N.

Tuning

Lectures Notes on CP violation

February 2020
ii
Table of contents

Introduction 1

1 CP Violation in the Standard Model 3

1.1 Parity transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
60
1.1.1 The Wu-experiment: Co decay . . . . . . . . . . . . . . . . . . . . 4
1.1.2 Parity violation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.3 CPT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2 C, P and T: Discrete symmetries in Maxwell’s equations . . . . . . . . . . 8
1.3 C, P and T: Discrete symmetries in QED . . . . . . . . . . . . . . . . . . . 9
1.4 CP violation and the Standard Model Lagrangian . . . . . . . . . . . . . . 12
1.4.1 Yukawa couplings and the Origin of Quark Mixing . . . . . . . . . . 12
1.4.2 CP violation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2 The Cabibbo-Kobayashi-Maskawa Matrix 17

2.1 Unitarity Triangle(s) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 Size of matrix elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3 Wolfenstein parameterization . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.4.1 The Lepton Sector . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3 Neutral Meson Decays 29

3.1 Neutral Meson Oscillations . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

iii
iv Table of contents

3.2 The mass and decay matrix . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3.3 Eigenvalues and -vectors of Mass-decay Matrix . . . . . . . . . . . . . . . . 31
3.4 Time evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.5 The Amplitude of the Box diagram . . . . . . . . . . . . . . . . . . . . . . 35
3.6 Meson Decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.7 Classification of CP Violating Effects . . . . . . . . . . . . . . . . . . . . . 40

4 CP violation in the B-system 43

4.1 β: the B 0 → J/ψKS0 decay . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2 βs : the Bs0 → J/ψφ decay . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.3 γ: the Bs0 → Ds± K ∓ decay . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.4 Direct CP violation: the B 0 → π − K + decay . . . . . . . . . . . . . . . . . 53
4.5 CP violation in mixing: the B 0 → l+ νX decay . . . . . . . . . . . . . . . . 54
4.6 Penguin diagram: the B 0 → φKS0 decay . . . . . . . . . . . . . . . . . . . . 55

5 CP violation in the K-system 57

5.1 CP and pions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.2 Description of the K-system . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.3 The Cronin-Fitch experiment . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.3.1 Regeneration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.4 Master Equations in the Kaon System . . . . . . . . . . . . . . . . . . . . 63
5.5 CP violation in mixing: ǫ . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.6 CP violation in decay: ǫ′ . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.7 CP violation in interference . . . . . . . . . . . . . . . . . . . . . . . . . . 67

6 Experimental Aspects and Present Knowledge of Unitarity Triangle 69

6.1 B-meson production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
6.2 Flavour Tagging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.3 Present Knowledge on Unitarity Triangle . . . . . . . . . . . . . . . . . . 73
Table of contents v

6.3.1 Measurement of sin 2β . . . . . . . . . . . . . . . . . . . . . . . . . 74

6.3.2 Measurement of ǫK . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.3.3 |Vub /Vcb | . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.3.4 Measurement of ∆m . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.4 Outlook: the LHCb experiment . . . . . . . . . . . . . . . . . . . . . . . . 78

References 80
Introduction

In these lectures we will introduce the subject of CP violation. This subject is often
referred to with the more general term “Flavour Physics” since all the interesting stuff
concerning CP violation happens in the weak (charged current) interaction when one
quark-flavour changes into another quark-flavour, q → W q ′ , even between different fami-
lies!
The charged current interactions q → W q ′ form a central element in the Standard Model.
Out of the 18 free parameters in the Standard Model, no less than four are related to the
coupling constants of the interaction q → W q ′ . In addition, we will see that the origin of
these coupling constants is closely related (through the Yukawa couplings) to the masses
of the fermions, which form another nine free parameters of the Standard Model. Both the
masses of the fermions and the coupling strength of the charged-current quark-couplings
form an intruiging, hierarchial, pattern for which some underlying mechanism must exist...
The CP operation changes particles into anti-particles, and changes the coupling constant
of q → W q ′ into its complex conjugate. It turns out that not all processes are invariant
under the CP operation and we will show how these complex numbers are determined.
In fact, the observation of CP violation allows us to make a convention-free definition
of matter, with respect to anti-matter 1 ! Maybe not surprising, CP violation is indeed
one of the requirements needed to create a universe that is dominated by matter (or by
anti-matter for that matter...).
Although CP violation was first discovered in the K-system in 1964, in recent years most
experimental and theoretical developments in the field of flavour physics occur in the
B-system and as a result the term “B-physics” is intimately related to flavour physics.
The study of B-mesons and their decays is not only interesting for the above mentioned
reasons. Many observables in B-physics are dominated by higher order diagrams, and
therefore these measurements are extremely sensitive to extra contributions from new,
virtual, heavy particles, such as the supersymmetric partners of the Standard Model
particles.

1
This could be of importance in a telephone call with aliens, before the first hand-shake. If they ask to
meet you, first ask them what the charge of the lepton is to which the neutral kaon preferentially decays.
If that is equal to the charge of the orbiting leptons in atoms, you are in business and can savely fix the
term...

1
2 Table of contents

Very interesting topics such as baryogenesis, sphalerons, the strong CP problem, or neu-
trino oscillations unfortunately fall beyond the scope of these lectures. These lectures will
focus on “normal” CP violation (also known as the Kobayashi Maskawa mechanism, for
which these gentlemen were awarded the Nobel Prize in 2008), and its direct connection
to the Standard Model, see Fig. 1.
The lectures are organized as follows. We start with the Standard Model Lagrangian and
see where the flavour (and even family) changing interactions originate. This leads to the
famous CKM-matrix which is discussed in chapter 2. We continue with the description of
neutral mesons and their decays in chapter 3. This will be of importance for the discussion
of measurements of some important B-decays in chapter 4. The historically important
but less instructive K-system is discussed in chapter 5. We conclude with a discussion on
experimental aspects and the present status of knowledge of CP violation in the Standard
Model.
Most facts in these notes are taken from two excellent books on the topic, Bigi & Sanda [1]
and Branco & Da Silva [2].

Figure 1: “Nature’s grand tapestry”. [1]

Chapter 1

CP Violation in the Standard Model

1.1 Parity transformation

The parity operator, P, inverts all space coordinates used in the description of a physical
process. Consider for instance a scalar wavefunction ψ(x, y, z, t). Performing the parity
operation on this wavefunction will transform it to ψ(−x, −y, −z, t), or
P ψ(x, y, z, t) = ψ(−x, −y, −z, t)
The parity transformation can be viewed as a mirroring with respect to a plane, (for
instance z → −z) followed by a rotation around an axis perpendicular to the plane (the
z-axis). As angular momentum is conserved, physics will be invariant under the rotation
and so the parity operation tests for invariance to mirroring w.r.t. a plane of arbitrary
orientation. Parity conservation or P-symmetry implies that any physical process will
proceed identically when viewed in mirror image. This sounds rather natural. After all
we would not expect a dice for instance to produce a different distribution of numbers if
one swaps the position of the one and the six on the dice.
Up until 1956 the general feeling was that all physical processes would conserve parity. In
this year, however, a number of experiments were performed which showed that at least
for processes involving the weak interaction this was not the case. For both experiments
which will be discussed the properties of the transformation of spin by the parity operation
played a crucial role, so let us consider how spin transforms.
Spin like angular momentum transforms as the cross product of a space vector and a
momentum vector.
~ = ~r × ~p
L
P ~r = −~r
P ~p = −~p
and so
~ =L
PL ~

3
4 Chapter 1 CP Violation in the Standard Model

In other words the parity operation leaves the direction of the spin unchanged. If one can
thus find a process which produces an asymmetric distribution with respect to the spin
direction one proves that P-symmetry is not conserved. Another way of looking at it is
by considering helicity which is the projection of the spin of a particle onto its direction
of motion,
1
h = ~σ  p̂
2
As helicity changes sign under parity transformation (~p → −~p) finding a process which
produces a particle with a prefered helicity also proves that P-symmetry is violated.

60
1.1.1 The Wu-experiment: Co decay

The experiment performed by Wu [3] in 1956 took a 60 Co source and placed it in a

magnetic field. The 60 Co nucleus has spin 5 and becomes polarised along the magnetic
field lines. The experimental aparatus is shown in Fig. 1.1a. The experimental method
(a) (b)

Figure 1.1: (a) The experimental configuration of the Wu experiment. The NaI coun-
ters monitor the state of polarisation by measuring the anisotropy of successive γ emis-
sions produced through the polarisation technique. The anthracene crystal measures the
β-electrons. (b) The result of the Wu experiment. The top plot shows the rate as a
function of time for the two NaI counters, the center shows the degree of polarisation
determined from the anisotropy. The lowest plot shows the measured β counting rates for
positive and negative magnetic field directions.
1.1 Parity transformation 5

A B
_
e ν

60 60 60 60
Co Ni Co Ni
(S=5) (S=4) (S=5) (S=4)

_
ν e

Figure 1.2: The possible transitions of 60 Co with spin 5 to 60 Ni with spin 4. The open
arrows denote the spin. Closed arrows denote the momentum vector. (a) The transition
which is forbidden in nature. (b) The allowed transition. The antineutrino is always
righthanded.

was then to measure the rate of β-electrons from the decay:

60
27 Co →60 −
28 Ni + e + ν̄e

in a small counter placed at small angles with respect to the field lines. By inverting the
magnetic field direction and thus the polarisation of the cobalt nucleus, a difference in
counting rate could be detected, as shown in Fig. 1.1b. Several control counters were also
read out so that the degree of polarisation and the absolute counting rate of the source
could be callibrated. The rate asymmetry shown in Fig. 1.1b was convincing evidence for
the violation of P-symmetry or parity.
It could be explained by the following argument: The transition from 60 Co(spin 5) to
60
Ni(spin 4) as shown in Fig. 1.2a apparently does not occur, but the transition shown
in Fig. 1.2b does. As the electron was known from other experiments to appear in nature
in both helicity states (±1/2), the only remaining conclusion was that the anti-neutrino
occured only in one single helicity state, namely +1/2.

1.1.2 Parity violation

A more elegant experiment was performed a few weeks later by Lederman [4] which
allowed the observation of parity violation in charged pion decay. The experimental setup
is shown in Fig. 1.3a. Charged pions of 85 MeV are created in pp collisions and separated
6 Chapter 1 CP Violation in the Standard Model
(a) (b)

Figure 1.3: (a) The experimental setup of the Lederman experiment. (b) The resulting
rate variation as a function of the applied magnetic field.

magnetically according to their charge. They are then allowed to decay according to

π + → µ+ + νµ

The remaining pions are absorbed. The penetrating muons are stopped in a carbon target
which is placed in a magnetic field, perpendicular to their line of flight. The muons will
start to precess in the magnetic field and after a while decay. The precession frequency
is given by
geB
ωL = (1.1)
2mµ
with B the magnetic field, e the charge of the muon, mµ its mass and g the gyromagnetic
ratio of the muon which for a spin 1/2 particle is approximately 2.
A counter placed at fixed angle w.r.t. the original flight direction is gated open with a
fixed delay after the entry of the muon into the carbon target. This counter detects the
positrons from the decay
µ+ → e+ + νe + ν̄µ
The experiment was repeated for several different settings of the magnetic field and thus
different precession frequency. The resulting rate is shown in Fig. 1.3b. A clear oscillation
is seen showing that the muons are produced with non-zero polarisation in the pion decay.
So also in pion decay parity is not conserved. Again the assumption of a single helicity
for the neutrino can explain the result. As an aside the curves also show an asymmetry
in the height of the oscillation caused by the violation of parity in the muon decay.
Furthermore the wavelength of the oscillation allowed for the first time the measurement
of the gyromagnetic moment of the muon, thus confirming the spin 1/2 nature of the
muon.
Let us now take a closer look at the π decay. Fig. 1.4 shows the effect of the parity
operation on the decay of a π + , which yields an unphysical result. If we now perform
1.1 Parity transformation 7

µ+ νµ ν-µ
π+ P π+ C π-
νµ + -
µ µ

CP
Figure 1.4: The physical π + decay is transformed via the parity operation to an unphysical
decay, the charge conjugation operation transforms this to a physically allowed situation
for π − decay. The solid arrows denote momentum vectors, the open arrows the spin.

a second operation, that of charge conjugation, C, the final result is again a physically
acceptable result. That this is correct could be verified by the Lederman experiment
by the use of π − mesons. So the combined application of the parity operation together
with the operation of charge conjugation (or more precisely particle-antiparticle exchange)
seems at least to provide a symmetry of nature.

