 5 ⊕ 3 ⊕ 1 ⊕ 3. Hence, the term from which we obtain
the singlet is the direct product of the Higgs triplet (4.32) and the following triplet, obtained from the direct
product of the two left-handed SU (2) doublets
 
T
νLi C −1 νLj
 √1 T −1 
T
 2 (νLi C lLj + νLj C −1 lLi ) , (4.33)
T
lLi C −1 lLj
with C −1 guaranteeing the Lorentz invariance. With these two triplets we can build the Lagrangian that
couples the fermions to the Higgs triplet

1
LY ∆ = y∆ij ∆++ lLi
T
C −1 lLj − √ ∆+ (νLi
T
C −1 lLj + νLj
T
C −1 lLi ) + ∆0 νLi
T
C −1 νLj + h.c. (4.34)
2
where yij is the Yukawa coupling, and includes the factor √1 arising from the decomposition into a irreducible
3
30
representation using the Clebsch-Gordan coefficients. When we consider a spontaneous symmetry breaking
(SSB) with the neutral component of the Higgs-triplet developing a VEV, h∆0 i = δ, we find from last
T
equation another mass term for neutrinos, yij δνLi C −1 νLj .
The VEV δ must be much smaller than the electroweak scale, in order to have small neutrino masses.
Hence we need a very small yij or δ (or both). Fortunately, the latter condition can be explained by the
seesaw mechanism type II, so we do not have to impose any condition to the Yukawa coupling. Furthermore,
the very good prediction of the SM to the W and Z boson masses can easily rule out any large enough
contribution of this Higgs triplet. That is clear when we consider the terms with both W and Z of SM’s
covariant, in eq. (4.4), applied to the scalar triplet
      
∆+ 0 (1 − 2sw 2 )∆++
     ig  
D∆ = ... + ig W +  ∆0  + W − ∆++  − Z  (−sw 2 )∆+  . (4.35)
cw
0 ∆+ −∆0
If one considers now both symmetry breakings, hφ0 i = v and h∆0 i = δ, we have mass terms for the vectorial
bosons obtained from the gauge kinetic Lagrangian of φ and ∆, (Dφ)† (Dφ) + (D∆)† (D∆), which are
2 2
− + g v
LmW = mSM
W + m ∆
W W W = + g 2 2
δ W −W + (4.36)
2

1 1 g2 v2 2g 2 δ 2
LmZ = mSM
Z + m ∆
Z ZZ = + ZZ (4.37)
2 2 2cw 2 cw 2
which violates the relation m2W = c2w m2Z very well experimentally established by Large Electron Positron
Collider (LEP). Accordingly, we must have δ v.
To see how the VEV of ∆ develops a tiny vale, consider below the triplet mass term
V1 = m2∆ ∆† ∆ = m2∆ (|∆++ |2 + |∆+ |2 + |∆0 |2 ) (4.38)
where m∆ is real. We are now interested in the next order terms with the Higgs triplet. These terms must,
of course, respect the weak hypercharge symmetry. Therefore, there will be a cubic coupling between the
T
triplet, with YW = 2, and two SM Higgs doublets φ̃ ≡ ϕ0† , −ϕ− , with YW = −1, as well as the Hermitian
conjugate of it. Taking in consideration that it is again a direct product 3 ⊗ 2 ⊗ 2, such potential would be
the following
√
V2 = m0 (ϕ− ϕ− ∆++ + 2ϕ− ϕ0† ∆+ + ϕ0† ϕ0† ∆0 ) + h.c. (4.39)
where m0 is complex in general and has mass dimension +1. The potential V2 (4.39) is linear in ∆0 , which is
a lower order in comparison to the mass term in V1 (4.38). This determines that, for small ∆0 , V2 dominates.
Therefore one has a SSB when m0 < 0 (one should distinguish from the SM, where the quadratic term µ|ϕ|2
is negative and the quartic term λ|ϕ|4 is positive).
∆0 gets a VEV when one has the minimum of the potential, which is given by
∂V ∂V1 + V2
0= ≈ = m2∆ ∆0† + m0 ϕ0† ϕ0† (4.40)
∂∆0 ∂∆0
31
in the vacuum. With hφ0 i = v and h∆0 i = δ, one has
m0 v 2
δ≈− . (4.41)
m2∆
It is clear now that when one considers a large scalar triplet mass m∆ , δ is small, supposing we have a normal
scale m0 . Comparing to seesaw mechanism type I, where one considers MR mD , the seesaw mechanism
type II works on the assumption that m2∆ is not only positive, but is also much larger than the electroweak
scale, i.e. m2∆ v 2 . In short, with seesaw mechanism type II, one can obtain a light Majorana mass term
for left-handed neutrinos just by introducing a very heavy scalar triplet.
Consequently, in the context of the seesaw mechanism one can include the Yukawa mass term (4.34) after
SSB
T
LY ∆ = y∆ij δνLi C −1 νLj + h.c.
= −νL Cm∗L νL T + h.c. (4.42)
in the mass matrix we previously considered, (4.20). Hence, the most general Majorana mass matrix that
mixes left- and right-handed neutrinos is
!
m∗L mD
M= , (4.43)
mD T MR
where m∗L = y∆ δ. Accordingly, eqs. (4.25), (4.26), (4.27) take the following form
K † m∗L K ∗ + S † mD T K ∗ + K † mD S ∗ + S † MR S ∗ = d (4.44)
K † m∗L R∗ + S † mD T R∗ + K † mD T ∗ + S † MR T ∗ = 0 (4.45)
†
R m∗L R∗ † T ∗ †
+ T mD R + R mD T + T MR T = D.∗ † ∗
(4.46)
Since we have a small mL following the consideration of seesaw mechanism type II, the approximation
S † ≈ −K † mD MR −1 . (4.47)
is still valid. Therefore, from eq.(4.44) one can get the seesaw type II formula
d ≈ K † (m∗L − mD MR −1 mD T )K ∗ . (4.48)
with mν defined as
∗ −1
mII
ef f ≈ mL − mD MR mD T . (4.49)
It is evident that we are only in the context of seesaw type II when the first term of eq. (4.48) dominates.
When both terms are comparable, we may call it mixed or hybrid seesaw. In the case the second term
dominates, one can neglect the first term, and work under the scope of seesaw type I.
4.3.3 A final word on seesaw mechanism

Type I and Type II are the two most popular realizations of the seesaw mechanism. However, it is possible to
define another form for this mechanism or some modifications on the previously presented ones. For instance,
32
we may consider two Higgs doublets, two Higgs doublets and a scalar singlet, etc. to perform the seesaw
mechanism type II. But there is an alternative seesaw mechanism realization that worth mentioning. It is
the so-called type III and involves the introduction of a triplet fermion. We will not analyse it deeply, but
it recycles the idea of considering a triplet, as the seesaw type II does, but this time the triplet is fermionic
and one of the doublets in the coupling is a Higgs doublet. We can use group theory again to check that this
is indeed the case. Under this hypothesis we would have
∆L(TF ) = yT LT CTF φ + MT TFT CT F + h.c., (4.50)
and in exactly the same manner as before in Type I, one gets a Type III seesaw [35] for MT v
−1
mIII T 2
ef f ≈ −yT MT yT v . (4.51)
As in Type I, and considering we have at least two massive neutrinos, one would need at least two of such
triplets (or a triplet and a singlet). Moreover, it is important to refer that under the assumption that a single
type of new particles are added to SM, no more types of seesaw can produce a Majorana mass for neutrinos.
In summary, we showed that within the seesaw framework, we can obtain an effective theory at low
energy scale with a Majorana mass term, which provides a window to new physics at high scale, taking in
consideration the particles introduced. Clearly, one traded coupling yν between physical, observable particles,
to the unknown yD (or y∆ or yT ) coupling and the unknown masses of those particles. However, one can
relate this with another physics phenomenon. Indeed, this is quite the same that was happening before the
discovery of the W boson, where Fermi theory was describing low energy phenomena in an appropriate way.
The bottom line of this section is that the Majorana neutrino mass is rather suggestive from the theoretical
point of view and provides new physics at a higher scale, as is the case of leptogenesis, chapter 6. The
prediction that this mass term allows is the lepton number violation, ∆L 6= 0, which can be violated, for
instance, in processes such as neutrinoless double beta decay (0νββ), which we explore in section 4.6, or
same sign dilepton pair production at colliders [36].
What happens if the neutrino has a pure Dirac mass? If that is the case, mν = yD v as any the other
fermion, and we simply require a small yD . This simply extends the puzzle regarding the smallness of the
electron Yukawa coupling to the neutrino sector. If this is indeed the case, we will see in the next chapter
that similarly to what we have in the quark sector, we would only have one CP violating phase, inasmuch as
we consider three neutrinos in our theory.
4.4 Leptonic mixing matrix in low energy physics

In this section we explore the leptonic mixing matrix, usually referred to as the Maki-Nakagawa-Sakata
matrix, or as the Pontecorvo-Maki-Nakagawa-Sakata (PMNS) matrix, in recognition of the pioneering con-
tributions of these scientists to the physics of mixing and oscillation [37, 38], and how it is parametrized.
For that purpose, we shall also define two different frameworks for our studies: one to use when we want
to relate low energy physics to hypothetical high energy processes, where we consider the full seesaw model
with the heavy neutrinos and which we will alternatively call ‘non-decoupling case’; and a second one, to
be used when we are only having in mind low energy processes, where all the CP violating phases on the
leptonic sector are contained in the 3 × 3 PMNS mixing matrix. Such model is also called the ‘decoupling
case’. Firstly, we should note that the Lagrangian introduced in the beginning of this chapter, obtained after
Spontaneous Symmetry Breaking, is diagonal in its leptonic charged gauge interactions, eq. (4.6). Such a
33
basis is usually called a weak basis (WB), i.e. a basis where all gauge currents are real and flavour diagonal.
If, for instance, one considers another basis, as it is the mass-eigenstate basis, then the fermion mass terms
would be real diagonal, but the flavour mixing in the charged currents would be non-trivial.
Associated to a weak basis there are several transformations. These, which are usually called WB trans-
formations, are defined as transformations of the fermion fields that keep the gauge currents flavour diagonal.
Considering the lepton fields in our extension, the following transformations can be performed
lL → U 0 lL , νL → U 00 νL , l R → V 0 lR , νR → W 0 νR , (4.52)
where U 0 , U 00 , V 0 and W 0 are arbitrary unitary matrices. Analysing all transformations, it is clear that a
WB transformation must have
U 00 = U 0 (4.53)
so that the charged weak current remains diagonal. Regarding the right-handed fields, there are no constraints
in this extension of SM. We should also recall that we can always use the property that any unitary matrix
can diagonalized by a bi-unitary transformation. For instance, in the case of charged lepton mass matrix, its
diagonalization could be achieved by a transformation such as
U 0† ml V 0 = dl (4.54)
where U 0 and V 0 are, in general, different, and refer to the decomposition of the left- and right-handed
charged leptons in mass eigenstates. By this we mean
0 0
liL = Uiα lαL , (4.55)
0 0
liR = Viα lαR , (4.56)
where i = 1, 2, 3 are the flavour eigenstates, and α = 1, 2, 3 are the mass eigenstates. Also note that, in
general, we can treat the theory as a set of matrices that transform as
m∗L → U 0† m∗L U 0∗ ,
MR → W 0T MR W 0 ,
(4.57)
mD → U 0† mD W 0 ,
ml → U 0† ml V 0 ,
and this is equivalent to (4.52).

Physics does not depend on the choice of WB, which means that all WB lead to the same physical
parameters of the theory. In our extension, these physical parameters are the fermion masses and mixing.
Therefore, it is always possible to have WB where ml is real, diagonal and positive, as we did in eq. (4.54).
This WB transformation can be performed without any loss of generality, and the matrix V that diagonalizes
M (eq. (4.22)) will have the physical meaning. Within this context, having in mind the relation between
34
neutrino weak-eigenstates and mass eigenstates defined as
 
! i = 1, 2, 3 = j
0 νjL  
νiL = Viα ναL = (K, R)  k = 1, 2, ..m , (4.58)
NkL
α = 1, 2, ..., m + 3
where i refer to the left-handed neutrinos in the flavour basis, and j and k refer to left-handed neutrinos
in the mass eigenbasis, the leptonic charged current interactions in the physical basis considering the high
energy effects are
g 0 g
Lhe−ph
W = √ lL γµ νL0 W µ + h.c = √ liL γµ Kij νjL + liL γµ Rij NjL W µ + h.c.. (4.59)
2 2
From the above equation follows that K and R give the charged current couplings of charged leptons to the
light neutrinos νj and to the heavy neutrinos Nj . When we want to study a connection between low energy
physics (e.g. neutrino oscillation) and high energy processes, we want to keep both terms K and R, otherwise
we would not be able find a relation between both. This is useful when, for instance, one tries to relate low
energy CP violation to CP violation relevant at higher energy scales, as we will consider later in this study
regarding leptogenesis.
mD
In the case one’s concern is about low energy physics, and recalling that we have a very small R ∼ MR ,
one neglects R, and considers that only charged leptons and light neutrino couplings to the W boson are
relevant. Moreover, from the fact that left-handed neutrinos and charged leptons are not diagonalized in the
same mass eigenbasis arises a nontrivial mixing matrix that can be either analogous to the quark sector, or
can contain more phases - the ‘Majorana’ phases - when compared to the one considered in the quark sector,
the so-called Cabbibo-Kabawashi-Maskawa (CKM) matrix. Such matrix is parametrised in the following,
and in such a convenient form which will allow us to separate the phases arising from a CKM-like matrix,
and the other (‘Majorana’) possible phases.
To obtain this matrix, consider a basis where both charged leptons and neutrinos are diagonalised, so
that the charged current interaction in the physical basis reads
g
Lle−ph
W = √ liL γµ Uij νjL W µ + h.c.. (4.60)
2
where U is the lepton mixing matrix and previously referred to as the Pontecorvo-Maki-Nakagawa-Sakata
(PMNS) matrix. This matrix is unitary by definition and physical meaningful since it relates neutrino
weak-eigenstates and mass eigenstates (redefines eq. (4.58)) as
!
0 i = e, µ, τ
νiL = Uiα ναL , (4.61)
α = 1, 2, 3
diagonalises mν , i.e. UPT M N S mν UP M N S and that, in the context of the seesaw mechanism, we identify
mν = m∗ef f , UP M N S = K ∗ , (4.62)
35
where U is usually presented as  
Ue1 Ue2 Ue3
 
U = Uµ1 Uµ2 Uµ3  (4.63)
Uτ 1 Uτ 2 Uτ 3
To meet the weak current Lagrangian in the physical basis, eq.(4.60), we take the mass term with diagonalised
fermion masses to be
1
Lle−ph
m =− νL d νLc + νLc d νL − lL dl lR − lR dl lL , (4.64)
2
where d and dl are real and positive matrices and refer to the physical masses of, respectively, light neutrinos
and charged leptons. Note that the general mass term for leptons in the decoupling limit is
1 0 ∗ 0c
0 m l0 − l0 m† l0
Lle
m =− νL mν νL + νL0c mν νL0 − lL l R R l L (4.65)
2

1 0 T

=− ν Cm∗ν νL0 − νL0T C −1 mν νL0 − lL0 m l 0 − l 0 m† l 0 , (4.66)
l R R l L
2 L
and the matrices transformations under a WB
mν → U 0T mν U 0 ,
(4.67)
ml → U 0† ml V 0 .
4.4.1 Parametrisation
Let us analyse the number of phases in the n × n PMNS matrix, and then particularise to 3 × 3. For that
purpose, recall that we have, in principle, freedom to rephase n charged lepton fields
li → li0 = exp(iθi )li (4.68)
with arbitrary θi ’s, leaving the charged lepton mass terms mi li li invariant. In the case of neutrinos, if we
consider their Majorana nature, the rephasing
νj → νj0 = exp(iϕj )νj (4.69)
T
with arbitrary ϕj ’s is not allowed, since it would not keep the Majorana mass terms νLj C −1 mj νLj invariant.
Note, however, that the rephasing of (4.69) by ϕj = nj π is still possible as long as we keep nk integer.
Naturally, when we consider a Dirac mass term νL mD NR for neutrinos, we can take an arbitrary ϕj .
Now consider the unitarity of PMNS, so that it is characterised by n2 conditions.2 These conditions
lead, for instance, to the unitary triangles, which will be subject to a later treatment, in section 5.5. The n2
parameters in the unitary mixing matrix can in principle be parametrised by
n(n − 1)
Mixing angles (4.70)
2
n(n + 1)
Phases. (4.71)
2
From this point on, the number of phases will depend on whether we are considering a Dirac mass or
2 Note that unitarity not only a very good approximation but also a very useful one, otherwise we would have to consider,
under the framework of the seesaw mechanism, n2 moduli and n(n − 1) phases, from a initial set of 2n2 parameters.
36
a Majorana mass for neutrinos. If we consider a Dirac-type, both the charged lepton rephasing (4.68) and
neutrino rephasing (4.69) are allowed. Therefore, one is free to rephase the charged current interaction
Lagrangian (4.6) by the following
n,n
X g
√ lLi γµ e−iθi Uij eiϕj νLj W µ + h.c., (4.72)
i,j
2
where we replaced the Einstein summation notation by an explicit summation. One is free to reorganize this
Lagrangian in terms of its phases as the following equation shows
n,n
X g
√ (lLi γµ e|−i(θ{z
e −ϕ1 ) −i(θi −θe ) i(ϕj −ϕ1 ) µ
} e| {z } Uij e| {z } νLj )W + h.c.. (4.73)
i,j
2
1 n−1 n−1
In the above equation it is clear how many phases can be eliminated. From e−i(θi −θe ) and ei(ϕj −ϕ1 ) and
since we are running i, j = 1, ..., n one can eliminate n − 1 for each (when θi = θe or ϕj = ϕ1 no phase can
be eliminated). Furthermore, the first exponential allows us to eliminate another phase, which means, in the
overall, the number of independent phases that can be eliminated is 1 + 2(n − 1) = 2n − 1. This number
will subtract to the number of phases the generic unitary matrix is parametrised (4.71), in order to have the
total of
(n − 1)(n − 2)
physical phases. (4.74)
2
Particularizing for the case of n = 3 we are interested in, one has

(
3 Mixing angles
n=3 (4.75)
1 Physical phase
which is the same we have in the quark sector. Analogously to that sector, one can parametrize U in different
ways. To obtain the standard one, let us consider a complex rotation in a − b parametrised as the following
!
cab eiφab sab
ωab = (4.76)
−e−iφab sab cab
This is equivalent to the redefinition of two fields (e.g. the charged lepton fields in (4.68)), but not the other
field (e.g. the neutrino field in (4.69)). This clearly leads to a 2 × 2 unitary matrix with one complex phase
as the above.
When one considers three neutrinos, the lepton mixing matrix can be parametrized as three complex
rotations around three axis, i.e. [10]
U = ω23 ω13 ω12 (4.77)
where each factor in the product of the ω’s is effectively 2 × 2, characterized by an angle and a CP phase.
Two of the three angles are involved in solar and atmospheric oscillations, so we set θ12 ≡ θ and θ23 ≡ θA .
In this parametrisation of the leptonic mixing matrix, U can be written as
 
c12 c13 s12 c13 e−iφ12 s13 e−iφ13
 
U = −s12 c23 eiφ12 − c12 s23 s13 ei(φ13 −φ23 ) c12 c23 − s12 s23 s13 ei(φ13 −φ12 −φ23 ) s23 c13 e−iφ23  . (4.78)
s12 s23 ei(φ12 +φ23 ) − c12 c23 s13 eiφ13 −c12 s23 eiφ23 − s12 c23 s13 ei(φ13 −φ12 ) c23 c13
37
where cij ≡ cos θij , sij ≡ sin θij and, without any loss of generality, all θij are in the first quadrant. This
matrix, the way it is presented, is a possible parametrisation of the leptonic mixing matrix if neutrinos are
Majorana fermions. However, since we are firstly interested in a Dirac mass term, we take φ12 = 0 = φ23 so
that we only have one CP violating phase - it is impossible to identify that as a rephasing of the lepton fields
(eq. (4.73)). This physical phase, crucial for the existence of CP violation in the quark sector, led to the
prediction of third quark generation, and it is usually referred to as a Dirac-type phase for obvious reasons.
One can now identify φ13 ≡ δ and present the standard parametrisation for a mixing matrix in a ‘Dirac’
neutrinos context [15]
 
c12 c13 s12 c13 s13 e−iδ
 
UD = −s12 c23 − c12 s23 s13 eiδ c12 c23 − s12 s23 s13 eiδ s23 c13  , (4.79)
s12 s23 − c12 c23 s13 eiδ −c12 s23 − s12 c23 s13 eiδ c23 c13
Notice that since δ is not a rephasing invariant quantity, it is only meaningful within a given parametrisation.
Let us now consider a Majorana-type mass for neutrinos. In this context, if the neutrinos are massive,
then the rephasing (4.69) is not allowed. Hence, one is only free to reorganize the charged current interaction
Lagrangian as the following
n,n
X g
− √ (lLi γµ e|−iθ µ
{z } Uij νLj )W + h.c..
i
(4.80)
i,j
2
n
In the above equation is again clear we can eliminate n phases, corresponding to our freedom to rephase
charged lepton fields. Subtracting this, to the total number of phases the unitary matrix in general parametrised
in (4.71), one has
n(n − 1)
Physical phases. (4.81)
2
when consider neutrinos to be Majorana. Then particularising for the case of n = 3 massive neutrinos we
are interested in, one has (
3 Mixing angles,
n=3 (4.82)
3 Physical phases.
Under these circumstances we have two more phases than before, therefore it is useful to consider a parametri-
sation where we keep the same ‘Dirac’ phase, and then add two more phases. One rather simple option is
the so-called standard parametrisation presented in the previous chapter in eq. (3.1), which is given as
UP M N S = UD P (4.83)
with
 