1.1.3 CPT

Sofar we have come across two basic symmetries P and C which both are violated maxi-
mally in the weak interaction. The neutrino has only one helicity state. A third symmetry
which many find an appealing symmetry is that of time reversal, T. Certainly there is a
very strong reason for requiring the combination of all three to be a symmetry of nature
as it has been proven that any Lorentz invariant local field theory must have the combined
CPT symmetry. This is such a basic requirement that it is hard to imagine any theory
in particle physics which does not conform to this symmetry. One of the consequences of
the CPT symmetry is that particle states i.e. mass eigenstates which are the solution of

Hψ − mψ = 0 (1.2)

will have an equivalent antiparticle mass eigenstate with the same mass eigenvalue. The
easiest way of conserving the CPT invariance would clearly have been the invariance of
physics to all three symmetries separately. As we have seen P-symmetry and C-symmetry
are both violated but CP seems for the time being a valid symmetry. The notion of time-
reversal invariance is thus closely coupled to that of CP invariance. If CP-invariance is
true then T invariance is also true, if CP symmetry is violated then so must timereversal
invariance be.
The discrete transformations parity (P), charge conjugation (C) and time reversal (T)
will be discussed in more detail in the following sections.
8 Chapter 1 CP Violation in the Standard Model

1.2 C, P and T: Discrete symmetries in Maxwell’s

equations

Consider first how the electric and magnetic fields, currents and charges behave under P,
C and T transformation. Under P transformation positions of charges will be exchanged
and so the electric field will change sign. Currents will flow in opposite direction so they
also will change sign. The magnetic field is proportional to ~j × ~r and so will conserve its
sign:

~ x, t) → P ~ x, t)
E(~ −E(−~
~ x, t) → P ~ x, t)
B(~ B(−~
P
~j(~x, t) → −~j(−~x, t)
P
∇ → −∇

Under T transformation the charges and positions will remain unchanged, whereas the
currents will flow in opposite direction, so we get:

~ x, t) → T ~ x, −t)
E(~ E(~
~ x, t) → T ~ x, −t)
B(~ −B(~
T
~j(~x, t) → −~j(~x, −t)
∂ T ∂
→ −
∂t ∂t

and using similar arguments, we get for the C transformation:

~ x, t) → C ~ x, t)
E(~ −E(~
~ x, t) → C ~ x, t)
B(~ −B(~
C
~j(~x, t) → −~j(~x, t)
C
ρ(~x, t) → −ρ(~x, t)

Finally under the combined CPT transformation the charges and currents change sign
and electric and magnetic field retain their sign. These properties can be summarised in
~
terms of the scalar potential φ and vector potential A:

P
~ x, t) → ~ x, t), A(~ T
~ x, t) → ~ x, −t), A(~ C
~ x, t) → ~ x, t).
A(~ −A(−~ −A(~ −A(~

P T C
φ(~x, t) → φ(−~x, t), φ(~x, t) → φ(~x, −t), φ(~x, t) → −φ(~x, t).
1.3 C, P and T: Discrete symmetries in QED 9

1.3 C, P and T: Discrete symmetries in QED

In this section we will derive expressions for the P, C and T operators. By definition the
transformed states ψP (~x, t), ψC (~x, t) and ψT (~x, t) are constructed such that they satisfy
the same equation of motions for free fields as ψ(~x, t). In the derivation of expressions
for the P, C and T operators we start from the (correct) assumption that electromagnetic
interactions are P, C and T symmetric. In other words, the Dirac equation should also
hold for the P, C and T transformed fields. Eventually we will see what CP invariance
implies for the weak interactions.
Let us consider the Dirac equation of a particle with charge e in an electro-magnetic field
 
µ ∂ µ
iγ − γ eAµ − m ψ(~x, t) = 0, (1.3)
∂xµ

where ψ(~x, t) is a four component spinor and the matrices γ µ are given by:

0 σi 1 0
   
i 0
γ = for i = 1, 3; γ = , with :
−σ i 0 0 −1
       
0 1 0 −i 1 0 1 0
1
σ = ; σ =2
; σ =3
; 1= .
1 0 i 0 0 −1 0 1

We now write out Eq. (1.3) as

     
0 ∂ i ∂
γ i − eφ(~x, t) − γ i − eAi (~x, t) − m ψ(~x, t) = 0 (1.4)
∂t ∂xi
The Dirac equation after parity transformation becomes:
     
0 ∂ i ∂
γ i − eφ(−~x, t) − γ i + eAi (−~x, t) − m ψ(−~x, t) = 0 (1.5)
∂t ∂(−xi )
Now, ψ(−~x, t) is not a solution of the Dirac equation, due to the additional -sign in front
of γi . Multiplying the Dirac equation (after parity transformation) from the left by γ 0 ,
we obtain the Dirac equation again:
     
0 0 ∂ i ∂
γ γ i − eφ(−~x, t) + γ i − eAi (−~x, t) − m ψ(−~x, t) = 0
∂t ∂xi

and then transport the γ 0 through the equation using the anti-commutation rules γ 0 γ i =
−γ i γ 0 for i = 1, 2, 3 we get:
     
0 ∂ i ∂
γ i − eφ(−~x, t) − γ i − eAi (−~x, t) − m γ 0 ψ(−~x, t) = 0 (1.6)
∂t ∂xi

We now see that the spinor γ 0 ψ(−~x, t) obeys the (original) Dirac equation. We come
to the conclusion that the original Dirac equation is obeyed by the simultaneous parity
10 Chapter 1 CP Violation in the Standard Model

transformation in Lorentz space (~x → −~x) and the transformation in Dirac space of the
spinor with γ 0 :
P
ψ(~x, t) −→ ψP (~x, t) = γ 0 ψ(−~x, t) = P ψ(−~x, t)
Of course also eiφ γ 0 ψ(−~x, t), with φ an arbitrary real phase, would provide a valid solution.
We will now take a look at the charge conjugation and investigate the interaction of a
particle of opposite charge with an electro-magnetic field. Starting again from Eq. (1.3)
and exchanging e → −e we find that the charge conjugate wave-function ψC (~x, t) must
satisfy:      
0 ∂ i ∂
γ i + eφ(~x, t) − γ i + eAi (~x, t) − m ψC (~x, t) = 0 (1.7)
∂t ∂xi
For our particular representation of the γ matrices we have the following properties:
γ 0∗ = γ 0 , γ 1∗ = γ 1 , γ 2∗ = −γ 2 and γ 3∗ = γ 3 . Then taking the complex conjugate of
Eq. (1.3) one obtains
      
0 ∂ 1 ∂ 2 ∂
−γ i + eφ(~x, t) + γ i + eA1 (~x, t) − γ i + eA2 (~x, t)
∂t ∂x1 ∂x2
  
3 ∂
+γ i + eA3 (~x, t) − m ψ ∗ (~x, t) = 0 (1.8)
∂x3

Now multiplying from the left with γ 2 and transporting it through the equation we get:
     
0 ∂ i ∂
γ i + eφ(~x, t) − γ i + eAi (~x, t) − m γ 2 ψ ∗ (~x, t) = 0 (1.9)
∂t ∂xi
comparing this result with Eq. (1.7) we can readily identify

ψC (~x, t) = γ 2 ψ ∗ (~x, t)

Again we can use the arbitrary phase which we now take to be i, causing the combination
iγ 2 to be real:
ψC (~x, t) = iγ 2 ψ ∗ (~x, t).
T
Rewriting this expression using ψ ≡ (ψ † γ 0 )T = γ 0T (ψ † )T = γ 0 ψ ∗ yields the widely used
expression for ψC (~x, t):

C T T
ψ(~x, t) −→ ψC (~x, t) = iγ 2 ψ ∗ (~x, t) = iγ 2 γ 0 ψ (~x, t) = Cψ (~x, t)

Similarly, using C = iγ 2 γ 0 , we find C = −C −1 and ψ(~x, t) → −ψ T (~x, t)C −1 .

Finally we take a look at time reversal. We now again start from the complex conjugate
equation and now multiply by γ 1 γ 3 we then get
T
ψ(~x, t) −→ ψT (~x, t) = iγ 1 γ 3 ψ ∗ (~x, −t) = T ψ ∗ (~x, −t)

where we again use the arbitrary phase to give the factor i.

1.3 C, P and T: Discrete symmetries in QED 11

For the CP operation we have:

CP ψ(~x, t) = ieiφ γ 2 γ 0 ψ ∗ (−~x, t)

and for CPT

CP T ψ(~x, t) = eiφ γ 5 ψ(−~x, −t)
using γ 5 = iγ 0 γ 1 γ 2 γ 3 . Check out the fact that the CP operation transforms an electron
into a positron with opposite momentum and opposite helicity.
In summary, we get the following properties of the transformed wave-functions:

Field P C
†
Scalar field φ(~x, t) φ(−~x, t) φ (~x, t)
0 2 0 T
Dirac spinor ψ(~x, t) γ ψ(−~x, t) iγ γ ψ (~x, t)
ψ(~x, t) ψ(−~x, t)γ 0 −ψ T (~x, t)C −1
Axial vector field Aµ (~x, t) −Aµ (−~x, t) A†µ (~x, t)

Table 1.1: C and P transforms of fields. Note that µ = 0, 1, 2, 3 and that Ak = −Ak and
A0 = A0 .

Because of Lorentz invariance, spinors typically occur in so-called bilinear forms in the
Lagrangian. For example, the bilinear ψ 1 γµ ψ2 transforms under C as follows (using
γ µ C = −Cγ µT and γ µ† γ 0 = γ 0 γ µ ) [5]:
C T T
ψ 1 γµ ψ2 −→ −ψ1T C −1 γµ Cψ 2 = ψ1T γµT ψ 2 = −(ψ 2 γµ ψ1 )T = −ψ 2 γµ ψ1 .

The minus-sign at the second step arises from interchanging the (anti-commuting) fermion
fields, and the transpose at the last step can be omitted because the entity is a ’one-by-one
matrix’.
For completenss the transformation properties of the bi-linear forms are listed below.

Bilinear P C T CP CPT
scalar ψ 1 ψ2 ψ 1 ψ2 ψ 2 ψ1 ψ 1 ψ2 ψ 2 ψ1 ψ 2 ψ1
pseudo scalar ψ 1 γ5 ψ2 -ψ1 γ5 ψ2 ψ 2 γ5 ψ1 -ψ 1 γ5 ψ2 -ψ2 γ5 ψ1 ψ 2 γ5 ψ1
vector ψ 1 γµ ψ2 ψ 1 γ µ ψ2 -ψ2 γµ ψ1 ψ 1 γ µ ψ2 -ψ2 γ µ ψ1 -ψ2 γµ ψ1
axial vector ψ 1 γµ γ5 ψ2 -ψ1 γ µ γ5 ψ2 ψ 2 γµ γ5 ψ1 ψ 1 γ µ γ5 ψ2 -ψ2 γ µ γ5 ψ1 -ψ2 γµ γ5 ψ1
tensor ψ 1 σµν ψ2 ψ 1 σ µν ψ2 -ψ2 σµν ψ1 -ψ 1 σ µν ψ2 -ψ2 σ µν ψ1 ψ 2 σµν ψ1

Table 1.2: C, P transforms of bilinears

12 Chapter 1 CP Violation in the Standard Model

1.4 CP violation and the Standard Model Lagrangian

1.4.1 Yukawa couplings and the Origin of Quark Mixing

Let us now have a close look at the Standard Model Lagrangian to see where CP violation
originates. The full Standard Model Lagrangian consists of three parts:

LSM = Lkinetic + LHiggs + LY ukawa .

The kinetic term describes the dynamics of the spinor fields ψ

Lkinetic = iψ̄(∂ µ γµ )ψ,

where ψ̄ ≡ ψ † γ 0 and the spinor fields ψ are the three fermion generations, each consisting
of the following five representations:

QILi (3, 2, +1/6), uIRi (3, 1, +2/3), dIRi (3, 1, −1/3), LILi (1, 2, −1/2), lRi
I
(1, 1, −1)

This notation [6] means that QILi (3, 2, +1/6) is a SU(3)C triplet, left-handed SU(2)L dou-
blet, with hypercharge Y = 1/6. The superscript I implies that the fermion fields are
expressed in the interaction basis. The subscript i stands for the three generations. Ex-
plicitly, QILi (3, 2, +1/6) is a shorthand notation for:
 I I I   I I I   I I I   I I I 
I ug , ur , ub ug , ur , ub cg , cr , cb tg , tr , tb
QLi (3, 2, +1/6) = I I I = I I I , I I I , .
dg , dr , db i dg , dr , db sg , sr , sb bIg , bIr , bIb

The interaction terms are obtained by imposing gauge invariance by replacing the partial
derivative by the covariant derivate

Lkinetic = iψ̄(D µ γµ )ψ (1.10)

with the covariant derivative defined as

D µ = ∂ µ + igs Gµa La + igWbµ σb + ig ′ B µ Y,

with La the Gell-Mann matrices and σb the Pauli matrices. Gµa , Wbµ and B µ are the
eight gluon fields, the three weak interaction bosons and the single hypercharge boson,
respectively.
We can now write out the charged current interaction between the (left-handed!) quarks:
i
Lkinetic,weak (QL ) = iQILi γµ ∂ µ + gWbµ σb QILi


2
 I
µ i µ

u
= i(u d)IiL γµ ∂ + gWb σb
2 d iL
g g
= iuIiL γµ ∂ µ uIiL + idIiL γµ ∂ µ dIiL − √ uIiL γµ W −µ dIiL − √ dIiL γµ W +µ uIiL + ...
2 2
1.4 CP violation and the Standard Model Lagrangian 13

using W + = √1 (W1 − iW2 ) and W − = √1 (W1 + iW2 ).

2 2

Next, the W and Z bosons aquire their mass through the mechanism of spontaneous
symmetry breaking. For this, the Higgs scalar field and her potential is added to the
Lagrangian:
LHiggs = (Dµ φ)† (D µ φ) − µ2 φ† φ − λ(φ† φ)2 (1.11)
with φ an isospin doublet
φ+
 
φ(x) = .
φ0
The coupling of the Higgs to the gauge fields follows from the covariant derivative in the
kinetic term. However, the interactions between the Higgs and the fermions, the so-called
Yukawa couplings, have to be added by hand:
−LY ukawa = Yij ψLi φ ψRj + h.c.
= Yijd QILi φ dIRj + Yiju QILi φ̃ uIRj + Yijl LILi φ lRj
I
+ h.c. (1.12)
with  0 
∗ φ
φ̃ = iσ2 φ = .
−φ−
The matrices Yijd , Yiju and Yijl are arbitrary complex matrices that operate in flavour space,
giving rise to couplings between different families, or quark mixing, and thus to the field
of flavour physics. It is interesting to note how intimately flavour physics is related to the
mass of the fermions, see Section 2.4. Since this is the crucial part of flavour physics, we
spell out the term Yijd QILi φ dIRj explicitly:

φ+
 
Yijd QILi φ dIRj = Yijd (u d)IiL dIRj =
φ
φ+
   +   +  
I I φ I φ
Y11 (u d)L 0 Y12 (u d)L 0 Y13 (u d)L 0
 φ+   φ+   φ+  
   I 
 dR
 φ φ φ 
 Y21 (c s)IL 0 Y 22 (c s) I
L 0 Y 23 (c s) I
L 0
   sIR 
 φ φ φ 
  +   +   +   bIR
 φ φ φ 
Y31 (t b)IL Y32 (t b)IL Y33 (t b)IL
φ0 φ0 φ0

After spontaneous symmetry breaking,

 +   
φ sym.breaking 1 0
φ(x) = −→ √ ,
φ0 2 v + h(x)
the following mass terms for the fermion fields arise:
−Lquarks d I I u I I
Y ukawa = Yij QLi φ dRj + Yij QLi φ̃ uRj + h.c.
v v
= Yijd dILi √ dIRj + Yiju uILi √ uIRj + h.c. + interaction terms
2 2
d I I u I I
= Mij dLi dRj + Mij uLi uRj + h.c. + interaction terms
14 Chapter 1 CP Violation in the Standard Model

The interaction terms of the fermion fields to the Higgs field, q̄qh(x), are omitted.
To obtain proper mass terms, the matrices M d and M u should be diagonalized. We do
this with unitary matrices V d as follows:
d
Mdiag = VLd M d VRd†
u
Mdiag = VLu M d VRu†

Using the requirement that the matrices V are unitary (VLd† VLd = 1) the Lagrangian can
now be expressed as follows:
−Lquarks I d I I u I
Y ukawa = dLi Mij dRj + uLi Mij uRj + h.c. + ...

= dILi VLd† VLd Mijd VRd† VRd dIRj + uILi VLu† VLu Miju VRu† VRu uIRj + h.c. + ...
= dLi (Mijd )diag dRj + uLi (Miju )diag uRj + h.c. + ...
where the matrices V are absorbed in the quark states, resulting in the following quark
mass eigenstates:
dLi = (VLd )ij dILj dRi = (VRd )ij dIRj
uLi = (VLu )ij uILj uRi = (VRu )ij uIRj
Note that we can thus express the quark states as interaction eigenstates dI , uI or as
quark mass eigenstates d, u.
If we now express the Lagrangian in terms of the quark mass eigenstates d, u instead of
the weak interaction eigenstates dI , uI , the price to pay is that the quark mixing between
families (i.e. the off-diagonal elements) appears in the charged current interaction:
g g
Lkinetic,cc(QL ) = √ uIiL γµ W −µ dIiL + √ dIiL γµ W +µ uIiL + ...
2 2
g g
= √ uiL (VL VL )ij γµ W diL + √ diL (VLd VLu† )ij γµ W +µ uiL + ...
u d† −µ
2 2
The unitary 3×3 matrix
VCKM = (VLu VLd† )ij (1.13)
is the Cabibbo-Kobayashi-Maskawa (CKM) mixing matrix [7].
By convention, the interaction eigenstates and the mass eigenstates are chosen to be equal
for the up-type quarks, whereas the down-type quarks are chosen to be rotated, going
from the interaction basis to the mass basis:
uIi = uj
dIi = VCKM dj
or explicitly:
dI
    
Vud Vus Vub d
 sI  =  Vcd Vcs Vcb   s  (1.14)
bI Vtd Vts Vtb b
1.4 CP violation and the Standard Model Lagrangian 15

The connection between the charged current couplings and the quark masses will be
discussed further in Section 2.4.
From the definition of VCKM , see Eq. (1.13), follows that the transition from a down type
quark to an up-type quark is described by Vud , whereas the transition from an up type
∗
quark to a down-type quark is described by Vud :

u d u d

Vud Vud
∗
Vud
∗
Vud
d u d u

W− W+ W+ W−

Figure 1.5: The definition of Vij and Vij∗ . Note that if the arrow of time points from left
to right, that the two right diagrams represent the situation for anti-quarks.

1.4.2 CP violation

CP violation shows up in the complex Yukawa couplings. We examine once more the
Yukawa part of the Lagrangian:

−LY ukawa = Yij ψLi φ ψRj + h.c.

= Yij ψLi φ ψRj + Yij∗ ψRj φ† ψLi

The CP operation transforms the spinor fields as follows:

CP (ψLi φ ψRj ) = ψRj φ† ψLi

So, LY ukawa remains unchanged under the CP operation if Yij = Yij∗ .

Similarly, if we look at the charged current coupling in the basis of quark mass eigenstates,
g g
Lkinetic,cc(QL ) = √ uiL Vij γµ W −µ diL + √ diL Vij∗ γµ W +µ uiL (1.15)
2 2
and the CP-transformed expression,
g g
LCP
kinetic,cc (QL ) = √ diL Vij γµ W
+µ
uiL + √ uiL Vij∗ γµ W −µ diL (1.16)
2 2
then we can conclude that the Lagrangian is unchanged if Vij = Vij∗ .
The complex nature of the CKM matrix is the origin of CP violation in the Standard
Model. In the following chapter the properties of the CKM mixing matrix will be examined
in detail.
16 Chapter 1 CP Violation in the Standard Model
Chapter 2

The Cabibbo-Kobayashi-Maskawa
Matrix

In the previous chapter we saw how the introduction of Yukawa couplings (i.e. the terms
where the Higgs couples to the fermions) led to off-diagonal elements in the 3×3 matrix be-
tween the different families. By diagonalizing the 3 × 3 Yukawa matrix, these off-diagonal
elements appear in the charged current coupling, in the Cabibbo-Kobayashi-Maskawa-
matrix. The CKM-mechanism is the origin of CP violation, and earned Kobayashi and
Maskawa the Nobel price in 2008, “for the discovery of the origin of the broken symmetry
which predicts the existence of at least three families of quarks in nature”.

2.1 Unitarity Triangle(s)

In this section we will discuss the properties of the unitary 1 CKM matrix VCKM . We
start by counting the number of free parameters for the CKM-matrix.

1) A general n × n complex matrix has n2 complex elements, and thus 2n2 real param-
eters.

2) Unitarity (V † V = 1) implies n2 constraints:

– n unitary conditions (unity of the diagonal elements);

– n2 − n orthogonality relations (vanishing off-diagonal elements).
u d
3) The phases of the quarks can be rotated freely: uLi → eiφi uLi and dLj → eiφi dLj .
Since the overall phase is irrelevant, 2n − 1 relative quark phases can be removed.
1
Remember from quantum mechanics the evolution of a wave function, |ψ(t)i = U (t)|ψ(0)i. The
unitarity condition implies conservation of probability: hψ(t)|ψ(t)i = hψ(0)|U † U |ψ(0)i = hψ(0)|ψ(0)i,
provided U † U = 1

17
18 Chapter 2 The Cabibbo-Kobayashi-Maskawa Matrix

Summarizing, the CKM-matrix describing the flavour couplings of n generations of up

and down type quarks has 2n2 − n2 − (2n − 1) = (n − 1)2 free parameters. Subsequently,
we can divide these free parameters into Euler angles and phases:

4) A general n × n orthogonal matrix can be constructed from 12 n(n − 1) angles de-

scribing the rotations among the n dimensions.

5) The remaining free parameters are the phases: (n−1)2 − 12 n(n−1) = 21 (n−1)(n−2).

For the Standard Model with three generations we find three Euler angles and one complex
phase.
At this point we make a short historical excursion. Before the third family was known,
Cabibbo suggested in 1963 the mixing between d and s quarks, by introducing the Cabibbo
mixing angle θC . This is the only free parameter for a 2×2 unitary matrix, and the
mixing matrix is a pure real matrix. To allow for CP violation the mixing matrix has
to contain complex elements, satisfying Vij 6= Vij∗ . This requires at least three families.
CP violation was first measured in 1964 by Cronin and Fitch (discussed in more detail
in Section 5.3). Subsequently, Kobayashi and Maskawa suggested in 1973 the possibility
that the existence of a third family could explain the CP violation within the Standard
Model. This happened at the time that not even the second family was completed! The 4th
quark, the charm quark was only discovered a year later, in 1974, in the form of the J/ψ
resonance. The bottom and the top quark were discovered in 1977 and 1994 respectively.
In 2008 Kobayashi and Maskawa were awarded the Nobel prize for the discovery of the
origin of the broken symmetry which predicts the existence of at least three families of
quarks in nature.
Let us now look at the consequences of the unitarity condition for the CKM-matrix:
 ∗
Vud Vcd∗ Vtd∗
   
Vud Vus Vub 1 0 0
V † V = V V † =  Vcd Vcs Vcb   Vus ∗
Vcs∗ Vts∗  =  0 1 0  (2.1)
∗ ∗ ∗
Vtd Vts Vtb Vub Vcb Vtb 0 0 1

This leads to the following three unitary relations:

∗ ∗ ∗
Vud Vud + Vus Vus + Vub Vub = 1
∗ ∗ ∗
Vcd Vcd + Vcs Vcs + Vcb Vcb = 1
Vtd Vtd∗ + Vts Vts∗ + Vtb Vtb∗ = 1 (2.2)

These relations express the so-called weak universality, because it shows that the squared
sum of the coupling strengths of the u-quark to the d, s and b-quarks is equal to the
overall charged coupling of the c-quark (and the t-quark). In addition, we see that this
sum adds up to 1, meaning that “there is no probability remaning” to couple to a 4th
down-type quark. Obviously, this relation deserves continuous experimental scrutiny.
2.1 Unitarity Triangle(s) 19

The remaining relations are known as the orthogonality conditions:

Vud Vcd∗ + Vus Vcs∗ + Vub Vcb∗ = 0

Vud Vtd∗ + Vus Vts∗ + Vub Vtb∗ = 0
∗ ∗
Vcd Vud + Vcs Vus + VcbVub∗ = 0
Vcd Vtd∗ + Vcs Vts∗ + VcbVtb∗ = 0
∗ ∗
Vtd Vud + Vts Vus + Vtb Vub∗ = 0
Vtd Vcd∗ + Vts Vcs∗ + Vtb Vcb∗ = 0 (2.3)

Three of the six equations are simply the complex conjugate version. An additional three
interesting equations arise from the unitarity relation V † V = 1:
∗
Vud Vus + Vcd∗ Vcs + Vtd∗ Vts = 0
∗
Vud Vub + Vcd∗ Vcb + Vtd∗ Vtb = 0
∗
Vus Vud + Vcs∗ Vcd + Vts∗ Vtd = 0
∗
Vus Vub + Vcs∗ Vcb + Vts∗ Vtb = 0
Vub∗ Vud + Vcb∗ Vcd + Vtb∗ Vtd = 0
∗
Vub Vus + Vcb∗ Vcs + Vtb∗ Vts = 0 (2.4)

Equations (2.3-2.4) give relations in which the complex phase is present. As these are
sums of three complex numbers that must yield zero they can be viewed as a triangle in
the complex plane.
In the literature there are many different parameterizations of the CKM matrix. A con-
venient representation uses the Euler angles θij with i, j denoting the family labels. With
the notation cij = cos θij and sij = sin θij the following parameterization was introduced
by Chau and Keung, and has been adopted by the Particle Data Group:

0 s13 e−iδ13
   
c12 s12 0 c13 1 0 0
VCKM =  −s12 c12 0   0 1 0   0 c23 s23  =
iδ13
0 0 1 −s13 e 0 c13 0 −s23 c23
s13 e−iδ13
 
c12 c13 s12 c13
 −s12 c23 − c12 s23 s13 eiδ13 c12 c23 − s12 s23 s13 eiδ13 s23 c13  (2.5)
iδ13 iδ13
s12 s23 − c12 c23 s13 e −c12 s23 − s12 c23 s13 e c23 c13