1 0 0
 
P = 0 eiα 0 , (4.84)
0 0 eiβ
where α and β are the ‘Majorana’ phases which arise very explicitly. It is now important to refer that we
have been considering that all neutrinos are massive. If one of the neutrinos is massless, which is not ruled
out experimentally, as is mentioned in chapter 3 (∆m221 , |∆m232 | =
6 0), then in (4.80) one would be able to
38
freely rephase one neutrino field and, accordingly, one of the ‘Majorana’ phases would vanish. In general,
for non-degenerate masses, the number of ‘Majorana’ phases is equal to the number of massive neutrinos
subtracted by one. However, if neutrino masses are degenerate, this relation is no longer valid, as we shall
explore in subsection 4.4.2.
Having in mind the experimental results presented in table 3.1 as well as the fact that θ13 is very close to
zero, which can be achieved by considering a vanishing |Ue3 | and perturbing it (see, for instance, [39], [40]),
the so-called Harrison, Perkins and Scott (HPS) matrix has been proposed in 2002 [41]
q 
2 √1 0
 3 3 
− √1 √1 √1  . (4.85)
 6 3 2 
− √16 √1
3
− 12
√
√
This matrix, alternatively designated as tribimaximal mixing, corresponds to tan θ12 = 1/ 2, θ23 = π/4 and
θ13 = 0.
4.4.2 Degenerate neutrino mass matrix

As mentioned before, the phases in the leptonic mixing matrix for degenerate neutrino masses differ from the
standard case. We can consider neutrino masses quasi-degenerate if we have mi of order & 0.1 eV, which,
despite being an approximation, can lead to interesting conclusions (see, for instance, [42]). This is one of
the plausible scenarios regarding neutrino mass scale referred in chapter 3.
Let us consider a WB where ml is diagonal. Recalling that any matrix can diagonalized by a bi-unitary
transformation, and that mν is, by construction, a symmetric matrix, therefore it is diagonalised by the same
unitary transformation on both sides, it is useful to define the following dimensionless matrix Z◦ = mν /µ in
the following way
Z◦ ≡ U◦∗ U◦† (Z◦ is symmetric unitary), (4.86)
where U◦ is the leptonic mixing matrix (PMNS) in the degenerate case. This will be our parametrization of
the degenerate mass matrix. This is clear using the following identities
dν = µ 1 = µU◦T U◦∗ U◦† U◦ = µU◦T Z◦ U◦ (4.87)
with µ being just a constant referring to the mass of all neutrinos.

Considering that the unitarity of Z◦ (symmetric) imposes 6 extra restrictions, those can be subtracted
from the 18-6=12 restrictions of a general symmetric mν matrix, allowing us to parametrize this matrix by
four phases and two angles [43]. Then, it is clear that we can rephase the charged leptonic fields as we did
in eq. (4.80), factoring out three phases, leaving only a residual phase. In short, one can parametrize mν
(therefore U◦ ) by only two angles and one single phase, concretely a ‘Majorana’ phase.
Let us particularize such mν matrix. Since Z◦ is a symmetric unitary matrix, a possible parametrisation
is either
mν = µ 1 (4.88)
39
for a trivial mixing with no phases,3 or the following [42]
   
1 0 0 cθ sθ 0 1 0 0
   
mν = µ 0 cφ sφ  sθ −cθ 0  0 cφ sφ  , (4.89)
iα
0 sφ −cφ 0 0 e 0 sφ −cφ
where cθ = cos θ and sθ = sin θ. From the above matrix one finds that the PMNS matrix for the degenerate
case, U◦ , reads
   
cos(θ/2) sin(θ/2) 10 0 0 1 0 0
   
U◦ =  sin(θ/2) − cos(θ/2) 0  0 cφ sφ  0 i 0 . (4.90)
0 0 e−iα/2 0 sφ −cφ 0 0 1
An interesting feature of this parametrisation comes from the fact that, as it is shown in [42], each
neutrino experiment constrains, independently, a single angle: double beta decay and solar neutrino data
only constrain θ, whereas atmospheric neutrino data only put a lower bound on φ.
4.4.3 Rephasing invariance

Has we have just checked, apart from discrete rephasing of (4.69), the only rephasing transformations allowed
are the charged lepton rephasing, eq. (4.68). Under these transformations the matrix elements of U transform
as
Uαi → eiθα Uαi . (4.91)
Then, since we know that physically meaningful quantities must be invariant under a rephasing of the fields,
it is not only clear that the minimal rephasing invariant terms are the products
∗
Uαi Uαj , (4.92)
but also that only functions of this terms may be measurable. The next-simplest invariants are built just by
∗ ∗
multiplying two of the minimal rephasing invariants, i.e. Uαi Uβj Uαk Uβl . One special case of those are the
‘quartets’, which will be very useful and are defined as
∗ ∗
Qαiβj = Uαi Uβj Uαj Uβi . (4.93)
As we shall see in section 5.5, considering we have three generations, the imaginary part of all ‘quartets’ of
the leptonic mixing matrix is equal up to a sign. Defining the imaginary part of the quartet Qe1µ2 as JCP ,
in the considered parametrisation (3.1) follows that
∗ ∗ 1
JCP = Im(Ue1 Uµ2 Ue2 Uµ1 )= sin(2θ12 ) sin(2θ13 ) sin(2θ23 ) cos(2θ13 ) sin δ, (4.94)
8
which shows JCP depends on the ‘Dirac’ phase, but not on any ‘Majorana’ phase. Therefore, one shall
associate any phenomenon dependent on these ‘quartets’ a possible source of ‘Dirac’ CP violation, but from
such phenomenon no conclusion whatsoever can be retrieved regarding ‘Majorana’ CP violation. This is due
3 Note that in such case, a similarity transform leaves m invariant, i.e. O T m O = µO T 1O = µ1, which means m is always
ν ν ν
kept diagonal.
40
to the fact that when we factorise ‘Majorana’ phases in this ‘quartet’, Uαi → eiθi Uαi , we always have both
the respective ‘Majorana’ phase and its conjugate, i.e.
∗ ∗
Uαi Uβj Uαj Uβi → ei(θi +θj −θj −θi ) Uαi Uβj Uαj
∗ ∗
Uβi ∗
= Uαi Uβj Uαj ∗
Uβi , (4.95)
which means the ‘quartet’ is not sensitive to ‘Majorana’ phases. As we shall see in section 4.5, this is relevant
when considering neutrino oscillations.
4.4.4 Non-Unitarity
Despite the fact that we consider, by definition, that PMNS matrix is unitary, it is important to observe that
K, which is related to U as eq. (4.62) shows, is not unitary in the context of seesaw mechanism type I. This
comes clear when we consider the V matrix (eq. (4.23)) unitarity condition V V † = 1, which is
KK † + RR† = 1, (4.96)
† †
SS + T T = 1, (4.97)
and as long as we are under the outlook of the seesaw mechanism, with mD 6= 0, we have both RR† , SS † 6= 0,
therefore KK † , T T † 6= 1 (note that 1 is unit matrix). It is, however, important to remind that not only this
deviation of K from unitarity is small, but that it is of order O (mD /MR ).
4.5 Neutrino Oscillations

The probability of finding a neutrino created in a given flavour state to be in the same state can oscillate with
time. The idea of the possible existence of neutrino oscillations was first introduced by Pontecorvo [38]. Our
main progresses on the understanding of neutrinos since then is mainly due to the study of this phenomenon,
as mentioned in chapter 3. Thus, it is crucial for us to understand it, and, if possible, try to relate it with CP
violation. Note that we will only consider vacuum oscillations, despite the fact that, as it was firstly pointed
out in [44], the presence of matter can strongly affect neutrino oscillation probabilities. As one shall see in
the following, this phenomenon only depends on Dirac phases, which means the possible ‘Majorana’ nature
4
of neutrinos cannot be evaluated based on it.
To understand this phenomenon and its results consider the vacuum neutrino oscillation experiment
schematically sketched in fig. 4.1. This figure represents the three stages of this experiment
1. A neutrino source produces, via W exchange (given by the Lagrangian in eq.(4.60)), the charged lepton
flavour i, plus an accompanying neutrino that, by definition, must be a νi ;
2. The neutrino propagates, in approximate vacuum, a distance L to a target/detector;
3. It interacts again via charged current weak interaction and produces a second charged lepton lj of
flavour j.
At that point, if the flavours i and j are different, we must conclude that during the trip from the source to
the detector, the neutrino’s flavour changed (or ‘oscillated’) from νi into a νj , and this is a consequence of
neutrino mixing.
4 In this study we are considering the quantum mechanical (QM) approach to neutrino oscillation, with neutrinos described
by the plane wave approximation. A consistent description of neutrino oscillations requires either QM approach or a quantum
field theoretic (QFT) treatment, where neutrinos are represented by propagators connecting production and detection vertices
in Feynmann diagram. A discussion and comparison between these two approaches can be found in [45].
41
Figure 4.1: Neutrino flavour change (oscillation) in vacuum. α and β are the flavour eigenstates of, re-
spectively, the created and the detected neutrino, while i and j are the mass eigenstates propagating with
different momenta due to its different masses. ‘Amp’ denotes amplitude. [46]
From above one sees that the amplitude of the process should depend on a product of three factors. The
first is the amplitude for the neutrino produced together with the charged lepton li to be a particular να .
The second factor, Prop (να ), reads the propagation from the source to the detector. The final factor is
the amplitude for the charged lepton created when the να interacts in the detector to be, in particular, an
lj . Then, we know that in the mass eigenstate basis, the particle that travels from the neutrino source to
the detector is one or another of the mass eigenstates να . Thus, we must consider a coherent sum over the
contributions of all the να , as shown in the lower part of fig. 4.1.
Let us consider neutrino mixing is given by eq. (4.61) and the plane wave approximation5 of the mass
eigenstates, given by
|να (t, x)i = ei(pα (x−x0 )−Eα (t−t0 )) |να i, (4.98)
where |να i ≡ |να (t = 0)i. Thus, the resulting time evolution of the flavour neutrino state under the stages
described before is
X
∗ i(pα (x−x0 )−Eα (t−t0 ))
|νi (t, x)i = Uiα e |να i
α
!
X X
∗ i(pα (x−x0 )−Eα (t−t0 ))
= Uiα e Ujα |νj i (4.99)
j=e,µ,τ α
Since the mixing matrix U is different from unity, the state |νi (t, x)i, which has pure flavour i at the initial
time t = 0, evolves in time into a superposition of different flavours. The quantity in parenthesis in eq. (4.99)
5 to keep things simple. An interesting discussion on different approaches, such as the consideration of a wave-packet, can be
found in [47] and references therein.
42
is the amplitude of νi → νj transitions at the time t after νi production. The probability P of νi → νj
transitions at the distance x = L and time t = T of neutrino detection is the following
2
X
∗ i(pα L−Eα T )
Pνi →νj (T, L) = |hνj |νi (T, L)i|2 = Uiα e Ujα
α
X
∗ ∗ i((pα −pβ )L−(Eα −Eβ )T )
= Uiα Ujα Uiβ Ujβ e . (4.100)
α,β
Now one only has to define phase of the above equation, φαβ , in order to obtain the amplitude of this process.
= L/c and pν ∼
Considering the ultra-relativistic nature of neutrinos, T ∼ = Eν , it follows
φαβ ∼
= (pα − pβ )L − (Eα − Eβ )L
p2α − p2β Eα2 − Eβ2
∼
= L− L
pα + pβ pα + pβ
L ∼ L
= (m2α − m2β ) = (m2α − m2β )
pα + pβ 2E
Under this assumption, one can rewrite eq. (4.100) as

!
X ∆m2αβ L
∗ ∗
Pνi →νj (L, E) = Uiα Ujα Uiβ Ujβ exp −i (4.101)
2E
α,β
3
X X
2 ∗ ∗
∆m2αβ L
= |Uiα Ujα | + 2 Re Uiα Ujα Uiβ Ujβ cos
i=1
2E
α<β
X ∆m2αβ L
∗ ∗
+2 Im Uiα Ujα Uiβ Ujβ sin .
2E
α<β
where ∆m2αβ = m2α − m2β . This expression is valid for any number of flavours and equal number of mass
eigenstates. One can also easily see that neutrino oscillation in vacuum from one flavour i into a different
flavour j implies nonzero mass splittings ∆m2αβ , and these mass splittings are important results we already
introduced in chapter 3. Evidently, the probability of finding the original flavour is given by
X
Pνi →νj = 1 − Pνi →νj . (4.102)
i6=j
After determining the amplitude and probabilities of the process, let us briefly analyse CP violation in
this process. By considering the amplitude of the CP-mirror image process να → νβ , one concludes that
∗ ∗
there will be CP violation when Im(Uαi Uβi Uαj Uβj ) 6= 0. As it is clear in eq. (4.95), this is a Dirac quartet,
which means it only contains Dirac phase(s). Thus we cannot decide, based on oscillations, whether neutrinos
present a ‘Dirac’ or ‘Majorana’ character. Moreover, based in the previous calculation of the imaginary part
of the quartet, J, we see that that there is no CP (or T) violation for three generation neutrino mixing in
vacuum if δ = 0, π and
i) one (or more) of the mixing angles is zero;
ii) two or more of the masses are degenerate.
Moreover, note that it is impossible to observe CP violation in the disappearance channels (i = j), since
νi → νi is related to νi → νi by CPT.
43
Zdesenko, Yu. G., O. A. Pnnkratenko, and V. I. Tretyak, 2001, J. Phys. G: Nucl. Part. Phys. 27,
2129.
Zdesenko, Yu. G., et al., 2005, Astropart. Phys. 23, 249.
Zuber, K., 2001, Phys. Lett. B 519, 1.
4.6 Neutrinoless double

Zuber, K., 2005, summary beta-decay
of the Workshop on: Nuclear matrix elements for neutrinoless double
beta decay (Durham UK), eprint nucl-ex/0511009.
Double β-decay is a nuclear transition in which an initial nucleus (Z, A), with proton number Z and total
nucleon number A, decays to (Z + 2, A) by emitting two electrons and possibly other light neutral particles.
Figures
This transition is possible and potentially observable because nuclei with even Z and N are more bound than
the odd nuclei, which have the same A and one proton more, Z + 1 (see fig. 4.2).
Z+1
0+
Z
0+
ββ
2+
0+
Z+2
Figure 4.2: Double-β transition. It is energetically more favourable a even transition (figure extracted
from [48]). 74
Considering only the particles we know about exist, there are two modes of the double-β decay. One
has already been observed and it is the straightforward case, predicted by the SM, with lepton number L
conservation and given by n + n → p + p + e− + e− + νe + νe . It has been reported to have a typical half-life
of ∼ 1019−20 years in nuclei with Q-values (the kinetic energy available to leptons) above 2 MeV. The second
mode is the neutrinoless double-β decay (0νββ) where only 2e− are emitted and nothing else. Such process
has ∆L = 2 and is clearly forbidden in the SM. Hence, its observation would be a signal of a ‘new physics’
and particularly a proof of the Majorana nature of neutrinos.
It is important to keep in mind that the detection of this process is an hard task. Apart from all
uncertainties, it is a second order of weak interactions (as the 2νββ), ∼ G4F , thus inherently slow. Moreover,
comparing to the ordinary double β decay, whose half-life is around ∼ 1019 years, the lower limit on 0νββ is
around ∼ 1025 , which is significantly higher and increases the difficulty of the task. One of the reasons for
that is its dependence on neutrino mass. To make this statement clear, let us analyse the process, and focus
in the leptonic part of this process (see fig 4.3). For a deeper analysis regarding the different aspects of this
phenomenon, namely its nuclear physics, see [49], [50]. The rate of the 0νββ decay is [48]
X Z d3 p1 d3 p2
0ν −1
[T1/2 ] = |M0ν |2 δ(Ee1 + Ee2 − Qββ ) (4.103)
spins
2π 3 2π 3
0ν
where T1/2 is the process half-life, M0ν is the amplitude and Qββ is the Q-value of the decay. The effective
√
low-energy interaction Hamiltonian is Hβ (x) = GF / 2 e(x)γµ (1 − γ5 )νe (x)JLµ (x) + h.c., where JLµ is the
charge-changing hadronic current and we neglected the momentum dependence in the W propagators, having
in mind that the Q-value is of the order of a few MeV. Considering that the decay is only mediated by the
exchange of three light neutrinos, the amplitude contains a lepton part (to be contracted with the hadron
44
part) of the form
X
e(x)γµ (1 − γ5 )Uej χj (x)e(y)γν (1 − γ5 )Uej χj (y) =
j
X
− e(x)γµ (1 − γ5 )Uej χj (x)χcj (y)γν (1 + γ5 )Uej ec (y) (4.104)
j
where in the last equality we charge conjugate both fields of the second term, used eq. (2.12) to transpose
γν and considered the anticommutative properties of the fermionic fields, which flipped the sign of the term.
Within our convention χcj = χj , eq. (2.30), we see that the contraction χj χcj = χj χj is the usual fermion
propagator. Hence, the leptonic part above becomes
Z X 4
i d q −iq(x−y) /q + mj c 2
− 4
e e(x)γµ (1 − γ5 ) 2 2 γν (1 + γ5 )e (y)Uej (4.105)
4 j
(2π) q − m j
with q as the 4-momentum transfer. Having in mind that, on the one hand, /q vanishes using the γ matrices
relations, and on the other hand that the mk in the denominator can be neglected for light neutrinos, the
amplitude is proportional to
X
2
|mee | ≡ Uei mi (4.106)
= | cos2 θ13 (|m1 |cos2 θ12 + |m2 |e2iα sin2 θ12 ) + sin2 θ13 |m3 |e2iβ |. (4.107)
where we considered a slight modification of the parametrisation in (4.78) (δ = φ13 − φ12 − φ23 ) to obtain
the last equality. To complete the calculation, one would have to multiply the lepton part of the amplitude
by the nuclear matrix element and integrate. It is out of the scope of this study but the reader can find this
treatment in the reviews already referred.
u
d u
W −
e e−
W
ν mee ∆
e− W
W
e−
d u
u
Figure 4.3: In the left-hand side is represented 0νββ decay leading order involving the ‘Majorana’ neutrino
ν defined as ν = νL + νLc in our convention. The diagram involving heavy neutrinos N is analogous. In the
right-hand side, the same decay considering the seesaw Type II, where ∆−− mediates the process.
From the factorisation of the ‘Majorana’ phases in (4.107), it comes out clear that by observing this
45
process one is not only able to decide if neutrinos are Majorana, but we can also be able to determine one
and only one combination of ‘Majorana’ phases, depending on the experimental and theoretical uncertainties,
as mentioned in chapter 3. Considering its dependence on the neutrino mass, which can be experimentally
challenging, there are also good expectations to find better constraints on this measurement, having in mind
that neutrino mixing angles θij and neutrino mass-squared differences ∆m2ij can be accurately determined
from oscillation experiments. It is also clear that cancellations among terms may occur due to those phases,
thus, in principle, they can conspire so mee is zero, and in that circumstance no 0νββ occurs.
One may wonder whether from the seesaw mechanism higher contributions for 0νββ may arise. Consid-
ering the term in (4.105) within all the approximations introduced, one checks that the effective coupling of
this term (G0ν ) is

1
G0ν ' G2F mν . (4.108)
q2
The expression in eq. (4.105) can be equivalently obtained when we substitute light Majorana neutrinos by
heavy Majorana neutrinos, in the context of seesaw mechanism Type I. In such case, one would consider
that the momentum transferred is negligible when compared to heavy neutrino mass, q Mj . Then one
can obtain the contribution of heavy neutrinos to the amplitude of 0νββ to be proportional to
G2F
G0N ' . (4.109)
M
Having in mind that the typical virtual-neutrino momentum is hq 2 i = 200 MeV, and taking m ∼ 1 eV, then
eq.(4.108) and eq. (4.109) would imply M & 107 GeV. Therefore, the effective coupling is likely to be weaker
than light neutrinos, having in mind that our mass scale for heavy neutrinos can be around 1014 or more (for
a deeper analysis see [51]).
Furthermore, when one considers the Type II seesaw, which is represented in fig. 4.3, in the same
approximation, one can estimate that the effective coupling is [52]
h∆ee δ
G∆ ' G2F . (4.110)
m∆
where the factor h∆ee refers to the coupling between the triplet and the electrons, δ from the coupling of the
triplet to the two W and 1/m2∆ is its propagator for m2∆ q 2 . This allows us again to see that we would
need m∆ < 107 GeV for 0νββ process to be dominated by doubly charged Higgs triplet.
4.7 Baryon Asymmetry of the Universe and Sakharov’s Condi-

tions
The next chapter of this study is dedicated to CP violation in leptonic sector. Therefore, it is important to
explain the reason for our focusing in such specificity of our theory. Indeed, all structures we see in universe
(stars, galaxies, etc.), consist of matter (namely baryons and electrons) whereas we find no antimatter in
appreciable quantity. Since various considerations suggest that the Universe have started out very symmet-
rically between matter and antimatter, then the observed Baryon Asymmetry of the Universe (BAU) should
have been generated dynamically. A scenario that creates this type of asymmetry is named baryogenesis. As
we shall see in the following, one of the ingredients for this phenomenon to take place is CP violation.
There are two distinct evidences that establish the domination of matter over antimatter. The first
46
arises from the analysis of the power spectrum of the cosmic microwave background. In particular, today’s
baryon-to-photon ratio yields [53]

CM B nB − nB̄ −10
ηB ≡ = (6.21 ± 0.16) × 10 , (4.111)
nγ 0
where nB , nB̄ and nγ are the number densities of, respectively, baryons, antibaryons and photons, and
the subscript 0 refers to the present time. The second evidence emerges from the concordance of the light
elements and the big bang nucleosynthesis, which gives [15]
BBN
ηB = (3.4 − 6.9) × 10−10 . (4.112)
Despite their different origin, both deduced values for ηB are in good agreement with one another. Having
this asymmetry in mind (well, not exactly in these terms, since this statement was made more than fifty years
ago, in 1967), Sakharov described the conditions for baryogenesis, stating in [11] the following: ‘According
to our hypothesis, the occurrence of C asymmetry is the consequence of violation of CP invariance in the
nonstationary expansion of the hot Universe during the superdense stage, as manifest in the difference
between the partial probabilities of the charge-conjugate reactions’. Today, this short extract is usually
dubbed as three necessary Sakharov’s conditions for baryon asymmetry generation from the initial charge
symmetric state in the hot Universe: at least one Baryon number (B) violating process, C and CP violation,
and breaking of thermal equilibrium. To understand these conditions, suppose that we know the state of the
Universe (the density matrix ρ0 at some moment t0 ) and know the theory of particle interactions and its
respective Hamiltonian H. Then, to determine the baryon asymmetry, one should find
B(t) = TrB̂ρ(t), (4.113)
where B̂ is the baryon number operator defined in (4.115) and the density matrix ρ satisfies the Liouville
equation
∂ρ(t)
i = [H, ρ] (4.114)
∂t
with the initial condition ρ(t0 ) = ρ0 . Then the three conditions are defined as follows.
1. B violation: There must exist a process of the form X → Y + B with net baryon number positive,
otherwise the state, initially symmetric with respect to the baryon number, cannot evolve into the state with
the nonzero baryon number that we observe today. Formally, if [B̂, H] = 0, then TrB̂ρ(t) = 0 provided
TrB̂ρ0 = 0.
2. C and CP violation: Another requirement is that the theory should distinguish between particles
and antiparticles. The baryon number operator
Z
1X
B̂ = d3 xq † (t, ~x)q(t, ~x) (4.115)
3 q
47
is C-odd and CP-odd, which can be shown considering the P, C and CP transformations previously derived
P B̂P −1 = B̂, C B̂C −1 = −B̂, CP B̂CP −1 = −B̂, T B̂T −1 = B̂. (4.116)
Thus, a non-zero expectation value hB̂i requires that the Hamiltonian violates C and CP. More intuitively,
C symmetry would guarantee that Γ(i → f ) = (i† → f † ), while CP symmetry would guarantee that
Γ(i → f ) = (i → f ) (x† has opposite charges and same chiralities as x, x has opposite charges and opposite
chiralities). Formally, if [C, H] = 0 or [CP, H] = 0, then TrB̂ρ(t) = 0, provided the initial state is C or CP
symmetric, [C, ρ0 ] or [CP, ρ0 ].
3. Departure from thermal equilibrium: As represented in fig. 4.4, this third requirement is
equivalent to the existence of arrow of time.
Y Y
X X
B B
time time
Figure 4.4: Consider a B violating process X ←→ Y + B. On the left, this process is in thermal equilibrium,
i.e. can happen in both directions. When the system breaks this equilibrium, then one of the reactions - in
the right-hand figure the X ←→ Y + B - is suppressed.
Suppose the thermal equilibrium state is characterized by only one parameter - the temperature, as usual
in statistical mechanics. The corresponding density matrix is ρeq = Z −1 exp −H/T , where Z is the partition
function. If some charge, specifically baryon number, is not conserved, then in thermal equilibrium, it is
equal to zero, independently of the conservation or nonconservation of C and CP. This is a consequence of
the CPT theorem, since it states that the CPT transformation does not change the Hamiltonian, but, as one
checks in (4.116) the baryon number is reversed by CPT , leading to TrB̂ρeq = 0.
In short, by the definition of the arrow of time and a B number violating process, considering CP vio-
lation (or equivalently T violation) one can finally fulfil the Sakharov conditions to have baryogenesis. It is
important to stress that all the conditions must occur simultaneously.
All these three conditions are known to be satisfied. It is certain that deviations from thermal equilibrium
appear in the expanding Universe, and the baryon number is nonconserved not only in hypothetical models,
but in the Standard Model itself. It is also a fact that CP is violated in the quark sector, although not enough
to generate the quantity of Baryon Asymmetry of the Universe (BAU) we observe [12]. For this reason it is
important to address our efforts to other sources of CP violation, namely in the leptonic sector as we do in
this work in chapter 5, and associate them with different physical processes and symmetries.
48
Chapter 5
CP violation in the leptonic sector
The only source of flavour mixing and CP violation in the Standard Model is the CKM matrix. However, as
we studied in the previous chapter of the study, neutrino oscillations are a experimental evidence of leptonic
flavour violation. These oscillations are consequence of the existence of a non-trivial flavour mixing matrix,
the PMNS matrix, and the fact that there seems to be no constraint on whether it is real or imaginary,
instigates the study and the possibility of a CP non-invariance of the Lagrangian. In this chapter we shall
proceed with our study on neutrino physics by analysing CP violation in the leptonic sector.
5.1 General CP conditions

Considering transformations in chapter 4.2, in the case CP-invariance holds in the mass Lagrangian, eq.
(4.2), the following conditions must be satisfied
mL = −U 0† m∗L U 0∗ , (5.1)
∗ 0T 0
MR = −W MR W , (5.2)
mD ∗ = U 0† mD W 0 , (5.3)
m∗l 0†
= U ml V . 0
(5.4)
This may lead to complicated conditions, but one will find some convenient ways to evaluate CP violation in
this sector. Furthermore, when one considers only low energy effects, whose mass term is in eq. (4.66), then
the condition
m∗ν = −U 0T mν U 0 , (5.5)
must hold, in order to have CP invariance.
5.2 CP violating phases

In this section one should analyse the number of CP violating phases in our theory. Considering the charged
current Lagrangian in eq. (4.59), and having in mind the CP transformations introduced in chapter 2.2 and
4.2, it is clear that the condition of CP invariance in the case of mixing and massive Majorana neutrinos
49
reads
Ulj = ±Ulj∗ (5.6)
where the relative sign depends on the CP parity of the Majorana neutrino, an issue we will address later in
this study, in chapter 5.3. Under these conditions, it is clear that phases in eq. (3.1) must be such that
δ = 0 mod π (5.7)
π
α, β = 0 mod (5.8)
2
where α can differ from β. At this point we should stress that the number of CP violating phases arising at
low energies is three, and correspond to the number of CP violating phases in the PMNS matrix. This is not,
as we already pointed out, the case corresponding to degenerate neutrino masses. In such case the number of
CP violating phases is only one, a ‘Majorana’ phase, which must obey to the condition introduced in eq. (5.8).
Regarding a further analysis regarding the extra phases arising from the seesaw mechanism, one should
choose a convenient weak-basis (WB), inasmuch as the conditions for CP conservation, defined in chapter
5.1, are rather complicated. In this case, a WB where both ml and MR are real, diagonal and positive is a
perfectly valid WB of the full mass Lagrangian of eq. (4.2) and is the right one for this study. Accordingly,
considering this WB condition, W 0 must be of the following form
W 0 = diag (exp(iα1 ), ..., exp(iαn )) , (5.9)
where we assumed that the eigenvalues of mL and MR are nonzero and nondegenerate, n denotes the number
of right-handed neutrinos generations, and αi must satisfy
π
αi = (2pi + 1) , (5.10)
2
where pi are integer numbers. Since we are considering a weak-basis where ml is also real diagonal, we can
derive the following condition for U 0 , just by multiplying (5.4) by its Hermitian conjugate
U 0 = diag (exp(iβ1 ), exp(iβ2 ), exp(iβ3 )) , (5.11)
with βj (j = 1, 2, 3) as arbitrary phases related to each flavour generation. This diagonal W 0 and U 0
form, derived from this particular WB, is very useful in order to define the following condition regarding
mD phases. Therefore, from (5.3), one can derive −arg(mDij ) = −βi + arg(mDij ) + αj , which means CP
invariance restricts mD matrix to
1
arg(mDij ) = (βi − αj ). (5.12)
2
In the following analysis we will understand how many independent phases are relevant for the Dirac
mass term mD . Therefore, if one notes that αi are fixed by (5.10) up to discrete ambiguities (pi is an integer
number) then one realizes that the CP invariance constrains the matrix mD to have only 3 free phases βi .
Since mD is an 3 × n arbitrary matrix (3 lepton flavours, n right-handed neutrinos), with 3n independent
phases, thus there will be 3n − 3 independent CP restrictions. This is also the number of CP violating phases
that appear in general in this model. In the WB we are analysing, these are also the 3(n − 1) independent
phases which we are unable to remove from mD .
50
Consider now the most common scenario, with the introduction of 3 neutrinos based in right-handed, we
have 6 CP violating phases, and 9 real parameters (arising from the 3 × 3 matrix mD ). Following the fact
that, in this WB, the matrices MR and ml are real diagonal, one has, in the overall, fifteen real parameters
and six CP violating phases. It will be interesting, as we shall see, that this counting of degrees of freedom
agrees with the one made in the physical basis, in terms of the neutrino and charged lepton masses, together
with mixing angles and CP violating phases entering in the leptonic mixing matrix. This leptonic mixing
matrix is present in many low energy processes which means some of these CP violating phases may be
detectable.
It is also useful to analyse the number of CP violating phases when mL is different from zero. From a
quick evaluation, and taking in consideration that a general complex symmetric matrix mL has six phases,
one would have 6 + 6 = 12 CP violating phases for a framework with three light plus three heavy neutrinos.
This is indeed the case. To obtain this result let us consider the same WB as before (mL and MR diagonal
real). Then from eqs. (5.1) and (5.2) we again derive (5.9) and (5.11), but this time both α and β are
restricted by
π π
αi = (2pi + 1) , βj = (2qj + 1) , (5.13)
2 2
where pi , qj are integers. Then eqs. (5.3) and (5.4) constrain mD and ml m†l respectively to
1
arg(mDij ) = (βi − αj ), (5.14)
2
1
arg[(ml m†l )ij ] = (αi − αj ). (5.15)
2
From the above equation one notes that phases of both mD and ml m†l are fixed up to discrete ambiguities,
i.e., they can only be 0, π2 , π, .... Having in mind that mD is 3 × n arbitrary complex matrix whereas ml m†l
is an arbitrary 3 × 3 Hermitian matrix, then one has 3n + 3(3 − 1)/2 independent CP restrictions. In other
words, in the WB we are considering, we have 3(2n + 3 − 1)/2 independent phases which we are unable to
remove.
Particularizing to the the framework with n = 3 heavy neutrinos, we have, as mentioned before, 12 CP
violating phases. We summarize the results in table 5.1, where we also generalize for the case of m left-handed
spinors and n right-handed ones.
Total CP violating phases

Scheme m LH and n RH 3 LH and 3 RH
Low Energy n(n − 1)/2 3

High Energy w/ mL = 0 m(n − 1) 6
General case mn + m(m − 1)/2 12
Table 5.1: Number of CP violating phases in each framework: in the low energy or ‘decoupling’ limit, and in
the ‘non-decoupling’ limit with vanishing and non vanishing mL . For the three frameworks we particularize
for the case we are mostly interested in: 3 light neutrinos and 3 heavy neutrinos.
Finally, it is also useful to note that this analysis has been general, without any assumptions regarding
the size of the various neutrino masses.
51
5.3 CP Parities
In a WB with diagonal charged lepton masses, in a low energy scheme, and taking into account the analysis
of chapter 4.4 about PMNS matrix U , one would naively require it to be a real matrix in order for CP to hold.
In fact, as we can check in (5.8), this is not correct due to the existence of massive Majorana neutrinos with
opposite CP eigenvalues, also called CP parities [54]. In order to illustrate this point consider the Majorana
mass term for active neutrinos previously introduced in eq. (4.64). Taking into account only one neutrino,
and applying a CP transformation to this mass term, we have
1 1
(CP) −νLc dν νL − νL dν νLc (CP)† = exp(−2iβν )νL dν νLc + exp(2iβν )νLc dν νL (5.16)
2 2
just by having in mind the results in (2.49), (2.51), (2.53) and (2.54). Therefore, when CP is conserved, one
1
must have exp(2iβν ) = −1 = exp(−2iβν ) ⇒ exp(iβν ) = ±i. Consequently, one arrives to the conclusion
that the CP parity of a left-handed neutrino mass eigenstate must be either i or −i, i.e.
(CP)νiL (CP)† = ±iCνiL

∗
, (5.17)
∗
where one applied the properties of the matrix C stated in appendix A to obtain γ 0 νiL
c
= γ 0 Cγ 0T νiL =
∗
−CνiL . Hence one can define the relative sign of ηiCP as the CP parity of the neutrino, with ηiCP being
defined as
ηiCP = ±i. (5.18)
Since we are considering a real matrix mν , it can be diagonalized by an orthogonal real transformation
O of the form
OT mν O = diag(m1 , m2 , m3 ) (5.19)
where mi are real, but can be either positive or negative. As one shall check now, regarding the CP parities
of neutrinos two cases are possible:
i) all mi have equal sign;
ii) one mi has a sign different from the other two.
Let us start with an example of case 2., for instance with m2 negative and both m1 and m3 positive.
Thus, the orthogonal transformation in eq. (5.19) reads
OT mν O = diag(|m1 |, −|m2 |, |m3 |). (5.20)
Then, in order to have all mass terms positive and diagonal one should not consider the real orthogonal
matrix O, but a complex matrix U to diagonalize the mass matrix, i.e.
U T mν U = diag(|m1 |, |m2 |, |m3 |), (5.21)
given by U = OP , with P = diag(1, i, 1). In the parametrisation we are considering, eq. (3.1), this can be
achieved, for instance, with α = π/2 and β = 0. Since we now have the diagonalisation of mν with all terms
1 This condition is also implicit in (5.5).
52
real positive, one can abandon the WB with the charged current interaction explicitly written as
  
O11 iO12 O13 ν1
g   
LW = − √ (e, µ, τ )L γ µ O21iO22 O23  ν2  Wµ+ −
2
O31iO32 O33 ν3 L
   (5.22)
O11 O21 O31 e
g µ  
− √ (ν1 , ν2 , ν3 )L γ −iO12 −iO22 −iO32  µ Wµ−
2
O13 O23 O33 τ L
As mentioned before, if one assumes CP invariance, then both the Majorana neutrinos and the charged
lepton should be CP eigenstates. Hence, the CP transformation of the mass eigenstates yields
(CP)νiL (CP)† = ηiCP CνiL

∗
, (CP)ljL (CP)† = ηjCP
l ∗
CliL . (5.23)
Now one only has to define ηiCP for one neutrino and use the structure of the charged weak interactions
(5.22) to fix the other relative CP Parities. Supposing that all Oij are nonvanishing and, for instance, take
the following as initial choice
(CP)ν1L (CP)† = +iCνiL

∗
, i.e. η1CP = +i (5.24)
then one obtains for (CP)eγµ O11 ν1L W +µ (CP)† = iηCP

e∗
ν1L γ µ O11
T
eWµ+ from which one can take ηCP
e
= +i.
µ τ
Similarly, applying CP transformations in the terms with O21 and O31 force ηCP = ηCP = i. In the
sequence, one can also take the couplings iOj2 to constrain ν2L to have η2CP = −i and the couplings Oj3 to
have η3CP = +i.
In short, one has η1CP = −η2CP = η3CP = i, which means we are forced to assign a CP parity different
from that of ν1 and ν3 for CP to hold. It is important to note that only relative CP parities are meaningful,
since it depends on an a priori arbitrary choice. Moreover it is clear that the case with two negative masses
in eq. (5.19) is analogous to the presented case and we should also have one neutrino with a different rela-
tive CP parity. Finally, it comes obvious that, for the case with all negative masses in eq. (5.19), one has
η1CP = η2CP = η3CP and a global phase we can rotate away, therefore the masses can be all set positive.
One should now return to the definition of a Majorana spinor introduced in (2.28). As mentioned in the
footnote, the most general Majorana spinor field for neutrinos is given by eq. (2.29), i.e.
χν = νL + exp(iξ)νLc . (5.25)
As things are stated, one can conclude working with the above definition for a Majorana spinor is equivalent
to include a phase ξ in the mass term. Accordingly, this mass term reads
c )(ψ + exp(iξ)ψ c )
LM = mL χν χν = mL (ψL + exp(−iξ)ψL L L
= mL exp(−iξ)νLT C −1 νL + h.c., (5.26)
where one can define mν = mL exp(iξ). Then, one obtains the conjugate spinor defined as
χcν = νLc + exp(−iξ)νL = exp(−iξ)χν . (5.27)
53
Thus, one can define the self-conjugate relation, stated before in eq. (2.30), for a general Majorana spinor
χ, which is
χ = exp(iξ)χc . (5.28)
And it is clear, as shown for the general case in (2.46), that above equations denote (χcν )c = χν .
We can proceed with the study of CP parities, particularly in their relation with the Majorana spinor,
whose definition we have just introduced. One can introduce the definition of the Majorana fields present in
the mass term (4.64)
X
c
χn = (Ona νLa + ηn Ona νLa ). (5.29)
a
Just by building the mass term it follows that to a real negative mi corresponds a relative negative sign
c
between νLa and νLa (which is expressed by ηn = −1), while the opposite stands for a real positive mi . In
short, one has the two cases previously introduced stated as
i) ηn = +1, Una = Ona ,
ii) ηn = −1, Una = iOna .
and the C transformation is naturally dependent on it, i.e.
Cχn C † = ηn χn . (5.30)
As as consequence, it is also clear that, apart from a factor, the CP transformation of χn is also dependent on
ηn . However, this different approach does not modify any of the conclusions regarding CP-parities of neutrinos
as the nonsignificance of the CP-even or CP-odd in a single Majorana neutrino or when all neutrinos have
the same parity (all mi with the same sign).
From the above we can conclude that the relative CP-parity of two neutrinos χi and χj is expressed as
the phase difference between the diagonal terms of this matrix, i.e. 2(Pii − Pjj ) = ±1, when considering the
parametrisation of P in eq.(3.1). After the characterisation of CP-parity, one can now complete the definition
of CP invariance of the leptonic charged current weak interaction, previously defined in (5.6), which yields
1
Ulj∗ = Ulj ρj , ρj = ηCP (χj ) = ±1, (5.31)
i
where ηCP (χj ) = iρj = ±i is the CP parity of the Majorana neutrino χj .
Finally, it is important to refer that [42] shows the impossibility of rotating away the leptonic mixing
matrix in the case of neutrinos with different CP parities, even in the limit of exact degeneracy and in the CP
conserving case. For this purpose, consider the parametrisation of the mass matrix for degenerate neutrino
masses (4.89). If one calculates the trace and the determinant of Z◦ , one obtains, respectively, Tr(Z◦ ) = eiα
and det(Z◦ ) = −eiα . Then, when one tries to rotate it away, which can be done by performing an orthogonal
transformation
Z◦ → O T Z◦ O (5.32)
54
as we did before in the non-degenerate case, obviously the trace and the determinant remain invariant.2 Thus,
one can only take the diagonalisation of Z◦ to be (1, −1, 1) or (1, −1, −1) (and trivial permutations), which
correspond to α = 0 and α = π. Since the case (1, 1, 1) has trace three, it obviously cannot be obtained by
parametrisation of (4.89). This is therefore equivalent to state the impossibility to rotate away the leptonic
mixing matrix.
Furthermore, we can also conclude that we parametrise the degenerate neutrino mass matrix as (4.88) if
the three neutrinos have the same CP parity, or as (4.89) if one neutrino has an opposite CP parity.
5.4 CP violating phases in seesaw mechanism type I