The phase can be made to appear in many elements, and is chosen here to appear in the
matrix describing the relation between the 1st and 3rd family.
20 Chapter 2 The Cabibbo-Kobayashi-Maskawa Matrix

2.2 Size of matrix elements

We will now briefly discuss the experimental evidence for the size of the matrix elements
of the CKM-matrix.

|Vud |: is determined from comparing nuclear β-decay rates or neutron decay rates to the µ-
decay rate, see Fig. 2.1. In the calculations there are some theoretical uncertainties
due to binding energy corrections in nuclei. The best value obtained by averaging
many experiments is:
|Vud | = 0.97420 ± 0.00021
νe νe

e− e−
W− W−
Vud 1
µ− νµ
d u
n p
d d
u u

Figure 2.1: Diagrams important for determining Vud .

|Vus |: is obtained by analysing semi-leptonic K-decays, shown in Fig. 2.2:

|Vus | = 0.2243 ± 0.0005
νe
νe

e− e−
W− W−
Vus 1
µ− νµ
s u 0
K −
π
u u

Figure 2.2: Diagrams important for determining Vus .

|Vcd |: was originally obtained by the analysis of neutrino and anti-neutrino induced charm-
particle production of the valence d-quark in a neutron (or proton) (see Fig. 2.3).
Averaged with measurements on semileptonic charm decays, and also fully leptonic
decays like D + → µ+ ν, yields
|Vcd | = 0.218 ± 0.004
 2.2 Size of matrix elements 21
µ −
νe
νµ
e−
W−
W 1
Vcd νµ
µ−
d
n d
d

u u

c
Figure 2.3: Diagrams important for determining Vcd .

|Vcs |: is the matrix element relevant for the dominant decay modes of the charm quark.
Here an analogous analysis is performed for semileptonic D-decays, or fully leptonic
Ds -decays (see Fig. 2.4). The final result is

|Vcs | = 0.997 ± 0.017

νe
νe

e+ e−
W+ W−
Vcs
∗
1
µ− νµ
0 c s
D̄ K −

u u

Figure 2.4: Diagrams important for determining Vcs .

|Vcb |: is determined from the exclusive B → D̄ (∗) µ+ νµ and inclusive B → D̄ (∗) Xµ+ νµ
decays (see Fig. 2.5). A large amount of data is available on these decays both
from LEP and from the lower energy e+ e− experiments BaBar and Belle, giving an
average result of
|Vcb | = 0.0422 ± 0.0008

|Vub |: is determined from the semi-leptonic exclusive B → πµ+ νµ and inclusive B →

Xu µ+ νµ decays , similar to the determination of |Vcb|.

|Vub | = 0.00394 ± 0.00036

|Vtd | and |Vts |: These elements cannot be measured from tree-level top-quark decays, and so these
elements are probed through loop diagrams such as the box-diagram, as will be
22 Chapter 2 The Cabibbo-Kobayashi-Maskawa Matrix

νe
νe

e+ e−
+
W W−
Vcb∗ 1
µ− νµ
b c
B0 D∗−

d d

Figure 2.5: Diagrams important for determining Vcb.

discussed in detail in Section 3.5. Using lattice calculations to take long-distance

effects into account, and assuming |Vtb | = 1, yields:

|Vtd | = 0.0081 ± 0.0005

|Vts | = 0.0394 ± 0.0023

|Vtb | : CDF, D0, ATLAS and CMS measured the ratio of branching ratios Br(t → W b)/Br(t →
W q), yielding the following 95% confidence level limit:

|Vtb | = 1.019 ± 0.025

Taking all the information above, a global fit with Standard Model constraints leads to
the following result for the absolute values of the elements:
   
0.97446 0.22452 0.00365 0.00010 0.00044 0.00012
VCKM =  0.22438 0.97359 0.04214 ±   0.00044 0.00011 0.00076  (2.6)
0.00896 0.04133 0.99911 0.00024 0.00974 0.00003

The strength of the charged current couplings seem to exhibit a hierarchy. This pattern
motivated Wolfenstein
q [8] to parametrize the CKM-matrix in powers of the parameter
md
λ ≈ sin θ12 ≈ ms , which is described in the next section.

1 λ λ3
 

|VCKM | ∼  λ 1 λ2 
λ3 λ2 1
2.3 Wolfenstein parameterization 23

2.3 Wolfenstein parameterization

Comparing the expressions (2.5) and (2.6) we see that typically the sin θij are small
numbers and that sin θ12 ≫ sin θ23 ≫ sin θ13 . This leads to a very popular approximate
parameterization of the CKM matrix proposed by Wolfenstein.
sin θ12 = λ (2.7)
sin θ23 = Aλ2 (2.8)
sin θ13 e−iδ13 = Aλ3 (ρ − iη) (2.9)
where A, ρ and η are numbers of order unity. The CKM matrix then becomes O(λ3 ):
1 − 12 λ2 Aλ3 (ρ − iη)
 
λ
VCKM =  −λ 1 − 21 λ2 Aλ2  + δV (2.10)
3 2
Aλ (1 − ρ − iη) −Aλ 1
The higher order terms in the Wolfenstein parametrization are of particular importance
for the Bs -system, as we will see in chapter 4, because the phase in |Vts | is only apparent
at O(λ4 ):
− 81 λ4
 
0 0
δV =  21 A2 λ5 (1 − 2(ρ + iη)) − 81 λ4 (1 + 4A2 ) 0  + O(λ6 ) (2.11)
1 5 1 4 1 2 4
2
Aλ (ρ + iη) 2
Aλ (1 − 2(ρ + iη)) − 2 A λ

Let us now return to the six orthogonality relations that give rise to the six unitarity
triangles. Only two out of the six equations have terms with equal powers in λ.
Vud Vub∗ + Vcd Vcb∗ + Vtd Vtb∗ = 0
O(λ3 ) O(λ3 ) O(λ3 ) (2.12)

∗ ∗ ∗
Vtd Vud + Vts Vus + Vtb Vub =0
3 3 3
O(λ ) O(λ ) O(λ ) (2.13)

These two triangles are relevant for B-decays. The other four equations contain terms
with different powers of λ and hence give rise to “squashed” triangles.
The relation shown in Eq. 2.12 is known as the unitarity triangle. By dividing the
three sides by |Vcd Vcb | and subsequently rotating the whole triangle (i.e. rephasing all
sides, without affecting the relative phases), yields the famous unitarity triangle shown
in Fig. 2.6. One side now has unit length and points along the real axis. The apex of the
triangle is located by definition at (ρ, η) 2 :
Vud Vub∗
ρ + iη ≡ .
Vcd Vcb∗
2
Occasionally the generalized parameters ρ and η are defined in the literature as the approximation
ρ ≡ ρ(1 − 12 λ2 ) and η ≡ η(1 − 21 λ2 ) [9].
 24 Chapter 2 The Cabibbo-Kobayashi-Maskawa Matrix

Im VudV*ub + Vcd Vcb* + Vtd V*tb = 0 Im VusV*ub + Vcs Vcb* + Vts V*tb = 0

(ρ,η) 0
Vtd V*tb VusV*ub π−γ βs 1 Re
VudV*ub α Vcs Vcb* Vts V*tb
Vcd Vcb*
Vcd Vcb*
Vcs Vcb*
γ β
0 1 Re

Figure 2.6: (a) “The” unitarity triangle. Shown in the complex plane is the relation
1 + Vtd Vtb∗ /Vcd Vcb∗ + Vud Vub∗ /Vcd Vcb∗ = 0. (b) The analogous unitarity triangle for the Bs0 -
system, with the d-quark replaced by the s-quark, 1 + Vts Vtb∗ /Vcs Vcb∗ + Vus Vub∗
/Vcs Vcb∗ = 0.

The parameters ρ, and η can be expressed in terms of the Wolfenstein parameters ρ and
η as follows:
1 1
ρ = ρ(1 − λ2 ) + O(λ4 ) η = η(1 − λ2 ) + O(λ4 ) (2.14)
2 2

The angles in “the” unitarity triangle are defined as follows:

Vtd Vtb∗ Vcd Vcb∗ ∗
Vts Vtb∗
       
Vud Vub
α ≡ arg − ∗
β ≡ arg − γ ≡ arg − βs ≡ arg −
Vud Vub Vtd Vtb∗ Vcd Vcb∗ Vcs Vcb∗
(2.15)
Note that these definitions are convention independent: any phase added to a specific
quark cancels out in either the product or the ratio of the CKM-elements. Equivalently,
the CKM triangles can be rotated and scaled in the complex plane, without affecting the
internal angles of the triangles.
In the Wolfenstein parametrization a phase convention is used such that the elements
Vtd , Vub and Vts have an imaginary component (to order O(λ4 )), and Vcd Vcb∗ is real and
negative, see Fig. 2.7.

β Im γ Im βs Im

*
VudVub
π π arg Vts
arg Vub* = γ
Vcd Vcb* Vcd Vcb* βs

β Re Re VcsVcb* Re
arg Vtd Vts Vtb*

Vtd Vtb*

Figure 2.7: The angles β, γ and βs using the phase convention as given by the Wolfenstein
parameterization. (a) β (b) γ (c) βs .
2.3 Wolfenstein parameterization 25

The expressions for the angles now become:

β ≈ π + arg(Vcd Vcb∗ ) − arg(Vtd Vtb∗ ) = π + π − arg(Vtd ) = − arg(Vtd )

∗
γ ≈ π + arg(Vud Vub ) − arg(Vcd Vcb∗ ) = π − arg(Vub ) − π = − arg(Vub )
βs ≈ π + arg(Vts Vtb∗ ) − arg(Vcs Vcb∗ ) = π + arg(Vts ) − 0 = arg(Vts ) + π

Alternatively, the Wolfenstein phase convention in the CKM-matrix elements can be

shown as:
|Vub |e−iγ
 
|Vud | |Vus |
VCKM,Wolfenstein =  −|Vcd | |Vcs | |Vcb |  + O(λ5 ) (2.16)
−iβ iβs
|Vtd |e −|Vts |e |Vtb |

As mentioned earlier, CP violation requires Vij 6= Vij∗ , which is satisfied if the triangle
has a finite surface in the complex plane. In fact, it turns out that the surface of all six
unitarity triangles have equal surface area.
This quantity denoted as J, also known as the Jarlskog invariant, can be derived in a
simple way from the CKM matrix. Remove one column and one row from the CKM
matrix and take the product of the diagonal elements with the complex conjugate of the
non-diagonal elements. The imaginary part of the product is then equal to J. In total
there will be nine possible expressions for J which all give the same result:

J = ℑ(V11 V22 V12∗ V21∗ ) = ℑ(V22 V33 V23∗ V32∗ ) = .... (2.17)

In the Wolfenstein parameterization the quantity J becomes

J = A2 λ6 η = 2 × area (2.18)

In the parameterization of Eq. (2.5) it is

J = c12 c213 c23 s12 s13 s23 sin δ13 (2.19)

From this form it is clear why this quantity occurs in all CP violation effects. It is zero if
any one of the mixing angles is zero. This would reduce the CKM matrix essentially to
a 2 × 2 matrix and allow the removal of the phase. Also if the complex phase would be
zero no CP violation is possible. As a final comment the quantity J is just equal to the
twice the surface area of the unitarity triangle.
26 Chapter 2 The Cabibbo-Kobayashi-Maskawa Matrix

2.4 Discussion

The strong hierarchy in the size of the matrix elements of the quark mixing matrix is
intriguing and its origin is not understood. To paraphrase Ikaros Bigi [10]: “ It has to
contain a message from nature - albeit in a highly encoded form.”
We have seen that the origin of the quark mixing matrix lies in the Yukawa couplings
between the Higgs field and the quark fields. At the same time, these Yukawa couplings
are responsible for the generation of the quark masses, which becomes obvious after
diagonalizing the matrix that describes the Yukawa couplings. Also the values of the quark
masses show a striking hierarchy, which makes the thought of an underlying connection
between the quark masses and the charged current quark couplings fascinating.

Yukawa Couplings

Couplings Masses
d s b

u u d
??
c c s

t t b

Figure 2.8: Both the charged current quark couplings and the quark masses originate from
the Yukawa couplings and both the couplings and the masses show an intriguing hierarchy.
Does this suggest an underlying connection between them?
.