In the scope of seesaw mechanism type I, where we consider the leptonic Lagrangian raising from spontaneous
symmetry breaking defined in (4.2) with mL = 0, one can derive more relations which can be useful later.
Thus, considering both the zero entry of (4.20) and the top left identity taken from M∗ V = V ∗ D (eq. (4.22)),
this exact relation can be derived
R = mD T ∗ D−1 . (5.33)
From (4.27), and keeping in mind that we are working in a WB where the right-handed neutrino mass
matrix MR is diagonal, one can easily conclude that T = 1 up to corrections of order O(mD 2 /MR 2 ). To
understand this one must recall that, since R is O(mD /MR ), the first two terms highest order in eq. (4.27)
must be O(mD 2 /MR ). To match this, the first correction of the third term must have the same order, namely

O mD 2 /MR . Therefore, we must have T = 1 − O(mD 2 /MR 2 ).
This consideration can lead us to the following excellent approximation
R = mD D−1 . (5.34)
From this equation it is now clear how the 3(n − 1) physical phases of the Dirac mass matrix mD come up in
the low energy interactions, namely in eq. (4.59). In our framework with three right-handed neutrinos νR we
can summarize the 18 parameters needed to define this matrix: 6 complex phases, 12 moduli - 3 eigenvalues
of MR and 9 moduli of our 3 × 3 mD matrix.
Let us make a detour to recall the useful property that states that for any complex matrix A of rank Na ,
we can find a unitary matrix which transforms it into a triangular form [55]
 
a1 0 0
 
A = U A4 = U a21 a2 0 , (5.35)
a31 a32 a3
where the diagonal elements are real, and the off-diagonal elements are complex in general. In this case, one
should parametrize the 3 × 3 Dirac mass matrix by [56]
mD = U Y4 (5.36)
with U as an unitary matrix, and Y4 as the terms arising from the Yukawa couplings that have the following
2 Just consider the property of invariance of the trace under cyclic permutations and the property det O T Z O =
◦
det OT det Z◦ det O where O is an orthogonal matrix with det O = 1.
55
triangular form  
y1 0 0
 
Y4 = |y21 | exp(iφ21 ) y2 0 , (5.37)
|y31 | exp(iφ31 ) |y32 | exp(iφ32 ) y3
where yi is real. Since we are considering, as mentioned above, six CP violating phases, we only have
to accommodate three phases in the unitary matrix U . In a WB this corresponds to a simultaneous phase
transformation in left-handed neutrino fields νL , and left- and right-handed charged lepton fields, respectively
lL and lR .
Furthermore, in order to isolate different phases, regarding their influence and origin in the model we are
taking into account, one can define another matrix Pβ = diag(1, exp(iβ1 ), exp(iβ2 )) and rewrite Y4 as
Y4 = Pβ† Ŷ4 Pβ . (5.38)
With β1 = φ21 and β2 = φ31 one can get

 
y1 0 0
 
Ŷ4 = |y21 | y2 0 , (5.39)
|y31 | |y32 | exp(iσ) y3
where σ = φ21 − φ31 + φ32 . Now considering the U matrix which contains three phases, we can rewrite it as
U = Ûρ Pγ , with Ûρ only having one phase ρ and Pγ = diag(1, exp(iγ1 ), exp(iγ2 )) as we usually do when we
parametrize the PMNS matrix. For instance, we can use the ‘standard’ parametrisation, as used in in the
case of the CKM matrix or in (3.1).
Additionally, having in mind eqs. (5.36), (5.38) and αi = γi − βi , the mass matrix mD can be written as
mD = Ûρ Pα Ŷ4 Pβ . (5.40)
With this transformation we have found a very elegant way of presenting the six mD matrix phases: one ρ
phase in Ûρ , two αi phases in Pα , one σ phase in Ŷ4 and two βi phases in Pβ . Moreover, when we consider
the WB where ml and MR are diagonal and real, these six phases are the only physical phases, which means
CP violation in this model is fully characterized by them.
Recalling now the excellent approximation (5.34) we got for the charged current couplings of charged
leptons to the heavy neutrinos Nj we obtain
R = Ûρ Pα Ŷ4 D−1 Pβ . (5.41)
From this result it is also clear that the phases β1 and β2 can be removed from R just by a phase redefinition
of the heavy Majorana neutrinos, N2 and N3 .
5.5 Leptonic unitary triangles

In this section we shall present the very useful concept of unitary triangles. Building these triangles, which
result from the unitarity condition of the mixing matrix, is an established procedure in the quark sector,
and we will analogously define them in the leptonic sector. As one shall see in the following, the area of
the unitarity triangle, S, is related to the Jarlskog invariant, JCP , which is a parametrisation independent
56
measure of CP-violation, as
1
S= JCP . (5.42)
2
So, to establish CP violation it is sufficient to show that the longest side of the triangle is smaller than the
sum of the other two.
Considering the unitarity of U , one can define triangles by multiplying two rows or two columns of U .
∗ ∗ ∗
Considering the former option, one has, for instance, Ue1 Uµ1 + Ue2 Uµ2 + Ue3 Uµ3 = 0. These are usually called
‘Dirac triangles’ and under rephasing (Uαi → eiθα Uαi , eq. (4.91)), these triangles rotate in the complex plane,
∗
so their orientation has no physical meaning. In other words, Uαi Uβi is not a rephasing invariant. As one
∗ ∗
will check for the ‘Majorana triangles’, they share a common area A = 1/2|ImUαi Uβj Uαj Uβi |. Considering
the 3 × 3 mixing matrix with one Dirac CP violating phase (δ in parametrization (3.1)), their area would
vanish in the case U matrix elements are real, so that the imaginary part of the ‘quartets’, which are defined
in (4.93), is zero and we have no CP violation. However, when one consider U to be ‘Majorana’, J = 0
only implies that the Dirac phase vanishes, and no considerations can be made regarding ‘Majorana’ phases,
which can still violate CP. Therefore, these triangles can provide a necessary but not sufficient conditions for
CP conservation and are not enough to completely describe CP violation.
Considering now the multiplication of two columns of U , one can also define three of the so-called ‘Ma-
jorana triangles’ [57]
∗ ∗
Ue1 Ue2 + Uµ1 Uµ2 + Uτ 1 Uτ∗2 = 0
∗ ∗
Ue1 Ue3 + Uµ1 Uµ3 + Uτ 1 Uτ∗3 = 0 (5.43)
∗ ∗
Ue2 Ue3 + Uµ2 Uµ3 + Uτ 2 Uτ∗3 = 0
In the complex plane, each term from the sums in (5.43) determines a vector. Since the sum is zero, it means
that the start and the end point coincides, corresponding to three unitary triangles. Contrary to the ‘Dirac
triangles’, a phase change that rotates them in the complex plane, correspondent to Uαi → eiθi Uαi , does not
leave the mass term invariant. Thus, their orientation is physically meaningful.
∗
If one multiplies, for instance, the first triangle by Uµ1 Uµ2 , the imaginary part of the second term vanishes
and one obtains
ImQe1µ2 = −ImQτ 1µ2 (5.44)
where Q is a ‘quartet’ defined in (4.93). In eq. (5.44) one can check that the two ‘quartets’ Qe1µ2 and Qτ 1µ2
have symmetrical imaginary parts. Proceeding in the same way, one can easily conclude that due to the
orthogonality of any pair of different rows or columns of V , the imaginary parts of all ‘quartets’ are equal up
to a sign [58] and therefore equal to J up to a sign.
Now considering the ‘Majorana triangles’ in (5.43), one can define the necessary and sufficient conditions
for CP conservation
i) Their common area A = J/2 vanishes;
ii) Orientation of all Majorana triangles along the direction of the real or imaginary axis.
When one considers all imaginary parts of the ‘quartets’ to vanish, then as happens in the quark sector
triangles, this is equivalent to the vanishing of the ‘Dirac’ phase. However, as mentioned before, this is not
sufficient for CP conservation. One also requires ii) so the Majorana phases do not violate CP. If the three
collapsed triangles are on the real axis, then the imaginary part of the rephasing invariants of (4.92) vanish,
57
∗
i.e. ImUij Uik = 0 for every possible combination, and obviously CP is conserved. Moreover, if one of these
triangles is parallel to the imaginary axis, that means the Majorana neutrino χi and χk whose U elements
were considered in that triangle, have opposite CP parities, but there exists no CP violation.
In short, on the one hand, from the vanishing of triangle’s area one concludes that the Dirac phase van-
ishes. On the other hand, by requiring that all the sides of the triangles stay on the real or the imaginary
axis, we require that the rephasing invariants in eq. (4.92) and therefore the ‘Majorana’ phases, do not
violate CP. Thus we have just defined the necessary and sufficient conditions for CP to hold.
Let us explore the unitary triangles we derived. In the case, for instance, of vanishing areas, we know
that all the triangles must be oriented either in the real or the imaginary axis for CP to hold. As mentioned
before, that depends on CP parity of neutrinos. Therefore, we only need two triangles considering that with
them one can define both the relative CP parities in case of CP conservation or decide whether there are CP
violating ‘Majorana’ phases. Two of the most notable cases where this happens arise when neutrinos have
degenerate masses and when there is one and only one zero in U . In the former case, one should remind
that we explored the leptonic mixing matrix and conclude that the ‘Dirac’ phase is zero. Thus, since the
imaginary part of the quartets vanishes, we only need two triangles to define CP parities or violation. In
the latter, when, for instance, Ue3 vanishes, as suggested by the tribimaximal mixing matrix in eq. (4.85),
having in mind that one of the ‘quartets’ is zero, then all ‘quartets’ must be real, which means all triangles’
areas vanish.
Consider now that one neutrino is massless, for instance, mν1 = 0. In this case, one can freely rephase ν1
leaving the mass term invariant. With this rephasal of the terms Ui1 , the orientation of the first and second
triangle in (5.43) is not physically meaningful anymore. Thus, for CP conservation, one must require that
the area of the last triangle is zero to eliminate the ‘Dirac’ phase δ, and that it is aligned to the real or
imaginary axis, so the only Majorana phase existent vanishes.3 The cases mν2 = 0 and mν3 = 0 are similar
for obvious reasons.
5.6 Low energy invariants in leptonic sector

As explored in the course of this study, sources of CP violation can be associated only with the phases in
the lepton mass matrices, assuming all gauge currents are real and flavour diagonal. Thus, it is clear that
by rephasing the lepton fields (choosing a different basis), the phases of the elements of the mass matrix
also change, although containing the same physics. Having in mind a quark sector with three generations,
Jarlskog introduced in [58] a condition for CP invariance. Later on, a general solution to this issue was
introduced in [59] by the definition of invariants under field rephasing and basis transformation, therefore
in a basis independent way. This procedure is particularly useful from the point of view of model building,
since it does not depend on any particular choice of basis, and was for the first time applied to the leptonic
sector in [60].
It is also important to note that we will be deriving invariants in a WB. As mentioned before, a WB only
rewrites the Lagrangian in terms of new fields, and physics are invariant under such change. Therefore, we
can have the result of any physical process dependent on WB invariant quantities. In this section we see,
for instance, that in a framework with only one Dirac phase we only need one WB invariant, as customary
in the context of the quark sector. However, if one considers neutrinos to be Majorana, then it was shown
3 As mentioned in chapter 4.4 for non-degenerate masses, the number of ‘Majorana’ phases is equal to the number of massive
neutrinos subtracted by one.
58
in [61] that in the low energy theory one needs three and only three WB invariants.
Let us first define the CP phases arising from the Lagrangian in eq. (4.65). Using unitary freedom (4.67),
we shall consider two different weak-basis
WB-A mν − real, positive and diagonal

(5.45)
ml − Hermitian
WB-B ml − real, positive and diagonal

(5.46)
mν − symmetric
Analysing the WB-A, where U 0 = V 0 , we only have the following three phases,
φ1 = arg[(ml )12 ] (5.47)

φ2 = arg[(ml )23 ] (5.48)
φ3 = arg[(ml )13 ] (5.49)
due to the Hermiticity of ml , so these phases must be physical. Therefore, the necessary and sufficient
conditions for CP conservation in the non-degenerate case are given by
φi = 0 (5.50)
and, consequently, this is equivalent to α = β = δ = 0, where α and β are the ‘Majorana’ phases, and δ is
the ‘Dirac’ phase (the phases are understood mod π).
Let us now construct Hermitian matrices transforming under the unitary symmetries of (4.67). The
following matrices
A ≡ ml m†l , (5.51)
B ≡ m∗ν mν , (5.52)
C≡ m∗ν (ml m†l )∗ mν , (5.53)
can be a good choice when building invariants, since they all transform the same way, as
Mi → U 0† Mi U 0 , (5.54)
where Mi = A, B, C. At this point, one should recall the WB invariants in the quark sector. It has proved
that the following condition
tr[A, B]3 = 0 (5.55)
is a necessary and sufficient condition for CP conservation in the SM, or, equivalently, when one considers
non-degenerate masses and one ‘Dirac’ phase [1]. However, in the most general case in this minimal extension
of the SM, we have three phases in the leptonic sector, and for that reason we surely would need at least
three conditions for CP invariance. We can define them by computing the Jacobian

∂Ii
Det (5.56)
∂φj
where Ii are the invariants we want to determine. By imposing a non-zero Jacobian, one checks not only
59
invariant’s dependence on each physical basis φ, but also whether those objects are independent. Accordingly,
in a straightforward albeit tedious way, one can define the following possible set of independent CP-odd
invariants [61]
Tr[A, B]3 , (5.57)

3
Tr[A, C] , (5.58)
Tr([A, B]C), . (5.59)

∂Ii
which obey to Det ∂φj 6= 0, and define the three independent conditions for CP conservation in the
non-degenerate case as
Ii = 0 (5.60)
where Ii (i = 1, 2, 3) denote the invariants (5.57)-(5.59).
Since Ii is WB basis invariant, we can now consider WB-B. This is a more suitable basis to evaluate the
invariants, having in mind that charged lepton masses are well known.
Let us calculate I1
I1 = Tr[A, B]3 = 6i(∆l )µe (∆l )τ µ (∆l )τ e Im(B12 B23 B31 ), (5.61)
with (∆l )ij = (m2i − m2j ) (i, j = e, µ, τ ). This invariant is analogous to the one usually considered in the
quark sector. Therefore, one can express it as frequently in that sector as [62]
I1 = −6i(∆l )µe (∆l )τ µ (∆l )τ e (∆ν )21 (∆ν )32 (∆ν )31 JCP (5.62)
where JCP was already introduced in (4.94) and we also identified U 0 = UP M N S for transformations in (5.54).
As expected, since it is a Jarlskog-type invariant, it is ‘Majorana’ phase independent, whereas proportional
to the ‘Dirac’ phase. Naturally, this is equivalent to the vanishing of unitary triangles’ area. With the current
experimental data, and considering θ13 in-between the bounds set by T2K, for a Dirac-type phase sin δ ∼ 1
the leptonic J would be of order 10−2 , much larger than the J = 2.91+0.19
−0.11 × 10
−5
[15] of the quark sector.
Having in mind that there are two more WB invariants, I2 and I3 , that must vanish so CP holds, their
calculation yields
I2 = 6i(∆l )µe (∆l )τ µ (∆l )τ e Im(C12 C23 C31 ), (5.63)

I3 = −2i [(∆l )µe Im(B12 C21 ) + (∆l )τ µ Im(B23 C32 ) + (∆l )τ e Im(B13 C31 )] . (5.64)
When one calculates the above invariants, one checks that they both depend on a non-trivial function of
both ‘Majorana’ and ‘Dirac’ phases.
Let us now consider the degenerate case. As mentioned before, in this situation we have only one possible
CP violating phase, which is ‘Majorana’. Clearly, the only non-vanishing invariant is I2 and is given by
(5.63) - to check that this is indeed the case consider the WB-B: B is a multiple of the identity matrix, so
commutes with A; both I1 and I3 depend on [A, B], so they both vanish. Considering that mν is expressed
60
by eq. (4.89), I2 yields
I2 = −6i∆m Im[(Z◦ )11 (Z◦ )22 (Z◦ )∗12 (Z◦ )∗21 ]

3i
=− ∆m cos(θ) sin2 (θ) sin2 (2φ), sin(α) (5.65)
2
where Z◦ ≡ U◦∗ U◦† and ∆m = µ6 (m2τ − m2µ )2 (m2τ − m2e )2 (m2µ − m2e )2 . The latter factor contains the square
mass difference between charged leptons and the common neutrino mass µ.
Let us analyse the different factors. Considering the moduli-squared of the elements of the PMNS matrix,
usually arranged in the tribimaximal matrix already introduced in eq. (4.85), one can find that the angles
in the parametrisation (4.90) are restricted to be
θ = 70.52o , φ = 45o (5.66)
by, for instance, |Ue1 |2 and |Uµ3 |2 (we considered, without any loss of generalisation, that the angles are in
first quadrant). Therefore, it is rather obvious that this invariant is very useful since, in general, it does not
vanish, even in the limit of exact degeneracy. In particular, we can see that it only vanishes in two cases:
if α = 0, π (and multiples), which would be expectable since in that case the element eα/2 is either real or
purely imaginary, or when the common mass of neutrinos, µ, vanishes, which is an hypothesis ruled out by
experimental results.
Since we have just shown the usefulness of I1 in the Dirac CP violation evaluation, and I2 in the degenerate
case, let us analyse I3 . For that purpose, consider now WB-A, eq. (5.45), but this time one shall not restrict
ml to be Hermitian. In this WB, it yields explicitly
I3 =m1 m2 (m2 2 − m1 2 )(A12 2 − A21 2 ) + m1 m3 (m3 2 − m1 2 )(A13 2 − A31 2 )+

+ m2 m3 (m3 2 − m2 2 )(A23 2 − A32 2 ). (5.67)
As previously mentioned, this invariant would vanish in the degenerate case. Furthermore, it is also vanishing
if Aij 2 = Aji 2 (i 6= j). Since A ≡ ml m†l is Hermitian, this condition is equivalent to set a purely real or
imaginary Aij (i 6= j). Considering the transformation set in eq. (4.67) for ml , we obtain in index notation
Aij = Uli∗ Umj (d2l )lm

∗ ∗
= Uei Uej m2e + Uµi Uµj m2µ + Uτ∗i Uτ j m2τ . (5.68)
From the above equation, it is clear that, in the case Aij is purely real, we have neutrinos νi and νj with
the same CP parity; if Aij is purely imaginary then we have opposite CP parities. In any of the cases, A12 2
must be real for CP to hold. Thus, the vanishing of I3 is equivalent to the vanish of
Im tr Q =Im trABC (5.69)