We have now set the framework for the incorporation of CP violation in the Standard
Model. The question remains of course whether all manifestations of CP violation can be
explained. Of course theoretically we can always incorporate new ideas such as supersym-
metry or an increase in the number of families to explain any deviations. Experimentally
it is now important to verify the Standard Model description. When looking at the uni-
tarity triangle we can see that the length of the sides of the triangle can be extracted
from measurable quantities. It is now necessary to investigate whether the angles of the
triangle can be measured in an independent way. Disagreement between the angles and
the lengths of the side would necessarily signal New Physics. At present many experi-
ments are either running or have been proposed which will be able to give answers to the
questions to a greater or lesser extent. In chapter 4 we will proceed to discuss the channels
which are considered to be the prime candidates for further investigation of CP violation.
Before that, we will introduce the concept of neutral meson oscillations, or mixing, which
plays a crucial role in many of the CP-measurements.
2.4 Discussion 27

2.4.1 The Lepton Sector

We only focussed on the quark couplings, and we will continue to do so in the rest of
these notes. Nevertheless it is both enlightening and intriguing to cast some light on the
lepton sector.
The discovery of neutrino oscillations [11] implies that neutrinos have non-zero mass, and
as a result a similar diagonalization of the Yukawa matrix can be done, compared to
the quarks (see Section 1.4.1). The lepton counterpart of the CKM-matrix is called the
PMNS-matrix, after Maki, Nakagawa, Sakata and Pontecorvo [12].
The first observation is that the leptons are commonly referred to as the flavour eigen-
states, in contrast to the mass eigenstates that we use for the quarks. For example, we
typically picture the W to couple purely to a (e, νe ) pair, whereas the coupling of the W
to the quarks we picture as the coupling to a (u, [d, s, b]) pair, ie. a mixture of d, s and b
quarks. The lepton-equivalent of the down-type mass eigenstates are ν1 , ν2 and ν3 .
The second, inspiring, observation is that the magnitude of the elements of the MNSP-
matrix show a completely different hierarchy [13]:
   
Ue1 Ue2 Ue3 0.82 0.55 0.15
UM N SP =  Uµ1 Uµ2 Uµ3  ≈  0.37 0.57 0.70  .
Uτ 1 Uτ 2 Uτ 3 0.39 0.59 0.69

Interesting numerology appears if we square the matrix elements, revealing the following
approximate composition (known as ’tri-bimaximal mixing’ [14]):
 2 1 
3 3
0
|UM N SP |2 ≈  16 31 21  ,
1 1 1
6 3 2

or alternatively:
 q q 
2 1
 q q   
2
0 1   1 q0 q0 0
 1 0 0 
3 3  q q 3 3 
1 1 
q
0
 q q
1 1
UM N SP ≈ − 1 0  0 1 0  q2 q 2  =  − 6 − 12 .
 
 2     
3 3 3
0 0 1
 
0 − 12 1
q q q
0 0 1 2 − 16 1
3
1
2

This comparison should make clear that the hierarchy in the CKM matrix, nor the fact
that the matrix is symmetric, is by any means “logical”, or “natural”?!
To date, no experiment has reached the sensitivity to measure complex phases on the
MNSP matrix elements, which would indicate CP violation in the lepton sector 3 .
3
The situation is slightly more complex if the neutrino’s are of Majorana nature, ie. if the neutrinos
are their own anti-particles. The smallness of the neutrino masses is typically explained with the see-saw
mechanism, which at the same time predicts a heavy right-handed sterile neutrino at the grand-unification
scale.
28 Chapter 2 The Cabibbo-Kobayashi-Maskawa Matrix
Chapter 3

Neutral Meson Decays

3.1 Neutral Meson Oscillations

The phenomenon of neutral meson oscillations is important for various reasons. Firstly, in
many measurements of CKM-parameters, the oscillations play a crucial role in providing
a second transition amplitude from the initial state to a given final state. This second
amplitude is needed to determine the relative phase difference between two amplitudes,
as described in chapter 4. Secondly, the observation of two K 0 particles with largely
varying lifetimes and the resulting discovery of CP violation is of historical importance,
see chapter 5, and is described in terms of a superposition of |Ki-states and its quantum-
mechanical evolution.
The formalism described in this section is valid for all weakly decaying neutral mesons:
K 0 , D 0 , B 0 and Bs0 . We will outline the framework in terms of a generical meson P 0 ,
which can be substituted at will by K 0 , D 0 , B 0 or Bs0 . Although we will see that the
difference in mass (and thus available phase space for the final state) and coupling strength
(CKM-elements) results in dramatically different phenomenology.

3.2 The mass and decay matrix

The states |P 0 i and |P̄ 0i which are eigenstates of the strong and electromagnetic interac-
tions with common mass m0 and opposite flavour content. Let us consider an arbitrary
superposition of the P 0 and P̄ 0 states, which has time-dependent coefficients a(t) and
b(t) respectively:
ψ(t) = a(t)|P 0 i + b(t)|P̄ 0 i
We can write ψ(t) in the subspace of P 0 and P̄ 0 as follows
 
a(t)
ψ(t) =
b(t)

29
30 Chapter 3 Neutral Meson Decays

The effective Hamiltonian that governs the time evolution is a sum of the strong, electro-
magnetic and weak Hamiltonians.

H = Hst + Hem + Hwk

The wavefunction ψ must then obey

∂ψ
i = Hψ
∂t
The Hamiltonian can then, in the (P 0 , P̄ 0) basis, be written as 2 × 2 complex matrix:
i
H =M− Γ
2
where both M and Γ are Hermitian matrices. M will provide a “mass” term and due to
the −i, Γ will provide the exponential decay. Note that due to the i, H is not hermitian
reflected in the property that the probability to observe either P 0 or P̄ 0 is not conserved,
but goes down with time:
  
d 2 2

∗ ∗ Γ11 0 a(t)
|a(t)| + |b(t)| = − (a(t) b(t) )
dt 0 Γ22 b(t)
If the weak part of the Hamiltonian did not exist the P system would be stable and so H
would reduce to  
mP 0 0
H→M =
0 mP̄ 0
where mP 0 = hP 0|Hst + Hem |P 0 i and mP̄ 0 = hP̄ 0|Hst + Hem |P̄ 0 i and the off-diagonal
elements are 0 through flavour conservation. With the weak interaction responsible for
the decay we get:
M11 − 2i Γ11
 
∂ψ i 0
i = Hψ = (M − Γ)ψ = ψ
∂t 2 0 M22 − 2i Γ22

If we now allow for the transitions P 0 → P̄ 0 , the off-diagonal elements are introduced:
M11 − 2i Γ11 M12 − 2i Γ12
 
∂ψ i
i = Hψ = (M − Γ)ψ = ψ
∂t 2 M21 − 2i Γ21 M22 − 2i Γ22

The off-diagonal elements consist of two parts, M12 and 12 Γ12 , which describe different
ways of the P 0 → P̄ 0 transition. M12 quantifies the short-distance contribution from
the (calculable) box diagram as will be discussed in Section 3.5. Γ12 is a measure of the
contribution from the virtual, intermediate, decays to a state f , see Fig. 3.1.
∗
If we now assume that CPT is valid then it follows that M11 = M22 , M21 = M12 and
∗
Γ11 = Γ22 , Γ21 = Γ12 meaning that mass and total decay width of particle and antiparticle
are identical.
M − 2i Γ M12 − 2i Γ12
 
∂ψ i
i = Hψ = (M − Γ)ψ = ψ (3.1)
∂t 2 ∗
M12 − 2i Γ∗12 M − 2i Γ
3.3 Eigenvalues and -vectors of Mass-decay Matrix 31

via off−shell states,

weak box−diagram
M 12

P0 P0
−i
Γ12
2
via on−shell states,
P0 f P0

Figure 3.1: The neutral meson oscillation consists of two contributions, namely through
off-shell states and on-shell states.
.

In general there can be a relative phase between Γ12 and M12 [15]:
 M 
12
φ = arg − (3.2)
Γ12

which is the relative phase difference between the on-shell (or dispersive) and off-shell (or
absorbative) transition. This leads to the relations

∆m = 2|M12 | (3.3)
∆Γ = 2|Γ12 | cos φ. (3.4)

If T is conserved then it follows that Γ∗12 /Γ12 = M12

∗
/M12 so that by introducing a free
phase we can make Γ12 and M12 real.
Under these assumptions we can now find the eigenvalues and eigenvectors of the Hamil-
tonian. These will describe the masses and decay widths and the P 0 , P̄ 0 superpositions,
that describe the physical particles.

3.3 Eigenvalues and -vectors of Mass-decay Matrix

Given the Schrödinger equation (3.1) we find the eigenvalues of the mass-decay matrix,
by solving the determinantal equation [16]:

 M − i Γ − λ M12 − i Γ12 
 
 2 2 =0
 M12∗
− 2i Γ∗12 M − 2i Γ − λ 
q
Using the shorthand notation F = (M12 − 2i Γ12 )(M12
∗
− 2i Γ∗12 ) we find the eigenvalues
λ± = M − 2i Γ ± F . Splitting the real and imaginary part by defining λ− = m1 + 2i Γ1 and
32 Chapter 3 Neutral Meson Decays

λ+ = m2 + 2i Γ2 , we obtain:

i i
m1 + Γ1 = M − ℜF − (Γ − 2ℑF )
2 2
i i
m2 + Γ2 = M + ℜF − (Γ + 2ℑF )
2 2
These expressions invite the use of the following notation:

∆m ≡ m2 − m1 = 2ℜ F
∆Γ ≡ Γ1 − Γ2 = 4ℑ F

If we express the eigenstates P1 and P2 as:

|P1 i = p|P 0i − q|P̄ 0i

|P2 i = p|P 0i + q|P̄ 0i

we find p and q by solving

M − 2i Γ M12 − 2i Γ12
    
p p
= λ±
∗
M12 − 2i Γ∗12 M − 2i Γ q q

yielding:
s
q ∗
M12 − 2i Γ∗12
=±
p M12 − 2i Γ12

The state |P1 i is the mass eigenstate with mass m1 and lifetime Γ1 . Similarly we obtain
the mass m2 and lifetime Γ2 for state |P2 i. The sign of q/p determines whether |P1 i or
|P2 i is heavier. The choice of a positive value of ∆m gives:
s
q ∗
M12 − 2i Γ∗12
= (3.5)
p M12 − 2i Γ12

Note that we have chosen the sign here, such that ∆m > 0, but that does not imply
anything for the sign of ∆Γ: experiment has to judge whether ∆Γ is positive or negative,
relative to the sign of ∆m.
We can also relate q/p to the mixing phase as introduced in Eq.(3.2) [15]:

|Γ12 | ∆Γ  |q| 
sin φ = tan φ = 2 1 − . (3.6)
|M12 | ∆m |p|

(This will turn out to be the size of a possible CP asymmetry for flavour-specific final
states, af s .)
3.4 Time evolution 33

3.4 Time evolution

We define the two mass eigenstates of the neutral mesons as 1 :

|PH i = p|P 0 i + q|P̄ 0i

|PL i = p|P 0 i − q|P̄ 0 i (3.7)

where the subscripts 1 and 2 are replaced by H and L, indicating the heavy and light
mass eigenstate, respectively. We can then decompose the P 0 and P̄ 0 states as
1
|P 0i = [|PH i + |PL i]
2p
1
|P̄ 0i = [|PH i − |PL i] (3.8)
2q

The states |PH i and |PL i are mass eigenstates and from the Schrödinger equation (with
diagonal Hamiltonian) the usual time dependent wave functions are obtained:
1
|PH (t)i = e−imH t− 2 ΓH t |PH (0)i
1
|PL (t)i = e−imL t− 2 ΓL t |PL (0)i (3.9)

By combining Eqs. (3.9), (3.8) and (3.7) we get:

1 n −imH t− 1 ΓH t 1
o
|P 0 (t)i = e 2 |PH (0)i + e−imL t− 2 ΓL t |PL (0)i
2p
1 n −imH t− 1 ΓH t 1
o
= e 2 (p|P 0 i + q|P̄ 0i) + e−imL t− 2 ΓL t (p|P 0 i − q|P̄ 0 i)
2p
1  −imH t− 1 ΓH t 1
 q  −imH t− 1 ΓH t 1

= e 2 + e−imL t− 2 ΓL t |P 0 i + e 2 − e−imL t− 2 ΓL t |P̄ 0 i
2 2p
 
q
= g+ (t)|P 0 i + g− (t)|P̄ 0 i (3.10)
p
where we define the functions
1  −imH t− 1 ΓH t 1
 1 −iM t  −i 1 ∆mt− 1 ΓH t 1 1

g+ (t) = e 2 + e−imL t− 2 ΓL t = e e 2 2 + e+i 2 ∆mt− 2 ΓL t
2 2
1  −imH t− 1 ΓH t 1
 1 −iM t  −i 1 ∆mt− 1 ΓH t 1 1

g− (t) = e 2 − e−imL t− 2 ΓL t = e e 2 2 − e+i 2 ∆mt− 2 ΓL t
2 2
1
There are some subtleties concerning the sign (or phase) convention. Let us assume CP symmetry,
|q/p| = 1. We can choose q/p = ±1 and CP |P 0 i = ±|P̄ 0 i. Once the sign of q/p is fixed, see Eq.(3.5),
experiment decides if PH is the state that is (more) even or odd, which fixes CP |P 0 i = ±|P̄ 0 i. In principle
this can be different for K 0 , B 0 and Bs0 . We choose the sign convention ∆mK > 0 and CP |K 0 i = −|K̄ 0 i
such that CP |KL i = −|KL i (or ∆ΓK = ΓS − ΓL > 0) according to experiment. This leads to the sign
convention in Eq.(3.7), and implies ∆mK = mL − mS . Also in the B-system the heavier mass eigenstate
BH is (more) CP odd, and the CP-even state in the Bs -system can decay to the final state Ds+ Ds− , and
has therefore a slightly shorter lifetime.
34 Chapter 3 Neutral Meson Decays

where M = (mH + mL )/2 and ∆m = mH − mL . Likewise, we get for the time evolution
of the state |P̄ 0i:  
0 p
|P̄ (t)i = g− (t) |P 0i + g+ (t)|P̄ 0i (3.11)
q