=m1 m2 (m2 2 − m1 2 )Im(A12 2 ) + m1 m3 (m3 2 − m1 2 )Im(A13 2 )+
+ m2 m3 (m3 2 − m2 2 )Im(A23 2 ), (5.70)
which was for the first time introduced in [60]. To proceed with our study of this invariant, we shall firstly
present a framework, introduced for the first time in [63], which fully defines the leptonic mixing matrix in
61
∗
terms of its rephasing invariant bilinears of the type Uij Uik , eq. (4.92), reflecting our freedom to rephase
∗
the three charged lepton fields. We designate arg(Uij Uik ) ‘Majorana phases’. These are the minimal CP
violation quantities in the case of Majorana neutrinos,4 and it can be seen, through their definition, that
there are only six independent Majorana-type phases - all the other can be obtained by combinations of these
six phases. We can define them as
∗ ∗
β1 ≡ arg(Ue1 Ue2 ), β2 ≡ arg(Uµ1 Uµ2 ), β3 ≡ arg(Uτ 1 Uτ∗2 ),
(5.71)
∗ ∗
γ1 ≡ arg(Ue1 Ue3 ), γ2 ≡ arg(Uµ1 Uµ3 ), γ3 ≡ arg(Uτ 1 Uτ∗3 ).
These are the most simple rephasing invariants, and all the others can be build based on them. Indeed,
considering the above phases, we can present the following set of four independent Dirac-type invariant
phases built up from the ‘quartets’ (eq. (4.93))
∗ ∗ ∗
12
σeµ ≡ arg(Ue1 Uµ2 Ue2 Uµ1 ), 12
σeτ ≡ arg(Ue1 Uτ 2 Ue2 Uτ∗1 ),
(5.72)
∗ ∗ ∗
13
σeµ ≡ arg(Ue1 Uµ3 Ue3 Uµ1 ), 13
σeτ ≡ arg(Ue1 Uτ 3 Ue3 Uτ∗1 ),
which can be presented equivalently built from the Majorana-type phases as follows
12 12
σeµ = β1 − β2 , σeτ = β1 − β3 ,
(5.73)
13 13
σeµ = γ1 − γ2 , σeτ = γ1 − γ3 .
We can obtain the result of Im tr Q, eq. (5.70), by simply determining Im(A23 2 ). For that purpose, we
just have to use the property Im(Uij ) = Im(|Uij |eiθ ) = |Uij | sin θ, which allows us to obtain
Im(Aij 2 ) = − |Uei 2 Uej 2 |me 4 sin 2α1 − |Uµi 2 Uµj 2 |mµ 4 sin 2α2 − |Uτ i 2 Uτ j 2 |mτ 4 sin 2α3 −
− 2|Uei Uµi Uej Uµj |me 2 mµ 2 sin(α1 + α2 ) − 2|Uei Uτ i Uej Uτ j |me 2 mτ 2 sin(α1 + α3 )− (5.74)
2 2
− 2|Uµi Uτ i Uµj Uτ j |mµ mτ sin(α2 + α3 )
where αk is given by 

 βk , i = 1, j = 2
αk = γk , i = 1, j = 3 (5.75)


γ k − βk , i = 2, j = 3
Thus, we now only have to plug this result into our original equation (5.70), to obtain the invariant. Since
it is a messy combination of phases and moduli, we present it in eq. (C.1) of Appendix C. Note that if
there is no CP violation (either with odd or even CP parities in the case neutrinos are Majorana), then this
invariant vanishes. To simplify this invariant analysis, let us consider the possibility of a normal hierarchy
with m1 = 0. In such case, eq. (C.1) is rather simplified, and yields
Im tr Q = − m2 m3 (m3 2 − m2 2 )[|Ue2 2 Ue3 2 |me 4 sin(2γ1 − 2β1 ) + |Uµ2 2 Uµ3 2 |mµ 4 sin(γ1 − β1 + σeµ
12 13
− σeµ )
+ |Uτ 2 2 Uτ 3 2 |mτ 4 sin(γ1 − β1 + σeτ
12 13
− σeτ ) + 2|Ue2 Uµ2 Ue3 Uµ3 |me 2 mµ 2 sin(2γ1 − 2β1 + σeµ
12 13
− σeµ )
+ 2|Ue2 Uτ 2 Ue3 Uτ 3 |me 2 mτ 2 sin(2γ1 − 2β1 + σeτ
12 13
− σeτ )
+ 2|Uµ2 Uτ 2 Uµ3 Uτ 3 |mµ 2 mτ 2 sin(2γ1 − 2β1 + σeµ
12 12
+ σeτ 13
− σeµ 13
− σeτ )].
(5.76)
4 Conversely, if neutrino are Dirac particles, the minimal CP violation quantities are the invariant ‘quartets’, eq. (4.93),
instead of the invariant bilinears. This is analogous to the quark sector - see [1].
62
Consider now the dependence of the neutrinoless double beta decay on |mee |, eq. (4.106). In terms of
Majorana-type phases we obtain
|mee | =m1 |Ue1 |2 + m2 |Ue2 |2 + m3 |Ue3 |2 + 2m1 m2 |Ue1 |2 |Ue2 |2 cos(2β1 )

+ 2m1 m3 |Ue1 |2 |Ue3 |2 cos(2γ1 ) + 2m2 m3 |Ue2 |2 |Ue3 |2 cos(2β1 − 2γ1 ). (5.77)
Again, it explicitly does not depend on any Dirac phase, but only on Majorana-type phases, β1 and γ1 .
Moreover, when we consider, as before, m1 = 0, it only depends on one Majorana-type phase β1 − γ1 , since
|mee | = m2 |Ue2 |2 + m3 |Ue3 |2 + 2m2 m3 |Ue2 |2 |Ue3 |2 cos(2β1 − 2γ1 ), (5.78)
∗ ∗
where the angle β1 − γ1 is the argument of Ue1 Ue2 Ue1 Ue3 , which is not a rephasing invariant Dirac-type
‘quartet’. As one can see, even in the case there is no Dirac-type CP violation, (5.76) does not vanish and
can be related to neutrinoless double-β decay through the Majorana-type phase β1 − γ1 . In other words, if
CP is conserved, we have β1 − γ1 = 0, π/2 for, respectively, same and opposite CP parities, and I3 vanishes
(sin 2(β1 − γ1 ) = 0, whereas cos 2(β1 − γ1 ) = 1), but if only Dirac CP conservation is assured, then I3 would,
in principle, be non-vanishing, and there would be a direct relation between I3 and |mee |. Finally, we should
note that that the application of these results is restricted by the scarcity of data regarding leptonic mixing
and CP violation, so the leptonic mixing matrix cannot be fully reconstructed through this bilinears from
a set of presently feasible experiments. A possible hope for this is the development in our understanding
of flavour, and specifically the leptonic one. If, for instance, theory of flavour implies direct constraint on
bilinears, then the relations set from Majorana phases can be fulcral in its study.
5.7 Leptogenesis related invariants

As we shall see in the next chapter, leptogenesis relies on CP violation to generate a lepton number asymmetry.
Thus, it should be possible to define a set of invariants useful to study the relation between CP violation and
leptogenesis, which should vanish in case CP is conserved in those processes.
As it will come clear in the next chapter, CP violation is dependent on the imaginary part of m† m (recall
that we dropped the subscript for notation simplicity, m = mD ). Considering the parametrization in eq.
(5.36), we shall see that the lepton number asymmetry is only sensitive to the CP violating phases in Y4 ,
since
Im[(m† m)jk ]2 = Im[(Y4 † Y4 )jk ]2 , (i 6= j), (5.79)
Explicitly, the complex elements of the Hermitian matrix m† m are
(m† m)12 = |y21 |y2 exp(iβ1 ) + |y31 y32 | exp(i(β1 + σ)),

(m† m)23 = |y32 |y3 exp(−i(β2 − β1 − σ)), (5.80)
†
(m m)31 = |y31 |y3 exp(−iβ2 ).
Which means that the three phases σ, β1 and β2 appearing in Y4 are the ones that generate the lepton
number asymmetry. Furthermore, we can also constitute a set of linear equations in terms of the three
variables Im[(m† m)ij ]2 and by these means assure CP is conserved. Analogously to the procedure of the
previous section, and this time considering the transformations in eqs. (4.57) we set the following possible
63
set of independent CP invariants [64]
Im Tr[m† mM † M M ∗ mT m∗ M ] (5.81)
Im Tr[m† m(M † M )2 M ∗ mT m∗ M ] (5.82)
† † 2 ∗ T ∗ †
Im Tr[m m(M M ) M m m M M M ] (5.83)
and define three independent conditions for CP conservation in the non-degenerate case
Ii = 0 (5.84)
where Ii (i = 1, 2, 3) denote the invariants (5.81)-(5.83). In the WB where M is real diagonal we have
I1 = M1 M2 (M2 2 − M1 2 )Im[(m† m)212 ] + M1 M3 (M3 2 − M1 2 )Im[(m† m)213 ]

+ M2 M3 (M3 2 − M2 2 )Im[(m† m)223 ], (5.85)
4 4 † 4 4 †
I2 = M1 M2 (M2 − M1 )Im[(m m)212 ] + M1 M3 (M3 − M1 )Im[(m m)213 ]
+ M2 M3 (M3 4 − M2 4 )Im[(m† m)223 ], (5.86)
3 3 2 2 † 3 3 2 2 †
I1 = M1 M2 (M2 − M1 )Im[(m m)212 ] + M1 M3 (M3 − M1 )Im[(m m)213 ]
+ M2 3 M3 3 (M3 2 − M2 2 )Im[(m† m)223 ]. (5.87)
As mentioned, the above equations are a set of linearly independent equations in terms of the variables
Im[(m† m)2ij ]. This can be checked by noticing that the determinant of the coefficients of this set
Det = M1 2 M2 2 M3 2 ∆M21
2 2
∆M31 2
∆M32 (5.88)
in non-vanishing, except when there is degeneracy or a different number of heavy neutrinos is considered, so
that a different number of conditions should be fulfilled for CP to be conserved (note that ∆Mij = Mi 2 −
Mj 2 ). In short, we see that the vanishing of I1 , I2 , I3 implies the vanishing of Im[(m† m)212 ], Im[(m† m)213 ],
Im[(m† m)223 ], which, as we shall see in the next chapter, are the phases sensitive to leptogenesis. Together
with section 5.6, this section showed clearly that building invariants which are zero when CP holds is a
valuable tool regarding the study of CP violation in different processes, energies and models.
64
Chapter 6
Leptogenesis
There is good evidence that the Universe has a predominance of matter over antimatter. To explain it, we
shall present in this chapter one of the most simple and elegant models, Leptogenesis [13]. A beautiful aspect
of this mechanism is the connection between baryon asymmetry and neutrino properties, which we have been
analysing. One of the requirements for a successful baryogenesis through leptogenesis yields on the masses of
neutrinos, and for that purpose we will not only present a qualitative study of leptogenesis, but also present
a quantitative constraint on the mass of heavy neutrinos, which suggests the plausibility of the mechanism.
We shall study one of the most popular possibilities, ‘thermal leptogenesis’, with hierarchical masses. In
this scenario, which can happen in the context of seesaw mechanism type I, the N1 particles are produced by
scattering in the thermal bath, so that their number density can be calculated from the seesaw parameters
and the reheating temperature of the Universe.
Let us recall, from eq. (4.111), the observed Baryon Asymmetry of the Universe (BAU), but this time in
a more suitable parameter (see Appendix D.1 for details on ηB to Y∆B conversion)

nB − nB̄ −11
Y∆B ≡ = (8.75 ± 0.23) × 10 , (6.1)
s 0
where s is the entropy.

The basic idea of this baryogenesis via leptogenesis scenario is that scattering processes can produce
enough heavy neutrinos at temperatures T > M (where Mi is the mass of the heavy neutrino i). But when
temperature drops below M , the equilibrium number density is suppressed and if the relevant interactions
are CP violating, there can be created a net lepton L asymmetry in their out-of-equilibrium decays. This
asymmetry is then partially transformed in the observed BAU through the SM B + L violating processes,
the so-called sphaleron interactions. Note that the B − L symmetry of SM holds.
Let us consider the following assumptions
1. The lepton asymmetry is produced in a single flavour α;
2. To keep things simple in terms of kinematics, heavy neutrino masses are, as mentioned before, hierar-
chical: M1 ∼ 109 GeV M2 , M3 ;
3. Thermal production of N1 and negligible production of N2 and N3 .
Moreover, let us write the relevant Yukawa couplings of the Lagrangian in the mass eigenbasis
1
L = −h∗β Lβ φ̃lRβ + λ∗αk Lα φ∗ Nk + N j Mj Njc + h.c., (6.2)
2
65
where α, β = e, µ, τ and !
†
†T ϕ0
φ̃ ≡ iτ2 φ = . (6.3)
−ϕ−
Looking closely at this Lagrangian, we see that leptogenesis is very likely to take place once we are in the
context of seesaw mechanism, since the Sakharov conditions are (likely to be) fulfilled
i) Baryon number violation: The Lagrangian in (6.2) violate L because lepton number cannot be consis-
tently assigned to N1 in the presence of λ and M . In the case L(N1 ) = 1, although λα1 term respects
L, M1 violates it by two units. If L(N1 ) = 0, then M1 respects L but λα1 violates it by one unit. After
the creation of L 6= 0 (while B = 0), we rely on the sphaleron processes, which leave B − L invariant,
to create a B asymmetry;
ii) CP violation: In the mass eigenbasis where this equation is written, h and M are diagonal and real
matrices, where λ is a complex matrix. As mentioned in section 5.4, this matrix is defined by 9 moduli
and 6 complex phases. This can provide the necessary sources of CP violation;
iii) Departure from thermal equilibrium: The interactions of the Ni are only of the Yukawa type. At a
certain point, those interactions can be slower than the expansion rate of the universe, which lead to a
decay out of equilibrium.
One of our objectives must be the definition of the baryon asymmetry so that we see if it matches the
result obtained from observation (eq. (6.1)). The baryon asymmetry can be approximated as [65] (see
appendix D for details)
nb − nb 135ζ(3) X
Y∆B ≡ ' αα × ηα × C. (6.4)
s 4π 4 g∗ α
The first factor is the equilibrium number density of N1 divided by the entropy density at T M1 and is of
O(4 × 10−3 ) when we take the number of relativistic degrees of freedom g∗ to be ' 106.75 in SM. The other
three factors on the right-hand side represent the following physics aspects
1. factor αα is the CP asymmetry in N1 decays. In every 1/ decays, there is one more L than there are
L’s;
2. factor ηα is the efficiency factor. We must consider the existence of inverse decays, ‘wash out’ processes
and inefficiencies in N1 , that reduce the asymmetry by 0 < η < 1;
3. factor C describes a further dilution of the asymmetry due to fast processes which redistribute the
asymmetry in the lepton doublets. These include gauge, third generation Yukawa and B + L violating
non-perturbative processes. There can be a more accurate treatment to this factor, but for simplicity
we will approximate it by a single number.
In the following we will try to define or estimate these three factors. They depend on the various stages
of leptogenesis explained along with the definition of these factors, and which we try to systematise as
1. Creation/Definition of a net N1 number density, which is mainly due to the scattering processes of
fig. 6.1. This asymmetry depends on whether these processes are fast enough compared to the Hubble
expansion rate - if they are, we are in the context of a strong washout, and we have, at T ∼ M1 , an
equilibrium number density of N1 ; otherwise, as we will see, we are in a weak washout scenario, and
initial conditions are important;
66
2. Production of a net lepton number L: if N1 is in thermal equilibrium at T ∼ M1 , then no asymmetry
arises from the creation of N1 (see Appendix D.5). However, if we have a CP asymmetry in the decay
of N1 (Γ(N1 → φLα ) 6= Γ(N1 → φLα )), then there is an asymmetry between particles and antiparticles
which means we have a net L. In a weak washout regime, the final asymmetry will depend on the
initial conditions. This can lead to a more complicated analysis, but in both cases we will consider an
approach that treats this asymmetry qualitatively;
3. Obtaining a baryon number B asymmetry: with the production of a net L, which is dependent both
on CP violation and on the efficiency factor (out of equilibrium dynamics), it is possible to obtain a
net B, since SM Lagrangian symmetry is B − L.
Regarding the two latter stages, we should note a more accurate treatment of them can be achieved by
considering a set of Boltzmann equations. It is, however, out of the scope of this study. Those are consistently
described in [65, 66].
6.1 CP violation ()

The first parameter we shall compute is CP violation. Since we are considering the decay of a heavy neutrino
N1 , CP violation manifests itself in a different decay rate to final states with particles and antiparticles. Such
parameter is the following
Γ(N1 → φLα ) − Γ(N1 → φLα )

αα ≡ (6.5)
Γ(N1 → φL) + Γ(N1 → φL)
where one recalls that, since N1 is a Majorana fermion, N1 = N1 . The processes that contribute to the
non-conservation of CP at leading order and next to leading order are reproduced in fig. 6.1.
φ Lβ
λ Lβ
λ N2,3 N2,3
N1 ∗ ∗
λ
λ φ λ λ λ
φ
Lα
Figure 6.1: The diagrams contributing to the CP asymmetry αα . Note that in the case the intermediate
neutrino would be N1 , the contribution would be real, therefore we do not have to account it. The flavour
of the internal Lβ is summed.
Due to CPT invariance and unitarity, there is no difference between the rate of some process and its
CP-opposite (see appendix D.2). Therefore, it is important to include tree level and one-loop diagrams, and
it is precisely from their interference that CP violation arises. Let us separate their coupling constant part,
c, from the amplitude part, A,1
M = M0 + M1 = c0 A0 + c1 A1 (6.6)
1 Take, for instance, the tree level decay in fig. 6.1, and see that c0 = λ∗α1 and A0 (N → φ† Lα ) = uLα PR uN .
67
and consider the same procedure for the CP conjugate
M = M0 + M1 = c∗0 A0 + c∗1 A1 . (6.7)
where2 |Ai |2 = |Ai |2 . Thus the CP asymmetry can be written as

R R
|c0 A0 + c1 A1 |2 δ̃dΠL,φ − |c∗0 A0 + c∗1 A1 |2 δ̃dΠL,φ
αα = P hR R i
|c0 A0 + c1 A1 |2 δ̃dΠL,φ + |c∗0 A0 + c∗1 A1 |2 δ̃dΠL,φ
β
R
[(c0 c∗1 − c∗0 c1 )(A0 A∗1 ) − (c0 c∗1 − c∗0 c1 )(A∗0 A1 )δ̃dΠL,φ ]
' PR
2 |c0 A0 |2 δ̃dΠL,φ
β
R
Im(c0 c∗1 ) 2 Im(A0 A∗1 )δ̃dΠL,φ
= P R , (6.8)
|c0 |2 |A0 |2 δ̃dΠL,φ
β
where
d3 p L d3 pφ
δ̃ = (2π)4 δ 4 (Pi − Pf ), dΠL,φ = dΠL dΠφ = , (6.9)
2EL (2π)3 2Eφ (2π)3
and Pi , Pf are, respectively, the incoming four-momentum (in this case PN ) and the outgoing four-momentum
(in this case PL + Pφ ). The β index accounts for the sum of all three flavours (e, µ, τ ). Note that in the de-
nominator of the second equality we did not considered the 1-loop contribution, which is negligible compared
to tree level.
Since it is not in the scope of this study to calculate the loop diagrams, we consider the limit M2 , M3 M1
to obtain a lower bound in M1 . Under this approximation, we see that the effects of N2 and N3 can be
represented by an effective dimension-5 operator, which corresponds to shrinking the heavy propagator in
the loop to a point3 . This point, which has a direct coupling of a Higgs doublet φ and a lepton doublet LL ,
has associated the following dimension-5 operator
(m)αβ α β β +
(νL φ0 − eα +
L φ )(νL φ0 − eL φ ) + h.c., (6.10)
2v 2
where we considered the effective interaction obtained from seesaw type I. From the tree level coupling and
the above operator we obtain the relevant coupling constants
X
c0 = λα1 , c1 = 3 λ∗β1 (m)βα /v 2 , (6.11)
β
where the factor of three comes from the sum over all possible lepton/Higgs combinations in the loop (three
lepton flavours and two Higgs scalars). Within all these considerations, by considering the integration in
(6.8) and the term c0 c∗1 , the following result holds [65]
3M1
αα = Im[(λ)α1 (m∗ λ)α1 ] (6.12)
16πv 2 (λ† λ)11
2 In the CP conjugate amplitude, A, the u
Lα spinors are replaced by vLα spinors. However, inasmuch as we consider
relativistic particles (m ' 0), then uLα uLα = p
/ = vLα vLα .
3 An historical reference to this procedure is Fermi interaction (four fermion interaction), which was a non-renormalizable
6-dimension operator, and was later UV completed by the development of electroweak theory.
68
where no summation on α is considered in this equation. Having in mind this result, we can now define the
total CP asymmetry as
3M1 X
= 2 †
Im( λα1 m∗αβ λβ1 ). (6.13)
16πv (λ λ)11
α,β
Noting that the three-vector
λα1
λ̂α = p (6.14)
(λ† λ)11
P
is a unit vector ( α |λ̂α1 |2 = 1) and diagonalizing mν by considering a transformation as in eq. (4.67), we
obtain
X
Im( λα1 m∗αβ λβ1 ) = Im(λ̂T U U † m∗ U ∗ U T λ̂) = Im[(U T λ̂)T d(U T λ̂)]
α,β
X
= Im(λ̂0T dλ̂0 ) = Im( λ̂02
α mα ) ≤ mmax (6.15)
α
where λ̂0 ≡ U T λ̂ is another unit vector and mmax is the heaviest light neutrino mass. Accordingly, one can
obtain an upper limit for the CP asymmetry
3M1 mmax
|| ≤ . (6.16)
16πv 2
6.1.1 A further look on CP violation