If we start from a pure sample of |P 0 i particles (e.g. produced by the strong interaction)
then we can calculate the probability of measuring the state |P̄ 0 i at time t:
 2
0 0 2 2
p
|hP̄ |P (t)i| = |g− (t)|  
q

with
1 −ΓH t
|g± (t)|2 = + e−ΓL t ± e−Γt (e−i∆mt + e+i∆mt )


e
4
1 −ΓH t
+ e−ΓL t ± 2e−Γt cos ∆mt


= e
4
e−Γt
 
1
= cosh ∆Γt ± cos ∆mt (3.12)
2 2

where Γ = (ΓL + ΓH )/2 and ∆Γ = ΓH − ΓL . Here we see that Γ fulfills the natural role of
decay constant, Γ = 1/τ , justifying the choice of 12 in the hamiltonian in Eq. (3.1). The
sign of ∆m is by definition positive, but the sign of ∆Γ has to be determined experimen-
tally.
3.5 The Amplitude of the Box diagram 35

3.5 The Amplitude of the Box diagram

The short distance contribution to the P 0 ↔ P̄ 0 transitions of neutral meson oscillations

is described by ∆m and can be represented by a Feynman diagram known as the box
diagram, and can be calculated in perturbation theory.
In this section we will calculate the value of ∆m by studying this so-called box diagram.
We will investigate the process of K 0 ↔ K̄ 0 using the CKM matrix. To describe mixing
between a K 0 which has strangeness S = 1 and a K̄ 0 which has S = −1 we must introduce
an amplitude which creates a ∆S = 2 transition. This must necessarily be a second order
weak interaction. The transition necessary for mixing is shown in Fig. 3.2. The calculation
of the box diagram is quite complicated but we will illustrate some of the features in the
calculation of the KL0 − KS0 mass difference.
The mass difference is given by

∆m = mKL0 − mKS0 = hKL0 |H|KL0 i − hKS0 |H|KS0 i (3.13)

As we saw in the previous section, the mass eigenstates can be expressed as a linear
combination of the flavour eigenstates. The amplitude hK 0 |H|K̄ 0i can now be calculated
via the box diagram of Fig. 3.2. As an example we use the Feynman rules to derive an
expression for the amplitude where both the intermediate quarks are u quarks:
 4
−igw ∗ ∗
Muu = i √ (Vus Vud Vus Vud )
2 2
d4 k −ig λσ − k λ k σ /m2W −ig αρ − k α k ρ /m2W
Z   

(2π)4 k 2 − m2W k 2 − m2W

  
5 k / + mu 5 5 k/ + mu 5
ūs γλ (1 − γ ) 2 γρ (1 − γ )ud v̄s γα (1 − γ ) 2 γσ (1 − γ )vd
k − m2u k − m2u

Here we readily recognise the weak coupling constant to the fourth power, the CKM matrix
elements for the vertices, the W propagator terms, the quark and anti-quark spinors and
the factors for the intermediate fermion lines.

Vus,cs,ts
∗
Vud,cd,td Vus,cs,ts
∗
Vud,cd,td
s u, c, t d s W d
0 u, c, t u, c, t 0
K0 W W K K0 K
d u, c, t s d W s
Vud,cd,td Vus,cs,ts
∗
Vud,cd,td Vus,cs,ts
∗

Figure 3.2: Box diagrams responsible for K 0 → K̄ 0 mixing.

36 Chapter 3 Neutral Meson Decays

Taking the sum of all amplitudes with all possible intermediate quark lines we get an
amplitude which is proportional to (assuming k 2 ≪ m2W ).
 ∗ 2
Vus Vud Vcs∗ Vcd Vts∗ Vtd
Z
4
M ∝ d k kµ kν + + 2 (3.14)
k 2 − m2u k 2 − m2c k − m2t
∗
Which with the aid of the equation Vus Vud + Vcs∗ Vcd + Vts∗ Vtd = 0 we can rewrite as
    2
1 1 1 1
Z
4 ∗ ∗
M ∝ d k kµ kν Vcs Vcd 2 − 2 + Vts Vtd 2 − 2
k − m2c k − m2u k − m2t k − m2u
This then finally leads to an answer that has three terms [17], one term depending on
m2c /m2W , one term depending on m2t /m2W and and a term which has a complicated depen-
dence on both m2c /m2W and m2t /m2W . The magnitude of the so-called Inami-Lim factor
these three terms is listed in Table 3.1, together with the size of the CKM-elements
involved in the box diagram.
This calculation only takes into account the quark level transitions and so the full calcu-
lation must take into account the transition from K 0 → ds̄ and gluonic corrections and
colour factors. Because |Vtd Vts | << |VcdVcs | the charm contribution in the loop dominates,
and the final answer becomes:
G2F m2W 2
h
2 2 2
i
∆mK = ηQCD BK fK mK S0 (mc /mW )|Vcd Vcs | (3.15)
6π 2
where GF is the Fermi coupling constant, ηQCD is the QCD correction (≈ 0.85), B andfK2
is the “bag-factor” and the decay constant, respectively, which describe the effect of the
transition from bound to free quarks and Vij are the CKM matrix elements.
In the B-system we have |Vtd Vtb | ∼ |Vcd Vcb |, but because mt >> mc now the top contri-
bution in the loop dominates. By replacing the internal charm quark with the top quark,
and replacing the strange flavour by the bottom quark we find for the B-system:

G2F m2W 2
h
2 2 2
i
∆mB = ηQCD BB fB mB S 0 (mt /mW )|V td V tb | (3.16)
6π 2

Internal Inami-Lim CKM factor

quarks factor K0 B0 Bs0
c, c 3.5 10−4 λ2 (2.7 10−2 ) A2 λ6 (7.4 10−5 ) A2 λ4 (1.4 10−3 )
c, t 3.0 10−3 A2 λ6 |1 − ρ − iη| (8.8 10−6 ) A2 λ6 |1 − ρ − iη| (7.3 10−5 ) A2 λ4 (1.5 10−3 )
t, t 2.5 A4 λ10 |1 − ρ − iη|2 (1.1 10−7 ) A2 λ6 |1 − ρ − iη|2 (7.2 10−5 ) A2 λ4 (1.5 10−3 )

Table 3.1: The magnitude of the three terms contributing to the box diagram, expressed sep-
arately for the Inami-Lim factors (depending on m2q /m2W ) and for the CKM elements [18].
Clearly, the charm-quark contribution dominates in the K-system, where the CKM-factor
compensates for the small Inami-Lim factor. In the B-systems the top-quark contribution
dominates.
3.5 The Amplitude of the Box diagram 37
K0 D0
Probability

Probability
1 1
P(K (t)→ K 0(t)) P(D (t)→ D0(t))
0 0

0 0
P(K (t)→ K (t)) P(D (t)→ D (t))
0 0
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
t (ns) t (ps)

B0 0
Bs
Probability

Probability
1 1
P(B s (t)→ B0s (t))
0
P(B (t)→ B0(t))
0

0 0
P(B s (t)→ B s (t))
0
P(B (t)→ B (t))
0
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
t (ps) t (ps)

Figure 3.3: If one starts with a pure P 0 -meson beam the probability
to observe a P 0 or a
P̄ 0-meson at time t is shown, Prob(t) = e 2 cosh 12 ∆Γt ± cos ∆mt .
−Γt

At this point we can see how the neutral mesons K 0 , D 0 , B 0 and Bs0 in reality oscillate and
what the differences are. As mentioned earlier, the oscillations consist of two components,
M12 and 12 Γ12 . As a general rule, all possible quark exchanges contribute to M12 , but only
actual final states contribute to Γ12 [16]. The short-distance, off-shell contribution from
M12 depends on the size of the CKM-elements at the corners of the box-diagram, and on
the mass of the particles in the box. In the case of D 0 -mixing, the mass of the heaviest
down-type quark in the box, mb is not large enough to compensate the suppression of
the CKM-elements |Vub Vcb|. As a result, the light quarks dominate the short-range D 0 -
mixing and proceeds proportional to ∼ |Vus Vcs |2 m2s ∼ λ2 m2s . As a consequence, the
mixing parameters are expected to be small, and the D-mesons decay before they have
the chance to oscillate.
The oscillation probability of D-mesons is clearly suppressed compared to B 0 -mixing,
see Fig. 3.3, which is proportional to ∼ |Vtb Vtd |2 m2t ∼ λ6 m2t . Bs0 -mixing on the other
hand, is more pronounced, see Fig. 3.3d), due to the magnitude of Vts : ∼ |Vtb Vts |2 m2t ∼
λ4 m2t . Finally, K 0 -oscillation is dominated by the (light) charm quark in the loop, ∼
|VcdVcs |2 m2c ∼ λ2 m2c . However, the kaons profit from the fact that their lifetime is much
higher compared to the B-mesons. Note that the sum of the B 0 and B̄ 0 distributions in
Fig. 3.3 give a perfect exponential decay, because the mass eigenstates BH and BL happen
to have equal lifetimes, ∆Γ = 0. In contrast, the sum of the K 0 and K̄ 0 distributions
results in the sum of two exponential distributions, corresponding to the KS and KL with
short and long lifetime, respectively.
38 Chapter 3 Neutral Meson Decays

Often the dimensionless variables x and y are used to express the mixing behaviour,
expressing the oscillation rate relative to the lifetime:
∆m ∆Γ
x= y=
Γ 2Γ
The oscillation parameters of the various neutral mesons are summarized in Table 3.2.

τ = 1/Γ ∆m x y
K-system 0.26 × 10−9 s 1 5.29 ns−1 0.477 -1
D-system 0.41 × 10−12 s 0.0019 ps−1 0.0046 0.0062
B-system 1.52 × 10−12 s 0.507 ps−1 0.769 0.0005 2
Bs -system 1.53 × 10−12 s 17.76 ps−1 26.8 0.068 2

Table 3.2: Oscillation parameters of the various neutral mesons.

1
Note that the average lifetime Γ is not a very meaningful quantity in the K-system due to the large
difference between the lifetimes of the two mass-eigenstates Ks and KL .
2
These numbers are theoretical values, rather than experimental measurements. The transition
T (Bs0 → D̄s Ds → B̄s0 ) is the largest contribution and proceeds proportional to |Vcb |2 . ∆ΓB 0 < ∆ΓBs0
because the transitions T (B 0 → (D̄D), (ππ), (Dπ) → B̄ 0 ) are all Cabibbo suppressed.
3.6 Meson Decays 39

3.6 Meson Decays

In this section we extend the formalism of neutral meson oscillations, and include the
subsequent decay of the meson to a final state f . We consider the following four decay
amplitudes

A(f ) = hf |T |P 0i Ā(f ) = hf |T |P̄ 0i

A(f¯) = hf¯|T |P 0i Ā(f¯) = hf¯|T |P̄ 0i

and define the complex parameter λf (not be confused with the Wolfenstein parame-
ter λ !):

q Āf 1 q Āf¯ 1
λf = , λ̄f = , λf¯ = , λ̄f¯ = (3.17)
p Af λf p Af¯ λf¯

The general expression for the time dependent decay rates, ΓP 0 →f (t) = |hf |T |P 0(t)i|2 ,
give us the probability that the state P 0 at t = 0 decays to the final state f at time t,
and can now be constructed as follows, using Eqs. (3.10) and (3.11):

ΓP 0 →f (t) = |Af |2 |g+ (t)|2 + |λf |2 |g− (t)|2 + 2ℜ[λf g+

∗


(t)g− (t)]
 2
q
ΓP 0 →f¯(t) = |Āf¯|2   |g− (t)|2 + |λ̄f¯|2 |g+ (t)|2 + 2ℜ[λ̄f¯g+ (t)g−∗


(t)]
p
 2
p
ΓP̄ 0 →f (t) = |Af |2   |g− (t)|2 + |λf |2 |g+ (t)|2 + 2ℜ[λf g+ (t)g− ∗


(t)]
q
2 ∗
|g+ (t)|2 + |λ̄f¯|2 |g− (t)|2 + 2ℜ[λ̄f¯g+


ΓP̄ 0 →f¯(t) = |Āf¯| (t)g− (t)] (3.18)

with

e−Γt
 
2 1
|g± (t)| = cosh ∆Γt ± cos ∆mt
2 2
−Γt
 
∗ e 1
g+ (t)g− (t) = sinh ∆Γt + i sin ∆mt
2 2
−Γt
 
∗ e 1
g+ (t)g− (t) = sinh ∆Γt − i sin ∆mt (3.19)
2 2

The terms proportional |A|2 are associated with decays that occurred without oscillation,
whereas the terms proportional to |A|2 (q/p)2 or |A|2 (p/q)2 are associated with decays
following a net oscillation. The third terms, proportional to ℜg ∗ g, are associated to the
interference between the two cases.
Combining Eqs. (3.18) and (3.19) results in the following expressions for the decay rates
40 Chapter 3 Neutral Meson Decays

for neutral mesons, also known as the master equations:

e−Γt
ΓP 0 →f (t) = |Af |2
 2 
2 1 1 2
(1 + |λf | ) cosh ∆Γt + 2ℜλf sinh ∆Γt + (1 − |λf | ) cos ∆mt − 2ℑλf sin ∆mt
2 2
 2 −Γt
p e
ΓP̄ 0 →f (t) = |Af |2   (3.20)
q 2
 
2 1 1 2
(1 + |λf | ) cosh ∆Γt + 2ℜλf sinh ∆Γt − (1 − |λf | ) cos ∆mt + 2ℑλf sin ∆mt
2 2
The sinh- and sin-terms are associated to the interference between the decays with and
without oscillation. Commonly, the master equations are expressed as:
−Γt
 
2 2 e 1 1
ΓP 0 →f (t) = |Af | (1 + |λf | ) cosh ∆Γt + Df sinh ∆Γt + Cf cos ∆mt − Sf sin ∆mt
2 2 2
 2 −Γt
 
2 p 2 e 1 1
 
ΓP̄ 0 →f (t) = |Af |   (1 + |λf | ) cosh ∆Γt + Df sinh ∆Γt − Cf cos ∆mt + Sf sin ∆mt
q 2 2 2
(3.21)
with
2ℜλf 1 − |λf |2 2ℑλf
Df = 2
C f = 2
Sf = . (3.22)
1 + |λf | 1 + |λf | 1 + |λf |2
For a given final state f we therefore only have to find the expression for λf to fully
describe the decay of the (oscillating) mesons. Examples of some final states will be
presented in chapter 4.