Let us now go beyond the effective theory and include the N2,3 states as dynamical degrees of freedom.
Considering the diagrams in fig. 6.1, we see that from the interference between tree level diagram and one
loop level we should obtain some dependence on λ2 λ∗ 2 . Indeed, for a thermal production of Ni neutrinos,
with Nj as an intermediate state, one obtains the following result for the total CP asymmetry [64, 67, 68]
1 1 X 2
= †
Im (λ† λ)ij g(xj ) (6.17)
8π (λ λ)11
j6=i
where
Mj2
xj ≡ , (6.18)
Mi2
and

√ 1 1+x
g(x) = x + 1 + (1 + x) log . (6.19)
1−x x
√
From this result and recalling that mij = λij v/ 2 (m = mD ), we see, as mentioned in chapter 5.7, that the
lepton asymmetry depends on combinations of
2
Im (m† m)jk (i 6= j). (6.20)
69
It is clear that if (m† m)12 , (m† m)13 , (m† m)23 vanish, then no leptogenesis occur. Furthermore, since they
depend on the same variables, we can also define the total CP asymmetry in terms of the three WB invariants
I1 − I3 or model build taking those invariants in consideration.
From this result, we can also see that, in general, one cannot directly relate the size of CP violation
responsible for leptogenesis with the strength of CP violation at low energies. Indeed, considering eqs. (4.30)
and (5.40), we see that not only the effective left-handed neutrino mass matrix, mef f , is given in terms of
the six phases ρ, α1 , α2 , σ, β1 , β2 , but also that neutrino oscillation, sensible to the ‘Dirac’ phase δ and
whose strength can be obtained through the invariant I1 , eq. (5.57), is also dependent on those six phases,
while, in general, leptogenesis only depends on three phases, σ, β1 , β2 , as it is clear by eqs. (5.80).
More specifically, this means that we cannot directly relate the size of the phase δ to the amount of CP
violation required to generate the observed BAU, eq. (6.1), except in some special cases. Such cases would
be, for instance, if one and only one of the phases σ, β1 or β2 are non-vanishing - [64] explores a GUT
scenario where this takes place - or some conditions where we have X real in Casas-Ibarra parametrisation,
eq. (D.20), corresponding to CP conservation in the right-handed neutrino sector [69].
The bottom-line is that, on the one hand we can relate low energy CP violation and leptogenesis in the
context of some model within some specific conditions, but, on the other hand, in general we are unable to
relate both. As a consequence, it is possible the existence of CP violation at low energy, but not at high
energies - in our parametrisation with σ, β1 , β2 = 0 and some general ρ, α1 , α2 - or the opposite, i.e. having
CP conservation at low energy and leptogenesis (see [70] for details).
6.2 Out-of-equilibrium dynamics (η)

Another requirement to have a net baryon asymmetry is the departure from thermal equilibrium - the third
Sakharov condition. Indeed, the non-equilibrium required is provided by the expansion of the Universe:
interaction rates of order (or slower) than the Hubble expansion rate H are not fast enough to equilibrate
particle distributions.
We can classify interactions as much slower, much faster, or of the same order of H. Since we consider the
time scale of leptogenesis H −1 (otherwise it would not occur), then we neglect the much slower interactions.
Regarding the faster than H ones, their behaviour can be described by statistical distributions, described by
thermal masses, chemical and kinetic equilibrium.
The N1 decay, represented by ΓD , is controlled mainly by the tree level diagram in fig. 6.1. Thus it is
given approximately by
(λ† λ)11 M1
ΓD ' . (6.21)
8π
From the classification of interactions just presented, we can define the following rule for N1 decay out of
equilibrium
ΓD < H(T = M1 ), (6.22)
where H(T = M1 ) is the expansion rate of the Universe at the time the temperature equals the mass of N1
(see Appendix D.3)

1/2 M12
H(T = M1 ) = 1.66g∗ . (6.23)
mP l T =M1
70
At this point, it is useful to introduce two parameters [71] of the order of the light neutrino masses, m̃ and
m∗ , and which will represent, respectively, the decay rate ΓD and the expansion rate H(T = M1 ) in a more
intuitive way
8πv 2 (λ† λ)11 v 2

m̃ ≡ ΓD = , (6.24)
M1 M1
8πv 2
m∗ ≡ H(T = M1 ) ' 1.1 × 10−3 eV. (6.25)
M1
It can be shown [72] that m̃ > mmin where mmin is the lightest neutrino mass (see Appendix D.4). It is also
explored in [73] that ‘usually’ m̃ & ∆m12 . The introduction of these two parameters, m̃ and m∗ , allows us
to set the rule for the N1 decay out of equilibrium (eq. (6.22)) in the following convenient way
m̃ < m∗ (6.26)
In the case m̃ & ∆m12 , then this condition is not satisfied, thus the N1 total decay is in equilibrium. Such
range of parameters is referred to as ‘strong washout’. The opposite, less likely, case of decay out of equilib-
rium is the so-called ‘weak washout’. Let us explore both cases in the following lines.
Strong washout scenario: In this regime, at T ∼ M1 , a thermal number density of N1 is obtained

independent of the initial conditions, and therefore we have a L symmetric situation (see Appendix D.5
where it is demonstrated, in terms of chemical equilibrium constants, that thermal equilibrium prevents the
generation of asymmetries). In other words, we know that any YL asymmetry in the production of N1 is
washed out.
When we proceed with our analysis, and with the temperature dropping below M1 , we see that the inverse
decays, which initially washout the asymmetry (since they are faster than H), start being suppressed by a
factor of e−M1 /T
1
ΓID ≡ Γ(φL → N1 ) ' ΓD e−M1 /T . (6.27)
2
And it is clear that the asymmetry shall only survive once the inverse decay (and lepton number violating
scattering processes - fig. 6.1) is out of equilibrium, i.e.
1/2 T2
ΓID < H = 1.66g∗ . (6.28)
mP l
At temperature TF , where eq. (6.28) is satisfied, the original equilibrium quantity is suppressed ∝ e−M1 /TF .
Then after that, the N1 decay is out of equilibrium and contributes to the lepton asymmetry. Thus, the
efficiency factor η can be estimated the identity in the above inequality
nN1 (TF ) m∗
η' ' e−M1 /TF ' , m̃ > m∗ , (6.29)
nN1 (T N1 ) m̃
where m∗ /m̃ = H(T = M1 )/ΓD ' ΓID /ΓD . Note that this approximation applies for m̃ > m∗ ≈ 1×10−3 eV.
Note that since m̃ > mmin , if light neutrinos are degenerate, for instance ∼ 0.1eV, the efficiency factor
becomes, at least 0.01, which leads to a strong suppression of the lepton asymmetry generated by the decays
of thermally produced heavy neutrinos.
71
Weak washout scenario: In this regime the total decay rate is small, m̃ < m∗ , and the final lepton
asymmetry depends on the conditions at T = M1 (and consequently on conditions at T M1 ).
Let us consider that the Universe starts symmetric and that the initial N1 number density is zero. Since
we have m̃ < m∗ , the number density nN1 does not reach equilibrium number density due to the fact that
the rate of production [65]
h2t (λ† λ)11

Γprod ∼ T (6.30)
4π
is lower than H, i.e. Γprod (T = M1 ) < H(T = M1 ). Note that this conclusion is valid since ht ∼ 1, which
means Γprod is of the same order of ΓD . Moreover, it is important to notice that to obtain (6.30), we have
in mind the more effective production of N1 by 2 → 2 scattering processes in the s or t channel of a Higgs
coupled to a top quark4 , qL tR → φ → LL N and LL tR → φ → qL N (see fig.6.1), than by inverse decays,
φLL → N1 .
From the above production rate one can make a rough estimate of N1 number density. Furthermore,having
in mind the close relation between Γprod and ΓD , and the Hubble time τU = 1/H, one can estimate that
at the time T ∼ M1 , η ∼ ΓD τU ∼ m̃/m∗ . However this is the result we need, since we are only taking
in consideration only one of the stages of the asymmetry. We must define the asymmetry obtained for the
second stage, T < M1 . At first sight one would suggest that the final lepton asymmetry is zero. A non-zero
asymmetry survives, however, since there is a small part of the asymmetry made during N1 production which
was washed out before N1 decay, i.e., for T > M1 . This suppression can be estimated as [65] a factor of order
m̃/m∗ , heading to an efficiency for thermal leptogenesis and a weak washout regime of
m̃2
η∼ , m̃ < m∗ , nN1 (T M1 ) ' 0. (6.31)
m2∗
Most of the times we consider the Universe to have initial N1 number density nN1 (T M1 ) ' 0. When
we consider a strong washout, that initial condition does not have any influence in equilibrium N1 number
density. However, when one arises the weak washout scenario, the initial N1 population is relevant. For
instance, if N1 is initially in equilibrium number density, we have η ' 1, or if N1 number density is much
larger than equilibrium density, then we have effectively η > 1, which can be the case of a inflaton field
decaying mostly into N1 [74].
Finally, it is important to mention that the analysis presented is rather simplified, and it was in order to
have a general picture regarding of how the out of equilibrium dynamics is processed and how the efficiency
factor behaves. A deeper analysis can be achieved, namely considering a set of Boltzmann equations to
compute the lepton asymmetry produced, and for that purpose see, e.g. [65].
6.3 Lepton and B + L violation (Csphal )

The renormalisable Lagrangian of the SM conserves the baryon number B and the three lepton flavour
numbers Lα . However, due to the chiral anomaly, there are non-perturbative gauge field configurations that
violate B + L (L = Le + Lµ + Lτ ). In the early universe, at temperatures in the symmetric phase (above the
electroweak phase transition - EWPT), such configurations, the so-called ‘sphalerons’, occur frequently and
4 We neglected the other fermions, since the strongest Higgs coupling is to top quark. Moreover we did not consider the
coupling involving gauge interactions because it is not only simpler whereas consistent, but also because the main conclusions
would still be valid. [65]
72
lead to a rapid violation of B + L.
Let us start at a classical level where, according to Noether’s theorem, baryon B and lepton number L
are conserved quantum numbers. Thus, their associated current must be conserved. In other words, we have
twelve global U (1) symmetries in the SU (2) part of the Lagrangian, each one corresponding to a left-handed
i
field ψL (six quarks and six leptons)
i
ψL (x) → eiβ ψL
i
(x), (6.32)
having the following associated current
i
jµi (x)ψL γµ ψL
i
. (6.33)
which is conserved at tree level according to Noether’s theorem, i.e. ∂ µ jµi = 0. However, in the full quantum
theory this is not true anymore. Considering the baryonic and leptonic current
1 X X
jµB ≡ (qL γµ qL + uR γµ uR + dR γµ dR ) (6.34)
3
colors generations
X
jµL ≡ (LL γµ LL + eR γµ eR ) (6.35)
generations
its divergence is given by the so-called ‘triangle anomaly’ (for more details see [75–77]), and it is given by
∂ µ jµB = ∂ µ jµL (6.36)

Ng
= −g 2 Wµν
a
W̃ aµν + g 02 Bµν B̃ µν , (6.37)
32π 2
where Ng is the number of generations, Wµa and Bµ are, respectively, the SU (2) and U (1) gauge fields with
gauge couplings g and g 0 , and G̃aµν = 1/2µναβ Gαβa . Having in mind the above equality, we can define a
symmetry B − L whose current divergence vanishes.
At zero temperature, the transitions between states with different baryon number (which are obviously
B − L symmetric) are described by instantons. If the ground state of the gauge fields is pictured as a periodic
potential, where minima are the baryon number integers, the instantons correspond to vacuum fluctuations
that tunnel between minima. These fluctuations are associated to changes in baryon and lepton number
multiples of ∆B = ∆L = ±3. Thus, in the SM, SU (2) instantons lead to an effective 12-fermion interaction
(see fig. 6.2).
Y
OB+L = (QLi QLi QLi LLi ). (6.38)
i=1...3
The rate of these processes is very slow, of order e−4π/αW ∼ 10−160 [78]. Consider the analogy of the
ground state of the gauge fields pictured as a periodic potential, where minima are the baryon number
integers; when we have finite temperature one can imagine that a thermal fluctuation of the field can climb
over the barrier. Such process is the so-called sphaleron. The associated B + L violating rate mediated by
sphalerons is therefore Boltzmann suppressed, Γsph ∝ e−Esph /T , with Esph (T ) ∼ 2mW /αW ∼ 8 − 14TeV.
One can see that this suppression is very intense.
However, in leptogenesis we are interested in the B + L violating rate far above the EWPT. The large
B + L violating gauge field configurations occur frequently at T mW , and their rate can be estimated
73
Lτ Q3 (G)
Q3(B)
Q3(R)
Q2 (G)
Lµ Q2 (B)
Q2 (R)
Q1(G)
Q1(B)
Le Q1(R)
Figure 6.2: Sphaleron process. The incoming quantum numbers are L = 3 and B = 0. The outgoing quantum
numbers are L = 0 and B = −3
as [79]
5
ΓB+L violation ' 250αW T. (6.39)
For temperatures below 1012 GeV and above EWPT, B + L violating rates are faster than Hubble expansion,
therefore they are in equilibrium. As mentioned before, firstly we have a generation of L leaving B unchanged
obtained from the out of equilibrium decay of heavy Majorana neutrinos. Later on, we consider sphaleron
processes, which are B + L violating (although B − L conserving) and create a baryon asymmetry at a cost
of decreasing lepton asymmetry. The final baryon asymmetry is the following [80]

28

 79 T > TEW P T ,
Y∆B ' Y∆(B−L) × (6.40)

 12
37 T < TEW P T .
Clearly we are interested in temperatures above EWPT, where the rate of sphaleron processes is enough for
them to be in equilibrium. The rest of this section is dedicated to showing how this result is obtained.
Let us define
1
Y∆α = Y∆B − Y∆Lα (6.41)
3
so that
X
Y∆(B−L) = Y∆α . (6.42)
α
Note that the asymmetry Y∆α is conserved after leptogenesis, and our aim is to determine the final ratio
between leptons and baryons, which depends on the chemical equilibrium of the different particle species.
Considering only SM interactions, if they are fast enough compared to the expansion rate, then we can
define a set of conditions by inspecting its Lagrangian. Those conditions are described by chemical equilibrium
- the sum of chemical potential, over all particles entering the interaction, should be zero. For instance, the
first term in the Lagrangian in (6.2), h∗α Lα φ̃lRα , changes an SU (2) singlet lepton into a Higgs and an SU (2)
doublet lepton. Such chemical equilibrium would be define through the condition
µlα − µLα + µφ = 0, (6.43)
74
where µX is the chemical potential of the particle X. Note that quark and leptons have, in principle, three
flavours, so the total chemical potential as
X
µX = µXi . (6.44)
i=1,2,3
After determining the chemical potentials, we can relate them to the asymmetries in particle number densities
by expanding the distribution functions for small µ/T (see eq. (D.5) in Appendix D.1)
g
2

 6s T µi
i
(fermions),
ni − nī
Y∆i ≡ = (6.45)
s 
 gi 2
3s T µi (bosons).
where we used the identity ζ(2) = π 2 /6. gi is the number of degrees of freedom of the particle, and it is
defined for each particle of the spectrum as
gQ = 2 × 3 = 6, gu = gd = 3, gL = 2, gl = 1, gφ = 2 (6.46)
where the subscript Q, u, d, L, l refer, respectively, to the quark SU (2) doublet, the up quark singlet, the
down quark singlet, the lepton SU (2) doublet and the singlet charged lepton. Moreover, the results include
the three quark colours and the two particles present in SU (2) doublets (the spin is also implicit in the result,
since we are considering chiral fields). A further note: although we are taking µ/T 1, we should keep in
mind that the plasma from the early Universe is very different from a weakly coupled relativistic gas, and
this aspect can have an important influence in some cases.
For T above the EWPT, all SM interactions are fast. Yukawa interactions lead to the following relations
µui = µQi + µφ ,
µdi = µQi − µφ , (6.47)
µei = µLi − µφ .
Recalling eq. (6.39), we can take sphaleron interactions to be fast. This consideration leads to
3µQ + µL = 0. (6.48)
We can find another condition by requiring hypercharge neutrality (see Appendix D.5 for details)
µQ + 2µu − µd − µL − µl + 2µφ = 0 (6.49)
The baryon and lepton number asymmetries are given in terms of particle number asymmetries as
X
Y∆L = Y eq (∆yLi + ∆yli ),
i
1X (6.50)
Y∆B = Y eq (∆yQi + ∆yui + ∆ydi ),
3 i
and, just by taking in consideration the relation of eq. (6.45), the above can be written in terms of chemical
75
potentials
T 2 y eq
Y∆L = (gL µL + gl µl ),
6 (6.51)
T 2 y eq
Y∆B = (gQ µQ + gu µu + gd µd ).
18
Since we already expressed asymmetries in terms of chemical potentials, we just have to match the various
gX with the values in eq. (6.46) to obtain the relevant asymmetries in terms of the same variables
17 2 eq
Y∆L = − T y µQ , (6.52)
14
2 2 eq
Y∆B = T y µQ , (6.53)
3
79 2 eq
Y∆(B−L) = T y µQ , (6.54)
42
so we finally obtain eq. (6.40)
28 28
Y∆B = Y∆(B−L) , Csphal = . (6.55)
79 79
6.4 Baryon asymmetry from leptogenesis

After investigating the different factors in eq. (6.4), we can put it all together so that we obtain an estimate
of the baryon asymmetry obtained from leptogenesis. From the squared mass differences obtained experi-
mentally, we see that the parameter region m̃ > 10−3 eV is favoured, so we consider that, in principle, we
are in the strong washout scenario. Thus, we can take η to be given by (6.29), as well as Csphal given by
(6.55), to obtain the following result
10−3 (eV)
Y∆B ∼ 10−3 . (6.56)
m̃
where accounts with all CP violation contributions and is given approximately by (6.13), and m̃ is given
by eq. (6.24).
A plausible range for m̃ is the one suggested by the range of hierarchical light neutrino masses, 10−3 −
10−1 eV, so we can expect a mild washout effect, one that takes the efficiency factor η ∼ 10−2 − 10−1 . Thus,
to account with the BAU given by (6.1), Y∆B ∼ 10−10 , we need || & 10−6 − 10−5 . From eq. (6.13), and
taking a WB where m is diagonal mi one estimates that
X 10−14 M1 (GeV) mαβ (eV)