3.7 Classification of CP Violating Effects

The following classification between the various types of CP violation can be made [6].

1) CP violation in decay. This type of CP violation occurs when the decay rate of
a B to a final state f differs from the decay rate of an anti-B to the CP-conjugated
final state f¯:
Γ(P 0 → f ) 6= Γ(P̄ 0 → f¯)
This is obviously satisfied (see Eq. (3.18)) when
 
 Āf¯ 
  6= 1. (3.23)
 Af 

An example of CP violation in decay for neutral mesons is decay B 0 → K + π − . A

sizeable CP-asymmetry has been observed
ΓB0 →K +π− − ΓB̄0 →K − π+
ACP = <0
ΓB0 →K +π− + ΓB̄0 →K − π+
3.7 Classification of CP Violating Effects 41

In charged mesons there is no mixing, so this is the only type of CP violation that
can occur in charged meson decays.

2) CP violation in mixing. This implies that the oscillation from meson to anti-
meson is different from the oscillation from anti-meson to meson:

Prob(P 0 → P̄ 0 ) 6= Prob(P̄ 0 → P 0 )

Experimentally this is searched for in the semi-leptonic decay of both the B̄ 0 and
the B 0 , coherently produced through Υ → B̄ 0 B 0 . The b̄-quark inside the B 0 -meson
decays weakly to a positively charged lepton, and vice versa. So, an event with two
leptons with equal charge in the final state means that one of the two B-mesons
oscillated. So, the asymmetry in the number of two positive and two negative leptons
allows us to compare the oscillation rates.

N++ − N−− |p/q|2 − |q/p|2

ACP = =
N++ + N−− |p/q|2 + |q/p|2

This type is violated if  

q 
  6= 1. (3.24)
p

In the B 0 - and Bs0 -system this is not the case, so |q/p| ≈ 1 both within the exper-
imental accuracy and theoretical expectation, but we will see that this type of CP
violation is active in the K-system, see chapter 5 2 .

3) CP violation in interference between a decay with and without mixing,

sometimes referred to as CP violation involving oscillations. This form of CP viola-
tion is measured in decays to a final state that is common for the B 0 and B̄ 0 -meson.
An interesting category are CP-eigenstates, f = f¯ (an example of a non-CP eigen-
state are the final states Ds± K ∓ in the Bs0 -system). CP is violated if the following
condition is satisfied:

Γ(P 0(❀P̄ 0 ) → f )(t) 6= Γ(P̄ 0 (❀P 0 ) → f )(t)

A direct consequence of f = f¯ is that there will be two amplitudes that contribute

to the transition amplitude from the initial state |B 0 i to a final state f , namely
A(B 0 → f ) and A(B 0 → B̄ 0 → f ). If we consider the case that |q/p| = 1, the
following expression is obtained, using Eqs. (3.21):

ΓP 0 (t)→f − ΓP̄ 0 (t)→f 2Cf cos ∆mt − 2Sf sin ∆mt

ACP (t) = = (3.25)
ΓP 0 (t)→f + ΓP̄ 0 (t)→f 2 cosh 12 ∆Γt + 2Df sinh 12 ∆Γt
2
p
p Normally expressed in terms of ǫ, for historical reasons: p = (1 + ǫ)/ 2(1 + |ǫ|2 ), q = (1 −
ǫ)/ 2(1 + |ǫ|2 ) and thus q/p = (1 − ǫ)/(1 + ǫ). The parameters p and q are normalized such that
|p|2 + |q|2 = 1.
42 Chapter 3 Neutral Meson Decays

This simplifies considerably if the transition is dominated by only one amplitude, i.e.
assuming that |Af | = |Āf | (or |λf | = 1), so that Df = ℜλf , Cf = 0 and Sf = ℑλf :

−ℑλf sin ∆mt

ACP (t) = 1 (3.26)
cosh 2
+ ℜλf sinh 21 ∆Γt
∆Γt

We conclude that CP violation can even occur when both |q/p| = 1 and |A(f )| =
|Āf |, namely when the following condition is satisfied:
 
q Āf
ℑλf = ℑ 6= 0 (3.27)
p Af

Commonly an alternative classification of direct and indirect CP violation is made [6].

Direct CP violation is defined as |A(f )| = 6 |Āf |. In terms of the above categories, direct
CP violation obviously appears in the CP violation in decay. In addition, the term direct
CP violation is used for the situation where Cf 6= 0, probed by the first term in Eq. (3.25),
since |A(f )| =
6 |Āf | → |λf | =
6 1 → Cf 6= 0. Indirect CP violation is the type of CP
violation that involves mixing in any way, either through |q/p| = 6 1 or via the second term
of Eq. (3.25). Historically this distinction originates from so-called superweak models that
predicted CP violation to appear only in mixing diagrams. The discovery of direct CP
violation excluded these superweak models.
Finally, we comment on the relative size of CP violation in the interference of mixing
and decay in the K and B system. The difference arises from the CKM-factor of the
box-diagram. The real part of the CKM-factor in the K-system is given by:

(Vcd Vcs∗ )2 = λ2

The imaginary part is proportional to A2 λ6 η. Therefore, we expect for the ratio of the
CP violating part to the CP non-violating part of ∆mK to be
ℑ∆m
∝ A2 λ4 η (3.28)
ℜ∆m
In the B system the CKM factor is given by

(Vtd Vtb∗ )2 = (1 − ρ − iη)2 A2 λ6

from which we can deduce that the ratio of CP violation to CP non-violation in the B
system is
ℑ∆m η(1 − ρ)
∝ (3.29)
|∆m| (1 − ρ)2 + η 2
In the B system we then have the strength of CP violation of the same order as CP
non-violation, whereas in the K system it is suppressed by a factor of λ4 ≈ 2 10−3 .
Chapter 4

CP violation in the B-system

In the previous chapter we have identified where CP violation occurs in the general for-
malism of meson decays, and classified the various categories. In the coming sections we
will investigate a few special decays with which CP violation is measured and the phases
of the CKM elements are determined [9].
Remember the Wolfenstein parametrization, Eq. (2.16), since it so widely used. This pa-
rameterization is very convenient to localize weak phase differences in Feynman diagrams:

|Vub |e−iγ
 
|Vud | |Vus |
VCKM,Wolfenstein =  −|Vcd | |Vcs | |Vcb |  + O(λ5 ) (4.1)
−iβ iβs
|Vtd |e −|Vts |e |Vtb |

In this chapter we will see how the angles β, βs and γ can be determined.
At first sight it might be remarkable that complex phases can be observed, because the
complex phase in an amplitude A = |A|eiϕ disappears in the expectation value, AA† =
|A|2ei(ϕ−ϕ) = |A|2 . However, several decay amplitudes Ai = |Ai |eiϕi might contribute to
the total amplitude A [19]. Each phase consists of a CP-odd phase φi originating from
complex coupling constants, and a CP-even phase δi , typically originating from gluon
exchange in the final state (and strong interactions are CP-conserving!). Therefore we
have for the CP-conjugated amplitude Āi = |Ai |ei(−φi +δi ) . Now we can calculate the
difference in the magnitude of the total amplitude |A(a → b)| and the CP-conjugate
|Ā(ā → b̄)|:

|A|2 = |A1 + A2 |2 = |A1 |2 + |A2 |2 + |A1 A2 | ei((φ1 +δ1 )−(φ2 +δ2 )) + ei(−(φ1 +δ1 )+(φ2 +δ2 ))

= |A1 |2 + |A2 |2 + 2|A1 A2 | cos(∆φ + ∆δ)

|Ā|2 = |Ā1 + Ā2 |2 = |A1 |2 + |A2 |2 + |A1 A2 | ei((−φ1 +δ1 )−(−φ2 +δ2 )) + ei(−(−φ1 +δ1 )+(−φ2 +δ2 ))

= |A1 |2 + |A2 |2 + 2|A1 A2 | cos(−∆φ + ∆δ)

An explicit example will be shown in Section 4.1.

43
44 Chapter 4 CP violation in the B-system

4.1 β: the B 0 → J/ψKS0 decay

In the case of decays into CP eigenstates (i.e. |f¯i = CP|f i = ηf |f i, with ηf = ±1)
only two independent amplitudes need to be considered: Af and Āf . We define the CP
asymmetry as (see Eq. (3.25)):

ΓB0 (t)→f − ΓB̄0 (t)→f

ACP (t) =
ΓB0 (t)→f + ΓB̄0 (t)→f

Let us now concentrate on specific decays to get an idea where the CKM phase enters the
asymmetry measurement. We start with the decay B 0 → J/ψKS0 [20] and will investigate
Eq. (3.26) further.
The first observation is that at the quark level the B 0 decay and the B̄ 0 decay have a
different final state, B 0 → J/ψK 0 and B̄ 0 → J/ψ K̄ 0 . As a result, we need to consider
the mass eigenstates in the K system, see Eq. (3.7), to obtain the same final state f
for the B 0 and B̄ 0 decay: |KS0 i = p|K 0 i + q|K̄ 0 i. (The details of the K-system will be
discussed in chapter 5.) Secondly, in the B 0 -system ∆Γ ≈ 0 (see Table 3.2), so Eq. (3.26)
can simply be written as:
ACP (t) = −ℑλf sin(∆mt) (4.2)

For a given final state f , the magnitude and phase of λf fully describe the decay and
oscillation of the B 0 and the B̄ 0 -meson. (If the final state is not a CP-eigenstate, we will
also need λf¯.) Starting from the definition of λf we write

q   ĀJ/ψKS0   q   Ā
J/ψK̄ 0
 p
λJ/ψKS0 = ηJ/ψKS0 =− (4.3)
p B0 AJ/ψKS0 p B0 AJ/ψK 0 q K0

The three parts in this equation correspond to the mixing of the B 0 -meson, (q/p)B0 , the
decay of the B 0 or B̄ 0 , Ā/A, and the mixing of the K 0 -meson, (q/p)K 0 . These three parts
are diagrammatically shown in Fig. 4.1. The factor ηJ/ψKS0 accounts for the CP-eigenvalue
of the final state. The J/ψ has spin-1 and is CP-even, while KS has spin-0 and is (almost)
CP-even. The B 0 is spin-0, and thus the particles in the final state must have a relative
angular momentum l = 1. As a result the final state J/ψKS0 is CP-odd, ηJ/ψKS0 = −1 1 .

The B 0 ↔ B̄ 0 mixing is induced by the box diagram shown in Fig. 4.1a). We have seen
that the mass matrix element M12 ∝ Vtb∗ Vtd Vtb∗ Vtd , (see Eq. (3.16)), and that we can neglect
the term Γ12 , see Table 3.2, and therefore:
r
∗
V ∗ Vtd
 
q M12
= = tb ∗ (4.4)
p B0 M12 Vtb Vtd

1
The analogous measurement can be performed with the decay B 0 → J/ψKL0 , with ηJ/ψKL0 = +1.
4.1 β: the B 0 → J/ψKS0 decay 45
c

J/ψ
Vtd u, c, t b Vbc
d c
Vcs
∗

B0 W W W−
Vtd
b W s, d
u, c, t d W

KS0

d, s

Figure 4.1: The diagrams that enter in the phase of the decay B 0 → J/ψKS0 , B 0 mixing,
B 0 decay and, K mixing.
.