Im(λα1 λβ1 ) & 10−6 − 10−5 , (6.57)
(λ† λ)11
α,β
This condition is quite natural. More concretely, taking mi ∼ 10−2 eV, we see that for a real and imaginary
part of λ† λ of the same order, we have a very plausible lower limit for M1 & 1010 − 1011 GeV.
Therefore, we can conclude that leptogenesis is attractive not only because all the required features are
qualitatively present, but also because the quantitative constraints are plausibly satisfied, as we checked
for only mild washout effects and m̃ ∼ 0.01eV, suggested by the range of parameters still observationally
unconstrained. Furthermore, we see that the required CP asymmetry is achieved for the majority of the
seesaw parameter space.
76
Chapter 7
Conclusions
In this study we discussed a few subjects that arise when considering that the Standard Model neutrinos are
massive. During the last few decades, a large set of experiments involving solar, atmospheric, reactor and
accelerator neutrinos not only established this idea, but have also determined, up to a good precision, two
squared mass differences (∆21 , |∆31 |), and two mixing angles (sin2 θ12 , sin2 θ23 ). Furthermore, regarding the
other mixing angle, θ13 , we briefly analysed it, focusing our attention in the results presented in the current
year, which for the first time consistently show that this angle in non-vanishing. We stressed that if θ13 was
indeed zero, then no Dirac CP violation would occur.
Being electrically neutral, massive neutrinos can either be Dirac or Majorana, and that distinctive features
come from this fact. If neutrinos are Dirac particles, that means neutrinos acquire mass as a consequence of
a small Yukawa coupling, but then we have no fulfilling explanation to the smallness of their masses. In such
framework, we would only have one CP violating phase when considering three massive neutrinos. However, if
neutrinos are Majorana, we would only have to rely on the heaviness of some particles (right-handed neutrinos
in type I seesaw, scalar triplet in type II and fermion triplet in type III) to obtain small masses for neutrinos.
In such case, as we have studied along this work, we would have n(n − 1)/2 phases, which correspond to
three phases for three light massive neutrinos. Furthermore and ironically, the seesaw mechanism provides a
better understanding to the lightness of neutrinos than SM can provide to the generational hierarchy factor,
namely the 106 factor between electron and top quark masses.
We also briefly described and obtained approximations to both neutrino oscillation and double-β decay.
The former have, as mentioned, provided several results on mixing angles, squared mass differences and can,
in principle, provide a result to the possible Dirac phase. The latter is experimentally more challenging, but
if effective neutrino mass is large enough, then 0νββ may well be observed in experiments currently under
construction or development.
If neutrinos are Majorana, there is a possibility that their Majorana phases are non-vanishing, even
∗
though CP is conserved. For that to take place, one would only have to require arg (Uαi Uαj ) = ±π/2 (mod
π) associated to such rephasing invariants, which is equivalent to assigning different CP parities. If there is
no phase difference, then CP parity of neutrinos is the same, but in both cases we see that the impossibility of
rotating away the leptonic mixing matrix in the case of neutrinos with different CP parities, even in the limit
of exact degeneracy and in the CP conserving case, means that CP parities must hold. As it is also discussed
in this study, CP parities are meaningless when CP is violated. If that is the case, we saw that the extent
of ‘Dirac’ violation is dependent on the area of the unitary triangles presented, whereas their orientation
is meaningful in terms of ‘Majorana’ phases. If we look for a necessary and sufficient condition condition
in a basis independent way for CP to hold, we just need the imaginary part of the following invariants,
77
Tr[ml m†l , m∗ν mν ]3 , Tr([ml m†l , m∗ν mν ]m∗ν (ml m†l )∗ mν ), Tr[ml m†l , m∗ν (ml m†l )∗ mν ]3 to vanish. Clearly, in the
case of degenerate neutrinos this condition is relaxed, and we only need the imaginary part of the latter
invariant to vanish.
In this study we also discussed one possible mechanism responsible for the origin of matter in the Universe,
leptogenesis. In our view, this scenario is very appealing for four reasons:
1. Explicit lepton number violation is very natural when one includes right-handed neutrinos in the Stan-
dard Model, whose presence is very natural in the context of seesaw mechanism type I;
2. We have a mechanism, the sphaleron processes, which can automatically turn a leptonic asymmetry
into a baryonic one;
3. If neutrino masses lie in range 10−3 eV < mi < 0.1 eV, as suggested by neutrino oscillation experi-
ments, then the leptonic asymmetry produced in thermal leptogenesis is independent of any pre-existing
asymmetry;
4. Along with the qualitative features, we see that quantitative constraints are satisfied. In some rough
calculations, we obtained a very plausible limit for M & 1010 − 1011 .
Thus, we see that leptogenesis can be quantitatively successful without any fine-tuning of the seesaw
parameters, although it seems impossible that any direct test can be performed. Indeed, in general we cannot
even relate CP violation at low energies to the observed BAU. To establish leptogenesis experimentally, we
would need to produce the heavy Majorana neutrinos and measure the CP asymmetry in their decays.
However, in the most natural seesaw scenarios, these states are either simply too heavy to be produced,
or if they are light, then their Yukawa couplings must be very tiny, again preventing any chance of direct
measurements.
78
Appendix A
Some properties of the matrix C
The matrix C has some interesting properties which we want to derive now. If one charge conjugates the
same spinor twice, then naturally one must have the original spinor. Explicitly
[ψ c ]c = Cγ 0T [ψ C ]∗ = Cγ 0T C ∗ γ 0† ψ. (A.1)
The spinor kinetic terms must hold. Therefore, one can take
Cγ 0T C ∗ γ 0† = 1. (A.2)
∗
Another property can be derived by noting that, using (2.12) one can take C −1 γ0∗ C ∗ = −γ0† . In the
sequence, considering both (2.3) and (2.5), the following result can be written
∗
Cγ 0T C ∗ γ 0† = −Cγ 0T C ∗ C −1 γ 0∗ C ∗
= −C ∗ Cγ 0T γ 0∗
= −C ∗ Cγ 0T γ 0T = −C ∗ C.
which leads to the second interesting property of the matrix C
C ∗ C = −1 (A.3)
or equivalently, just by using C unitarity
C = −C T (A.4)
which means C must be an antisymmetric matrix in any representation of the Dirac matrices.
One can construct the matrix C just by specifying a representation of the Dirac algebra. Let us consider
the Dirac representation
! !
1 0 0 σi
γ0 = , γ0 = ; (A.5)
0 −1 −σi 0
where σi are the Pauli matrices. As one can easily check, γ0 is a real symmetric matrix, γ1 and γ3 are real
79
antisymmetric, and γ2 is imaginary symmetric. Therefore, considering relations in eq. (2.12), one can define
C = γ2 γ0 (A.6)
in the Dirac representation.
80
Appendix B
General mass term
To show that the general mass term describes two Majorana particles just consider Dirac-Majorana mass
term in eq. (2.34), with real mass for simplicity, and write it as follows
2LDM = mD (χω + ωχ) + mL χχ + MR ωω (B.1)

! !
mL mD χ
= (χω) . (B.2)
mD MR ω
Such mass term can be diagonalized to obtain its respective mass eigenvalues
mL + MR ± [(mL − MR )2 + mD 2 ]1/2
M1,2 = , (B.3)
2
which corresponds to the following Majorana mass eigenstates [81]
η1 = (cos θ)χ − (sin θ)ω

(B.4)
η2 = (sin θ)χ + (cos θ)ω,
with tan 2θ = 2mD /(mL − MR ). The above eigenstates clearly show that the most general mass term (B.2)
in fact describes two Majorana particles (B.4) with distinct masses. For instance, even when one considers
mL = 0, it describes two Majorana particles.
A particular assumption for this mass term, mL = MR = 0 recovers the usual four-component Dirac for-
malism. In that case we have θ = π/2, the mass eigenvalues, given by eq. (B.3), are ±mD and the eigenstates
√
are (χ ± ω)/ 2. Since we usually present the Dirac masses as positive, we apply a chiral transformation so
that we recover that formalism. Such chiral transformation is the following
! ! !
χ χ0 c
−ψL + ψL
γ5 = = , (B.5)
ω ω0 c
ψR − ψR
and flips the sign of the left-handed components, having in mind that γ5 PL = γ5 (1 − γ5 )/2 = (−1 + γ5 )/2 =
−PL . We can now introduce the fields which have the have the same eigenvalue mD
1 1 c c
√ (χ + ω) = √ (ψL + ψL + ψR + ψR ) ≡ ξ1 , (B.6)
2 2
1 1
√ (χ0 − ω 0 ) = √ (−ψL + ψLc
− ψR + ψRc
) ≡ ξ2 , (B.7)
2 2
81
and with this definition one can write the Dirac mass term as
c ψ c ) + h.c.
2LDM (mL = MR = 0) = mD (ψL ψR + ψR (B.8)
L
= mD (ξ1 ξ1 + ξ2 ξ2 ). (B.9)
Now, if one redefines the fields within the parentheses by π/4 as follows (both ξ1 and ξ2 have the same
eigenvalue, so it is an acceptable choice like any another)
1
ξ10 = √ (ξ1 − ξ2 ) = ψL + ψR (B.10)
2
1
ξ20 = √ (ξ1 + ξ2 ) = ψL
c c
+ ψR (B.11)
2
we explicitly retrieve the usual Dirac mass term mD ψψ with ψ = ψR + ψL .

Therefore, we can state that a Dirac fermion really corresponds to the degenerate limit of mL = MR = 0
when we consider the more general case, with two Majorana particles. Furthermore, we only need either
a non-vanishing mL or MR to have Majorana particles. However, since Majorana mass terms mL and
MR violate any U (1) charge (namely electric charge) all elementary fermions, except neutrinos, must have
mL = MR = 0.
82
Appendix C
The third low energy invariant
When we plug eq. (5.74) into eq. (5.70), we obtain, in the case of three generations
Im tr Q = − m1 m2 (m2 2 − m1 2 )[|Ue1 2 Ue2 2 |Me 4 sin 2β1 + |Uµ1 2 Uµ2 2 |mµ 4 sin 2β2 +
+ |Uτ 1 2 Uτ 2 2 |mτ 4 sin 2β3 + 2|Ue1 Uµ1 Ue2 Uµ2 |me 2 mµ 2 sin(β1 + β2 )+
+ 2|Ue1 Uτ 1 Ue2 Uτ 2 |me 2 mτ 2 sin(β1 + β3 ) + 2|Uµ1 Uτ 1 Uµ2 Uτ 2 |mµ 2 mτ 2 sin(β2 + β3 )]−
− m1 m3 (m3 2 − m1 2 )[|Ue1 2 Ue3 2 |me 4 sin 2γ1 + |Uµ1 2 Uµ3 2 |mµ 4 sin 2γ2
+ |Uτ 1 2 Uτ 3 2 |mτ 4 sin 2γ3 + 2|Ue1 Uµ1 Ue3 Uµ3 |me 2 mµ 2 sin(γ1 + γ2 )+
(C.1)
+ 2|Ue1 Uτ 1 Ue3 Uτ 3 |me 2 mτ 2 sin(γ1 + γ3 ) + 2|Uµ1 Uτ 1 Uµ3 Uτ 3 |mµ 2 mτ 2 sin(γ2 + γ3 )]−
− m2 m3 (m3 2 − m2 2 )[|Ue2 2 Ue3 2 |me 4 sin 2(γ1 − β1 ) + |Uµ2 2 Uµ3 2 |mµ 4 sin 2(γ2 − β2 )+
+ |Uτ 2 2 Uτ 3 2 |mτ 4 sin 2(γ3 − β3 ) + 2|Ue2 Uµ2 Ue3 Uµ3 |me 2 mµ 2 sin(γ1 + γ2 − β1 − β2 )+
+ 2|Ue2 Uτ 2 Ue3 Uτ 3 |me 2 mτ 2 sin(γ1 + γ3 − β1 − β3 )+
+ 2|Uµ2 Uτ 2 Uµ3 Uτ 3 |mµ 2 mτ 2 sin(γ2 + γ3 − β2 − β3 )].
Let us check this result in the case of two generations. Considering the parametrisation presented in [60]
! !
U11 U12 cos θ exp(iα1 ) − sin θ exp(iγ + iα1 )
= , (C.2)
U21 U22 sin θ exp(−iγ + iα2 ) cos θ exp(iα2 )
we define the Majorana-type phases as
∗ ∗
β1 ≡ arg(Ue1 Ue2 ) = −γ, β2 ≡ arg(Uµ1 Uµ2 ) = −γ (C.3)
For two generations, eq. (5.74) yields
Im(Aij 2 ) = −|U11 2 U12 2 |m1 4 sin 2β1 − |U21 2 U22 2 |m2 4 sin 2β2 − 2|U11 U21 U12 U22 |m1 2 m2 2 sin(β1 + β2 ) (C.4)
Thus, when we use the definition of U of (C.2) in the above equation, it yields
Im(H12 2 ) = −[m1 4 cos2 θ sin2 θ − m2 4 cos2 θ sin2 θ − 2 cos2 θ sin2 θm1 2 m2 2 ] sin(−2γ)
1
= (m2 2 − m1 2 )2 sin2 2θ sin 2γ,
4
83
from which we obtain the final result when we consider only two generations
1
Im tr Q = m1 m2 (m2 2 − m1 2 )(M2 2 − M1 2 )2 sin2 2θ sin 2γ. (C.5)
4
84
Appendix D
Calculating CP violation in Leptogenesis
D.1 The baryon asymmetry as an initial condition

The results of baryonic asymmetry were presented considering a parameter which relates the difference
between baryons and antibaryon divided by the total number of photon. However, this is not the most suitable
parameter, having in mind that, according to the Big Bang model, matter decoupled from Cosmic Background
Radiation (CBR) at the average kinetic energy of 0.3 eV, which means the production of CBR photons was
suppressed at that moment. Therefore, due to space-time expansion, the photon density decreases.
A more suitable parameter can be obtained by considering that the Universe is an adiabatically expanding
homogeneous media, which means the entropy s is approximately constant. Such parameter would be
nb − nb
Y∆B = , (D.1)
s
To determine s, we firstly introduce the energy density to be [15]

!
X 7X π2 4
ρ= gB + gf T (D.2)
8 30
F
where the first term is defined as g∗ and refers to the sum of relativistic degrees of freedom (d.f), having gB
as the bosonic d.f and gF the fermionic ones. Since we are considering to be above the electroweak phase
transition, we taking into account all the fermions and bosons of the SM (T > mtop ), we obtain g∗ ' 106.75.
Having in mind that at this point we are in a radiation dominated universe, we can also take pγ = ργ /3, so
that we define the entropy s to be
ρ+p 4ρ 2π 2
s= = = g∗ (T )T 3 . (D.3)
T 3T 45
We will also have to derive the equilibrium number densities, which can be done by integrating over the
Fermi-Dirac (+) or Bose-Einstein (-) distributions
eq 1
fi,± (p) = (D.4)
e(Ei −µi )/T ± 1
85
which, expanding to leading in the chemical potential µ/T and neglecting effects of order mi /T , gives
 µi
Z 3

 ζ(3) + T ζ(2) (bosons),
gi gi T
neq
i,± = 3
d eq
pfi,± (p) = 2 × (D.5)
(2π)3 π 
3 1 µi
4 ζ(3) + 2 T ζ(2) (fermions).
In the case of photons, the number density yields
2ζ(3) 3
nγ = T ' 0.2436T 3 , (D.6)
π2
where ζ(n) is the Riemann zeta function. We also define the d.f. at the time when heavy Majorana neutrinos
N were in equilibrium
3
7 Tν 21 4 43
g0 = 2 + 3 × 2 × =2+ = , (D.7)
8 T 4 11 11
which results from the two polarizations of the gauge field and from the three neutrinos with two spin states
(3×2). Considering the above result we can obtain the ratio of photon number density to the entropy density,
which is
s0
' 7.04, (D.8)
nγ0
and from this result we can find the relation between ηB (eq. (4.111)) and Y∆B (eq. (D.1)) at the present
time, which is ηB ' 7.04Y∆B .
To obtain (6.4), we just use the equilibrium number density for massless fermions from (D.5)
3ζ(3)gi T 3
neq
i = , (D.9)
4π 2
where gi is the number of internal d.f. of the particle. In our approximation, the thermal production of N2
and N3 is negligible, thus we only have to consider N1 . Having in mind that N1 are Majorana fermions, we
take gN1 = 2 and can finally obtain the factor in eq. (6.4)
neq
N1 135ζ(3)
' . (D.10)
s 4π 4 g∗
Note that the equilibrium number density of N1 is the maximum that can arise via thermal N1 production,
and we must account with a sufficient large λα1 . The three other factors in eq. (6.4) will naturally decrease
the ideal value.
D.2 Implication of unitarity and CPT in decays

Unitarity and CPT induce some important constraints on CP violation. Indeed, we must take into account
loop level processes for the account of CP asymmetry. To show such statement, we must consider, from CPT ,
that
|M(a → b)|2 = |M(b → a)|2 . (D.11)
86
P †
Moreover, we also take into account the unitarity condition at tree level b ha|S|bihb|S |ai = 1, which implies
X X
|M(a → b)|2 = |M(b → a)|2 . (D.12)
b b
Combining both conditions above, one obtains

X X
|M(N1 → φLα )|2 = |M(N1 → φLα )|2 . (D.13)
α α
It is important to notice that, despite the total CP invariance in the process, there can be CP asymmetry in
a partial decay rate, i.e., for a specific Lα .
D.3 Hubble expansion rate

To obtain eq. (6.23), one must take in consideration Friedmann equations
k 8πGρ Λ
H2 + 2
= + (D.14)
a 3 3
when one considers the Universe to be flat (k = 0) and the cosmological constant to be negligible (Λ = 0)
then the positive solution for the above equation is the following
1/2
8πGρ
H= . (D.15)
3
Considering the following identity regarding the Planck mass mP l
mP l ≡ G−1/2 , (D.16)
and eq. (D.2), then eq. (D.15) yields
1/2 M12
H ' 1.66g∗ (D.17)
mP l
D.4 Decay rate minimum

From the usual seesaw type I formula (4.29), we can define in the WB where MR and ml are diagonal, real
and positive (MR = D)
−1/2 T
X = vMR λ K ∗ d−1/2 . (D.18)
Clearly X follows the condition
X T X = 1. (D.19)
Solving for λ we find the so-called Casas-Ibarra parametrization [82]
1 1/2 1/2 T
λT = M Xd K . (D.20)
v R
87
With this parametrization we can see that the parameter m̃ in eq. (6.24) is bounded from below as follows
X
m̃1 = mi |U1i |2
i
X
> min(mj ) |U1i |2
i
X
2
≥ min(mj ) U1i = min(mj ). (D.21)

i
In the last equation we have applied (D.19).
D.5 Thermal equilibrium and the third Sakharov condition

We shall show, by the use of chemical equilibrium constrains imposed by fast interactions, that no asymmetry
may be generated in thermal equilibrium. For simplicity consider only a single generation whose relevant
chemical potentials µ are the following
µQ , µu , µd , µL , µl , µφ . (D.22)
where we do not consider the heavy neutrinos N , since at this point since the N1 ’s previously in equilibrium,
have already decayed. As we considered in section 6.3 - eq. (6.47) and (6.48)- the fast Yukawa and sphaleron
interactions lead to the following four constraints
µQ − µu + µφ = 0,
µQ − µd − µφ = 0,
µQ − µu − µφ = 0, (D.23)
µL − µφ + µN = 0,
3µQ + µL = 0.
The thermal equilibrium of the plasma is also affected by its charge (in this case the weak hypercharge),
which restricts the plasma to obbey to
X
gi Yxi µxi = 0 (D.24)
xi
where Yxi is the hypercharge of the particle xi . In table D.1 we present the hypercharges of the different
particles. By considering both the degrees of freedom of particles, eq. (6.46), and hypercharges in table D.1,
we can define hypercharge neutrality condition, eq. (D.24), as
µQ + 2µu − µd − µL − µl + 2µφ = 0. (D.25)
We now have seven conditions for seven chemical potentials, which have the solution
µQ = µu = µd = µL = µl = µφ = µN = 0, (D.26)
showing that no asymmetry is generated. If, on the other hand, the singlet neutrino is out of equilibrium,
88
Fermionic Field Q I3 Y Bosonic Field Q I3 Y
νL 0 1/2 -1 φ+ 1 1/2 1
lL -1 -1/2 -1 φ0 0 -1/2 1
νR 0 0 0 B0 0 0 0
lR -1 0 -2 W+ 1 1 0
uL 2/3 1/2 1/3 W− -1 -1 0
dL -1/3 -1/2 1/3 W3 0 0 0
uR 2/3 0 4/3
dR -1/3 0 -2/3
Table D.1: SU (2) × U (1) quantum numbers above electroweak phase transition.
then µN 6= 0 and the condition (D.23) evolves, allowing non-vanishing asymmetries.

Finally, it is interesting to refer that if one of the interactions is slow, then the related condition on
chemical potential is relaxed, and a corresponding conservation law arises [83]. This would be the case of
slow EW sphaleron processes, which naturally would lead to a baryon number conservation, ∆B = 0.
89
Bibliography
[1] G. C. Branco, L. Lavoura, and J. P. Silva, CP violation. Oxford University Press, Oxford, UK, 1999.
[2] R. N. Mohapatra and G. Senjanovic, “Neutrino Mass and Spontaneous Parity Violation,”
Phys.Rev.Lett. 44 (1980) 912.
[3] S. Glashow, “The Future of Elementary Particle Physics,” NATO Adv.Study Inst.Ser.B Phys. 59
(1980) 687. Preliminary version given at Colloquium in Honor of A. Visconti, Marseille-Luminy Univ.,
Jul 1979.
[4] M. Gell-Mann, P. Ramond, and R. Slansky Proceedings of the Supergravity Stony Brook Workshop
(1980) . Eds. P. Van Niewenhuizen, D. Freeman (North-Holland, Amsterdam).
[5] T. Yanagida, “Horizontal Symmetry and Masses of Neutrinos,” Proceedings of the Workshop on
Unified Theories and Baryon Number in the Universe (1979) . Eds. A. Sawada, A. Sugamoto, KEK
Report No. 79-18.
[6] E. Majorana, “Theory of the Symmetry of Electrons and Positrons,” Nuovo Cim. 14 (1937) 171–184.
[7] T. Schwetz, M. Tortola, and J. W. F. Valle, “Where we are on θ13 : addendum to ’Global neutrino data
and recent reactor fluxes: status of three- flavour oscillation parameters’,” arXiv:1108.1376
[hep-ph].
[8] T2K Collaboration Collaboration, K. Abe et al., “Indication of Electron Neutrino Appearance from
an Accelerator-produced Off-axis Muon Neutrino Beam,” Phys.Rev.Lett. 107 (2011) 041801,
arXiv:1106.2822 [hep-ex].
[9] M. Kobayashi and T. Maskawa, “CP Violation in the Renormalizable Theory of Weak Interaction,”
Prog.Theor.Phys. 49 (1973) 652–657.
[10] J. Schechter and J. Valle, “Neutrino Masses in SU(2) x U(1) Theories,” Phys.Rev. D22 (1980) 2227.
[11] A. D. Sakharov, “Violation of CP Invariance, c Asymmetry, and Baryon Asymmetry of the Universe,”
Pisma Zh. Eksp. Teor. Fiz. 5 (1967) 32–35. [JETP Lett.5:24-27,1967].
[12] A. Dolgov, “NonGUT baryogenesis,” Phys.Rept. 222 (1992) 309–386.
[13] M. Fukugita and T. Yanagida, “Baryogenesis Without Grand Unification,” Phys.Lett. B174 (1986) 45.
[14] J. D. Bjorken and S. D. Drell, “Relativistic quantum fields,”. International Series in Pure and Applied
Physics. McGraw-Hill Book Company, New York-St.Louis-San Francisco-Toronto-London-Sydney
(ISBN-10: 0-07-005494-0).
91
[15] Particle Data Group Collaboration, K. Nakamura et al., “Review of particle physics,” J. Phys. G37
(2010) 075021.
[16] R. N. Mohapatra et al., “Theory of neutrinos,” arXiv:hep-ph/0412099.
[17] T. Schwetz, M. Tortola, and J. Valle, “Global neutrino data and recent reactor fluxes: status of
three-flavour oscillation parameters,” New J.Phys. 13 (2011) 063004, arXiv:1103.0734 [hep-ph].
[18] G. Fogli, E. Lisi, A. Marrone, A. Palazzo, and A. Rotunno, “Evidence of θ13 > 0 from global neutrino
data analysis,” Phys.Rev. D84 (2011) 053007, arXiv:1106.6028 [hep-ph].
[19] R. Mohapatra, S. Antusch, K. Babu, G. Barenboim, M.-C. Chen, et al., “Theory of neutrinos: A
White paper,” Rept.Prog.Phys. 70 (2007) 1757–1867, arXiv:hep-ph/0510213 [hep-ph].
[20] W. Rodejohann, “Neutrino-less Double Beta Decay and Particle Physics,” arXiv:1106.1334
[hep-ph].
[21] C. Kraus, B. Bornschein, L. Bornschein, J. Bonn, B. Flatt, et al., “Final results from phase II of the
Mainz neutrino mass search in tritium beta decay,” Eur.Phys.J. C40 (2005) 447–468,
arXiv:hep-ex/0412056 [hep-ex].
[22] K. Abazajian, E. Calabrese, A. Cooray, F. De Bernardis, S. Dodelson, et al., “Cosmological and

Astrophysical Neutrino Mass Measurements,” arXiv:1103.5083 [astro-ph.CO].
[23] C. Weinheimer, “Direct determination of Neutrino Mass from H-3 beta Spectrum,” arXiv:0912.1619
[hep-ex].
[24] S. Pascoli, S. Petcov, and T. Schwetz, “The Absolute neutrino mass scale, neutrino mass spectrum,
majorana CP-violation and neutrinoless double-beta decay,” Nucl.Phys. B734 (2006) 24–49,
arXiv:hep-ph/0505226 [hep-ph].
[25] SNO Collaboration Collaboration, Q. Ahmad et al., “Measurement of the rate of nu/e + d –¿ p + p
+ e- interactions produced by B-8 solar neutrinos at the Sudbury Neutrino Observatory,”
Phys.Rev.Lett. 87 (2001) 071301, arXiv:nucl-ex/0106015 [nucl-ex].
[26] SNO Collaboration Collaboration, B. Aharmim et al., “Electron energy spectra, fluxes, and
day-night asymmetries of B-8 solar neutrinos from measurements with NaCl dissolved in the
heavy-water detector at the Sudbury Neutrino Observatory,” Phys.Rev. C72 (2005) 055502,
arXiv:nucl-ex/0502021 [nucl-ex].
[27] M. Thomson, “Long Baseline Neutrino Oscillation Experiments,” eConf C080625 (2008) 0005.
[28] V. D. Barger, K. Whisnant, and R. Phillips, “CP Violation in Three Neutrino Oscillations,”
Phys.Rev.Lett. 45 (1980) 2084.
[29] H. Nunokawa, S. J. Parke, and J. W. Valle, “CP Violation and Neutrino Oscillations,”
Prog.Part.Nucl.Phys. 60 (2008) 338–402, arXiv:0710.0554 [hep-ph].
[30] W. Grimus and M. N. Rebelo, “Automorphisms in gauge theories and the definition of CP and P,”
Phys. Rept. 281 (1997) 239–308, arXiv:hep-ph/9506272.
92
[31] J. Schechter and J. Valle, “Neutrino Decay and Spontaneous Violation of Lepton Number,” Phys.Rev.
D25 (1982) 774.
[32] G. Lazarides, Q. Shafi, and C. Wetterich, “Proton Lifetime and Fermion Masses in an SO(10) Model,”
Nucl.Phys. B181 (1981) 287.
[33] R. N. Mohapatra and G. Senjanovic, “Neutrino Masses and Mixings in Gauge Models with
Spontaneous Parity Violation,” Phys.Rev. D23 (1981) 165.
[34] C. Wetterich, “Neutrino Masses and the Scale of B-L Violation,” Nucl.Phys. B187 (1981) 343.
[35] R. Foot, H. Lew, X. He, and G. C. Joshi, “Seesaw Neutrino Masses Induced by a Triplet of Leptons,”
Z.Phys. C44 (1989) 441.
[36] W.-Y. Keung and G. Senjanovic, “Majorana Neutrinos and the Production of the Right-handed
Charged Gause Boson,” Phys.Rev.Lett. 50 (1983) 1427.
[37] Z. Maki, M. Nakagawa, and S. Sakata, “Remarks on the unified model of elementary particles,”
Prog.Theor.Phys. 28 (1962) 870–880.
[38] B. Pontecorvo, “Neutrino Experiments and the Problem of Conservation of Leptonic Charge,”
Sov.Phys.JETP 26 (1968) 984–988.
[39] E. K. Akhmedov, G. Branco, F. Joaquim, and J. Silva-Marcos, “Neutrino masses and mixing with
seesaw mechanism and universal breaking of extended democracy,” Phys.Lett. B498 (2001) 237–250,
[40] S. Pakvasa, W. Rodejohann, and T. J. Weiler, “Unitary parametrization of perturbations to

tribimaximal neutrino mixing,” Phys.Rev.Lett. 100 (2008) 111801, arXiv:0711.0052 [hep-ph].
[41] P. Harrison, D. Perkins, and W. Scott, “Tri-bimaximal mixing and the neutrino oscillation data,”
Phys.Lett. B530 (2002) 167, arXiv:hep-ph/0202074 [hep-ph].
[42] G. C. Branco, M. N. Rebelo, and J. I. Silva-Marcos, “Degenerate and quasi degenerate Majorana
neutrinos,” Phys. Rev. Lett. 82 (1999) 683–686, arXiv:hep-ph/9810328.
[43] C. Jarlskog, “Recursive parameterisation and invariant phases of unitary matrices,” J.Math.Phys. 47
(2006) 013507, arXiv:math-ph/0510034 [math-ph].
[44] L. Wolfenstein, “Neutrino Oscillations in Matter,” Phys.Rev. D17 (1978) 2369–2374.
[45] E. K. Akhmedov and J. Kopp, “Neutrino oscillations: Quantum mechanics vs. quantum field theory,”
JHEP 1004 (2010) 008, arXiv:1001.4815 [hep-ph].
[46] B. Kayser, “Neutrino Oscillation Phenomenology,” arXiv:0804.1121 [hep-ph].
[47] E. K. Akhmedov and A. Y. Smirnov, “Paradoxes of neutrino oscillations,” Phys.Atom.Nucl. 72 (2009)

1363–1381, arXiv:0905.1903 [hep-ph].
[48] I. Avignone, Frank T., S. R. Elliott, and J. Engel, “Double Beta Decay, Majorana Neutrinos, and
Neutrino Mass,” Rev.Mod.Phys. 80 (2008) 481–516, arXiv:0708.1033 [nucl-ex].
[49] P. Vogel, “Nuclear physics aspects of double beta decay,” arXiv:0807.2457 [hep-ph].
93
[50] R. Mohapatra and P. Pal, “Massive neutrinos in physics and astrophysics. Second edition,” World
Sci.Lect.Notes Phys. 60 (1998) 1–397.
[51] M. Mitra, G. Senjanovic, and F. Vissani, “Neutrinoless Double Beta Decay and Heavy Sterile
Neutrinos,” arXiv:1108.0004 [hep-ph].
[52] W. Rodejohann, “Inverse Neutrino-less Double Beta Decay Revisited: Neutrinos, Higgs Triplets and a
Muon Collider,” Phys.Rev. D81 (2010) 114001, arXiv:1005.2854 [hep-ph].
[53] WMAP Collaboration Collaboration, E. Komatsu et al., “Seven-Year Wilkinson Microwave

Anisotropy Probe (WMAP) Observations: Cosmological Interpretation,” Astrophys.J.Suppl. 192
(2011) 18, arXiv:1001.4538 [astro-ph.CO].
[54] L. Wolfenstein, “CP Properties of Majorana Neutrinos and Double beta Decay,” Phys. Lett. B107
(1981) 77.
[55] T. Morozumi, T. Satou, M. N. Rebelo, and M. Tanimoto, “The top quark mass and flavour mixing in
a seesaw model of quark masses,” Phys. Lett. B410 (1997) 233–240, arXiv:hep-ph/9703249.
[56] J. Hashida, T. Morozumi, and A. Purwanto, “Neutrino mixing in seesaw model,” Prog. Theor. Phys.
103 (2000) 379–391, arXiv:hep-ph/9909208.
[57] J. Aguilar-Saavedra and G. Branco, “Unitarity triangles and geometrical description of CP violation
with Majorana neutrinos,” Phys.Rev. D62 (2000) 096009, arXiv:hep-ph/0007025 [hep-ph].
[58] C. Jarlskog, “Commutator of the Quark Mass Matrices in the Standard Electroweak Model and a
Measure of Maximal CP Violation,” Phys. Rev. Lett. 55 (1985) 1039.
[59] J. Bernabeu, G. Branco, and M. Gronau, “CP Restrictions on Quark Mass Matrices,” Phys.Lett. B169
(1986) 243–247.
[60] G. C. Branco, L. Lavoura, and M. N. Rebelo, “Majorana Neutrinos and CP Violation in the Leptonic
Sector,” Phys. Lett. B180 (1986) 264.
[61] H. K. Dreiner, J. S. Kim, O. Lebedev, and M. Thormeier, “Supersymmetric Jarlskog invariants: The
neutrino sector,” Phys. Rev. D76 (2007) 015006, arXiv:hep-ph/0703074.
[62] G. C. Branco and M. N. Rebelo, “Neutrino Physics and CP violation,” Acta Phys. Polon. B38 (2007)
3819–3850, arXiv:0711.2650 [hep-ph].
[63] G. C. Branco and M. Rebelo, “Building the full PMNS Matrix from six independent Majorana-type
phases,” Phys.Rev. D79 (2009) 013001, arXiv:0809.2799 [hep-ph].
[64] G. C. Branco, T. Morozumi, B. M. Nobre, and M. N. Rebelo, “A bridge between CP violation at low
energies and leptogenesis,” Nucl. Phys. B617 (2001) 475–492, arXiv:hep-ph/0107164.
[65] S. Davidson, E. Nardi, and Y. Nir, “Leptogenesis,” Phys.Rept. 466 (2008) 105–177, arXiv:0802.2962
[hep-ph].
[66] W. Buchmuller, R. Peccei, and T. Yanagida, “Leptogenesis as the origin of matter,”

Ann.Rev.Nucl.Part.Sci. 55 (2005) 311–355, arXiv:hep-ph/0502169 [hep-ph].
94
[67] T. Endoh, T. Morozumi, T. Onogi, and A. Purwanto, “CP violation in seesaw model,” Phys.Rev. D64
(2001) 013006, arXiv:hep-ph/0012345 [hep-ph].
[68] L. Covi, E. Roulet, and F. Vissani, “CP violating decays in leptogenesis scenarios,” Phys.Lett. B384
(1996) 169–174, arXiv:hep-ph/9605319 [hep-ph].
[69] G. Branco, R. Gonzalez Felipe, and F. Joaquim, “A New bridge between leptonic CP violation and
leptogenesis,” Phys.Lett. B645 (2007) 432–436, arXiv:hep-ph/0609297 [hep-ph].
[70] M. Rebelo, “Leptogenesis without CP violation at low-energies,” Phys.Rev. D67 (2003) 013008,
[71] W. Buchmuller and M. Plumacher, “Baryon asymmetry and neutrino mixing,” Phys.Lett. B389 (1996)
73–77, arXiv:hep-ph/9608308 [hep-ph].
[72] M. Fujii, K. Hamaguchi, and T. Yanagida, “Leptogenesis with almost degenerate majorana neutrinos,”
Phys.Rev. D65 (2002) 115012, arXiv:hep-ph/0202210 [hep-ph].
[73] W. Buchmuller, P. Di Bari, and M. Plumacher, “The Neutrino mass window for baryogenesis,”
Nucl.Phys. B665 (2003) 445–468, arXiv:hep-ph/0302092 [hep-ph].
[74] G. Giudice, A. Notari, M. Raidal, A. Riotto, and A. Strumia, “Towards a complete theory of thermal
leptogenesis in the SM and MSSM,” Nucl.Phys. B685 (2004) 89–149, arXiv:hep-ph/0310123
[hep-ph].
[75] G. ’t Hooft, “Symmetry Breaking Through Bell-Jackiw Anomalies,” Phys.Rev.Lett. 37 (1976) 8–11.
[76] G. ’t Hooft, “Computation of the Quantum Effects Due to a Four-Dimensional Pseudoparticle,”

Phys.Rev. D14 (1976) 3432–3450.
[77] J. Callan, Curtis G., R. Dashen, and D. J. Gross, “The Structure of the Gauge Theory Vacuum,”
Phys.Lett. B63 (1976) 334–340.
[78] S. R. Coleman, “The Uses of Instantons,” Subnucl.Ser. 15 (1979) 805.

5
[79] P. B. Arnold, D. Son, and L. G. Yaffe, “The Hot baryon violation rate is O(αW T 4 ),” Phys.Rev. D55
(1997) 6264–6273, arXiv:hep-ph/9609481 [hep-ph].
[80] S. Khlebnikov and M. Shaposhnikov, “The Statistical Theory of Anomalous Fermion Number
Nonconservation,” Nucl.Phys. B308 (1988) 885–912.
[81] T. Cheng and L. Li, “Gauge Theory of Elementary Particle Physics,”.
[82] J. Casas and A. Ibarra, “Oscillating neutrinos and muon —¿ e, gamma,” Nucl.Phys. B618 (2001)
171–204, arXiv:hep-ph/0103065 [hep-ph].
[83] C. S. Fong, M. Gonzalez-Garcia, E. Nardi, and J. Racker, “Supersymmetric Leptogenesis,” JCAP

1012 (2010) 013, arXiv:1009.0003 [hep-ph].
[84] J. W. F. Valle, “Neutrino physics overview,” J. Phys. Conf. Ser. 53 (2006) 473–505,
arXiv:hep-ph/0608101.
[85] P. B. Pal, “Dirac, Majorana and Weyl fermions,” arXiv:1006.1718 [hep-ph].
95
[86] Y. Burnier, M. Laine, and M. Shaposhnikov, “Baryon and lepton number violation rates across the
electroweak crossover,” JCAP 0602 (2006) 007, arXiv:hep-ph/0511246 [hep-ph].
[87] G. Senjanovic, “Neutrino mass: From LHC to grand unification,” Riv.Nuovo Cim. 034 (2011) 1–68.
96

Aspects of Neutrino Physics and CP Violation

Enviado por

Dados do documento

Descrição original:

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Aspects of Neutrino Physics and CP Violation

Enviado por

Direitos autorais:

Formatos disponíveis

Aspects of Neutrino Physics and CP Violation

Pedro Miguel Lopes Loreto dos Santos

Dissertação para a obtenção de Grau de Mestre em

Presidente: Prof. Maria Teresa Haderer de la Peña Stadler

Palavras-chave: fı́sica de neutrinos, partı́culas de Majorana, violação CP, leptogénese.

Keywords: neutrino physics, Majorana particles, CP violation, leptogenesis.

4 Minimal extension of the Standard Model of particle physics 23

5 CP violation in the leptonic sector 49

A Some properties of the matrix C 79

B General mass term 81

C The third low energy invariant 83

D Calculating CP violation in Leptogenesis 85

3.1 Neutrino parameters summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5.1 Number of CP violating phases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

D.1 SU (2) × U (1) quantum numbers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

2.1 Difference between massive Dirac and Majorana particles. . . . . . . . . . . . . . . . . . . . . 8

3.1 Neutrino mass patterns as indicated by the current data. . . . . . . . . . . . . . . . . . . . . 16

4.1 Neutrino flavour oscillation in vacuum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

6.1 The diagrams contributing to the CP asymmetry αα . . . . . . . . . . . . . . . . . . . . . . . 67

2.1 Majorana Spinor

2.1.1 Dirac Equation

(i~γ µ ∂µ − mc)ψ(x, t) = 0. (2.1)

γ0† = γ0 , γi† = −γi , (2.5)

γµ0 = Sγµ S −1 (2.6)

2.1.2 Charge conjugation

2.1.3 Lorentz transformation

with Λµν = δ µν + ω µν . A fermion field transforms as follows

S(Λ)−1 γ µ S(Λ) = Λµν γ ν . (2.19)

Cγ0T γµ∗ (Cγ0T )−1 = −γµ . (2.21)

With above equality one obtains

2.1.4 Dirac and Majorana mass terms

Lm = mψψ = m(ψL + ψR )(ψL + ψR ) (2.24)

And to the usual definition of a Majorana spinor2

where both χ and ω are Majorana spinors.

2.1.5 A few words on Majorana mass term

v) Majorana mass terms are n symmetric by definition.

2.2 Discrete transformations

|hf 0 |S|i0 i|2 = |hf |S|ii|2 . (2.36)

where the same operator S is used for the transformed states.

Pψ(x, t)P † = P ψ(x, t). (2.37)

From (2.4) one can take

Pψ(t, ~x)P † = exp(iβp )γ 0 ψ(t, −~x). (2.39)

Pψ(t, ~x)P † = Pψ † (t, ~x)P † γ 0

P 2 ψ(t, ~x)P †2 = exp(2iβp )ψ(t, ~x). (2.41)

The P is an operator in the Hilbert space and is shown to be unitary

PAµ (t, ~x)P † = gµν Aν (t, −~x) (2.43)

Considering the electromagnetic current LA µ

2.2.2 Charge Conjugation C

With the above result, one can also derive

Cψ(t, ~x)C † = exp(−iβc )ψ c (t, ~x). (2.45)

Finally, as in the case of the P operator, the C is also unitary

CAµ (t, ~x)C † = −Aµ (t, ~x). (2.48)

(CP)ψ(t, ~x)(CP)† = exp(iβψ )γ 0 ψ c (t, −~x) (2.49)

(CP)ψ(t, ~x)(CP)† = exp(−iβψ )ψ c (t, −~x)γ 0 (2.51)

(CP)ψ c (t, ~x)(CP)† = − exp(−iβψ )γ 0 ψ (2.53)

(CP)Aµ (t, ~x)(CP)† = −gµν Aν (t, −~x). (2.55)

2.2.5 Time Reversal T

|hi0 |S|f 0 i|2 = |hf |S|ii|2 , (2.56)

T0 γi∗ T0 −1 = −γi , T0 γ0∗ T0 −1 = γ0 . (2.59)

T Aµ (t, ~x)T † = gµν Aν (−t, ~x). (2.62)

2.2.6 CPT Theorem

Figure 3.1: Neutrino

∆m221 [10−5 eV 2 ] 7.59+0.20

6.1 The diagrams contributing to the CP asymmetry αα . . . . . . . . . . . . . . . . . . . . . . . 67