For the ratio of the decay amplitudes we find on inspection of the diagram in Fig. 4.1b)
that
Vcb V ∗
 
Ā
= ∗ cs
A Vcb Vcs
At this point we have however not produced a KS0 but either a K 0 or a K̄ 0 to finally make
a prediction of the CP violation in the decay B 0 → J/ψKS0 we also have to take into
account the K 0 ↔ K̄ 0 mixing. This adds a factor in analogy to Eq. (4.4) (see Fig. 4.1c):
s
Vcs Vcd∗
 
p M12
= ∗
=
q K M12 Vcs∗ Vcd

Taking everything together we find for the parameter λJ/ψKS0 :

 V ∗V   V V ∗   V V ∗  Vtb∗ Vtd Vcb Vcd∗
td cb cs cs cd
λJ/ψKS0 = − tb ∗ = − (4.5)
Vtb Vtd Vcb∗ Vcs Vcs∗ Vcd Vtb Vtd∗ Vcb∗ Vcd

and for its imaginary part

 ∗
Vtb Vtd Vcb Vcd∗ Vcb Vcd∗
    
ℑλJ/ψKS0 = − sin arg = − sin 2 arg ≡ sin 2β, (4.6)
Vtb Vtd∗ Vcb∗ Vcd Vtb Vtd∗
or
SJ/ψKS0 = −ηJ/ψKS0 sin 2β, (4.7)
where β is defined as in Eq. (2.15). In short we can also write λJ/ψKS0 = −e−2iβ .

To recapitulate, the CP-asymmetry of the decay B 0 → J/ψKS0 is given by the imaginary

part of λJ/ψKS0 :
ACP, B0 →J/ψKS0 (t) = − sin 2β sin(∆mt) (4.8)
46 Chapter 4 CP violation in the B-system

Using the Wolfenstein parameterization we see that the CKM-element Vtd is the only
component with a non-vanishing imaginary part, leading to Eq. (4.6). We conclude that
the CP-asymmetry in the decay B 0 → J/ψKS0 arises from the phase difference of the
amplitudes B 0 → J/ψKS0 and B 0 → B̄ 0 → J/ψKS0 . The phase difference arises from the
CKM-elements Vtd (in the Wolfenstein parametrization) originating from the box-diagram
that is responsible for the B̄ 0 ↔ B 0 oscillations.
The value of sin 2β has been determined very accurately by the BaBar and Belle experi-
ments with the process e+ e− → Υ → B 0 B̄ 0 . A remarkable feature of this process is that
the B 0 B̄ 0 -pair is produced coherently, which means that the B 0 -clock only starts ticking
when the B̄ 0 has decayed. The lifetime of the B 0 -meson is thus expressed as the time
diffrence between the two decays, ∆t. The number of decaying B 0 -mesons is determined
by requiring that the other B had decayed as a B̄ 0 . This number is called the number of
tagged B 0 -mesons, NB0 . The asymmetry is given by:
ΓB0 (∆t)→f − ΓB̄0 (∆t)→f Γtag=B̄0 − Γtag=B0
ACP (∆t) = = = ηf sin 2β sin ∆m∆t.
ΓB0 (∆t)→f + ΓB̄0 (∆t)→f Γtag=B̄0 + Γtag=B0

After correcting for imperfect tagging, we see that the amplitude of the asymetry gives
the value of sin 2β. The present world average is: sin 2β = 0.68 ± 0.03.
4.1 β: the B 0 → J/ψKS0 decay 47

200 a)
Evts. / 0.4 ps

B 0 tags

B 0 tags

-5 0 5

0.5 b)
Raw asym.

-0.5
-5 0 5

c)
Evts. / 0.4 ps

100 B0 tags

B 0 tags

-5 0 5

0.5 d)
Raw asym.

-0.5
-5 0 5
∆t (ps)
Figure 4.2: Number of ηf = −1 candidates (mainly J/ψKS0 ) in the signal region with a B 0
tag NB0 and with a B̄ 0 tag NB̄0 , and b) the raw asymmetry (note the inverted convention
here, compared to our notes...) ACP (t) = (NB̄0 − NB0 )/(NB̄0 + NB0 ), as functions of ∆t.
Figs. c) and d) are the corresponding plots for the ηf = +1 mode J/ψKL0 . The shaded
regions represent the estimated background contributions. From [21].
.
48 Chapter 4 CP violation in the B-system

Measuring complex numbers

Before we continue, we can reflect on the principle behind the measurement of the complex
phase β. Let us show once more how the complex phase appears as an observable, starting
from the |B 0 i wave function and the two decay amplitudes. Remember our wave function
of the decaying, oscillating neutral meson, Eq. (3.10):
     
0 0 q 0 −iM t−i 21 ∆Γt ∆mt 0 ∆mt q 0
|B (t)i = g+ (t)|B i+ g− (t)|B̄ i = e cos |B i + i sin |B̄ i
p 2 2 p
     
0 p 0 0 −iM t−i 12 ∆Γt ∆mt p 0 ∆mt 0
|B̄ (t)i = g− (t) |B i+g+ (t)|B̄ i = e i sin |B i + cos |B̄ i
q 2 q 2
again using ∆Γ ≈ 0 for the B 0 case, see Table 3.2. The factor pq accounts for the B 0 → B̄ 0
oscillation. We saw that for the B 0 -mesons holds | pq | = 1, more specifically, qp = e−i2β .
How is this phase factor e−i2β measurable, in general? If we would measure the number
of B 0 -mesons (i.e. produced as a B 0 ) and compare that to the number of B̄ 0 -mesons
π
(i.e. produced as a B̄ 0 ) at time t = 2∆m , then both the unoscillated and the oscillated
amplitudes are of equal magnitude, and the CP asymmetry can be written as:
 π  |1 + ie−i2β |2 − |ie+i2β + 1|2
ACP t= = = sin 2β (4.9)
2∆m |1 + ie−i2β |2 + |ie+i2β + 1|2

The situation is schematically shown in Fig. 4.3. The total amplitude of the CP-conjugated
situation will have a different magnitude if there are two phases, of which one flips sign
under CP transformation!
c c

J/ψ J/ψ
b Vbc
∗
Vbc Vtd∗ u, c, t
b c d c
Vcs
∗
Vcs
B̄ 0 W− B̄ 0 W W W+
Vtd∗
W
d W s, d b s, d
W u, c, t d W

KS0 KS0

d, s d, s

c c

J/ψ J/ψ
Vbc∗ Vtd u, c, t b Vbc
b c d c
Vcs Vcs
∗
+ W
B0 W B0 W W
−

Vtd
d W s, d b W s, d
W u, c, t d W

KS0 KS0

d, s d, s

Figure 4.3: Adding two amplitudes results in a Atot with different magnitude under CP.
.
4.2 βs : the Bs0 → J/ψφ decay 49

4.2 βs : the Bs0 → J/ψφ decay

The decay Bs0 → J/ψφ is the Bs0 analogue of the decay B 0 → J/ψKS0 , with the spectator
d-quark replaced by an s-quark. However, there are four major differences:

I Vts vs Vtd . Since the spectator d-quark is replaced by an s-quark, the CKM-
element responsible for the CP-asymmetry (in the Wolfenstein parameterization) is
now Vts , instead of Vtd , see Fig. 4.4. In contrast to Vtd the imaginary part of Vts is
no longer of comparable size as the real part, see Eqs. (2.10-2.11), and the predicted
CP asymmetry is therefore small, arg(Vts ) ∼ ηλ2 .

II No K-oscillations. The final state, containing the mesons J/ψ and φ, is the same
for the Bs0 and the B̄s0 -meson, and hence we do not need the extra K-oscillation
step as in the B 0 system.

III ∆Γ 6= 0. In contrast to the B 0 case, the Bs0 -system has non-vanishing ∆Γ. This
is caused by the existence of a final state common to Bs0 and B̄s0 , with a large
±(∗) ∓(∗)
branching fraction around 5%, namely the CP-eigenstate Ds Ds . Since this
is a CP-eigenstate with eigenvalue +1 this decay channel is only accessible for the
CP-even eigenstate Bs,H and not for Bs,L . Hence the different lifetime for Bs,H and
Bs,L with a predicted value of ∆Γ/Γ ∼ 0.1. (A similar situation for the B 0 case
does not occur, because the branching ratio for B 0 → D ± D ∓ is Cabibbo suppressed,
A ∼ |Vcd |.)

IV Vector-vector final state. The final state now contains two vector-particles with
spin-1. As a result the final state is not a pure CP-eigenstate, in contrast to B 0 →
J/ψKS0 . The spin of the final state particles J/ψ and φ can be pointing parallel,
c c

J/ψ J/ψ
Vbc∗ Vts u, c, t b Vbc
b c s c
Vcs Vcs
∗

Bs0 W+ Bs0 W W W−
Vts
s s b s
u, c, t s
φ φ

s s

Figure 4.4: The two interfering diagrams of the decay Bs0 → J/ψφ, with phase differ-
ence 2βs .
.
50 Chapter 4 CP violation in the B-system

orthogonal, or opposite, which need to be compensated by an orbital momentum,

of l = 2, 1 and 0, respectively, to obtain the spin of the initial state, SBs = 0. The
CP-eigenvalue of the final state now depends on the orbital momentum due to the
factor (−1)l in the total wave function;

CP|J/ψφil = (−1)l |J/ψφil

The fact that the predicted CP-asymmetry is so small in the Standard Model, makes
this decay particularly sensitive to new particles participating in the box-diagram. Any
deviation from the Standard Model value would signal New Physics.
The asymmetry for the decay of the Bs0 -meson to the common final state J/ψφ is given
by Eq. (3.26):

ΓBs0 (t)→J/ψφ − ΓB̄s0 (t)→J/ψφ −ℑλJ/ψφ sin ∆mt

ACP (t) = = (4.10)
ΓBs0 (t)→J/ψφ + ΓB̄s0 (t)→J/ψφ cosh 2 ∆Γt + ℜλJ/ψφ sinh 12 ∆Γt
1

where
q   ĀJ/ψφ   V ∗V   V V ∗ 
l tb ts cb cs
λJ/ψφ = ηJ/ψφ = (−1) ∗
(4.11)
p Bs0 AJ/ψφ Vtb Vts Vcb∗ Vcs
and
Vcb Vcs∗
  
l
ℑλJ/ψφ = (−1) sin 2 arg (4.12)
Vtb Vts∗
= (−1)l sin 2βs (4.13)
= +ηJ/ψφ sin 2βs (4.14)

By comparing Eqs. (4.6) and (4.14) a relative minus sign occurs due to the definition of
β and βs : β is defined with Vtd in the denominator, whereas βs has Vts in the numerator,
see Eqs. (2.15).
A complication arises from the above mentioned vector-vector final state. The contri-
butions from the terms with different orbital momentum, Ak , A⊥ and A0 , for values of
the orbital momentum of 2, 1 and 0, respectively, need to be disentangled statistically
by examining the angular distributions of the final state particles, J/ψ → µ+ µ− and
φ → K +K −.
4.3 γ: the Bs0 → Ds± K ∓ decay 51

4.3 γ: the Bs0 → Ds±K ∓ decay

CP violation in interference between a decay with and without mixing is most simply
realized by considering a final state that is a CP eigenstate. In that case the amplitudes
B → f and B → B̄ → f occur and interfere. In addition, the formulas simplify because
|Af | = |Āf | = |Āf¯| = |Af¯|.
The decay Bs0 → Ds± K ∓ is a final state that is not a CP eigenstate. The interference
can however occur when both the Bs0 and the B̄s0 decay to the same final state, albeit
with different amplitudes this time. We will first examine the pair Bs0 → Ds− K + and
Bs0 → B̄s0 → Ds− K + in a similar way as in the previous sections. This is then followed by
the pair Bs0 → Ds+ K − and Bs0 → B̄s0 → Ds+ K − . The information from both pairs allows
for the extraction of the angle γ in the unitarity triangle.
By examining Fig. 4.5 we see that the amplitude of the decay Bs0 → Ds− K + proceeds
proportional to Af ∼ Vcb∗ Vus whereas the decay B̄s0 → Ds− K + proceeds proportional to
Āf ∼ Vub Vcs∗ . At this point we should note three important aspects:

I Although both the Bs0 -decay and the B̄s0 -decay are equally Cabibbo suppressed,
|ADs− K + | ∼ λ3 and |ĀDs− K + | ∼ λ3 , nevertheless they are not completely equal
|ADs− K + | =
6 |ĀDs− K + |. If we split off the part from the weak couplings, and in-
troduce the hadronic amplitude including effects from the strong interactions in the
final state, A1 and A2 , we get:
 Ā   V V ∗  A 
Ds− K + ub cs 2
= (4.15)
ADs− K + Vcb∗ Vus A1

II Both amplitudes will not only differ by their magnitude, but also by a relative
u s

Vus Vcs
∗

K + Ds−
Vcb∗ Vts Vub
b W+ s s W− c
u, c, t
Bs0 Bs0 W W

u, c, t
s c b u
Vts
Ds− K+

s s

Figure 4.5: The two interfering diagrams of the decay Bs0 → Ds− K + , with phase difference
−γ + 2βs .
.
52 Chapter 4 CP violation in the B-system

phase γ, originating from Vub , see Eq. (4.1), and therefore

ADs− K + |A − + |
= Ds K e−iγ . (4.16)
ĀDs− K + |ĀDs− K + |

III In fact, since the transitions Bs0 → Ds− K + and B̄s0 → Ds− K + proceed in a different
way, an extra relative phase δs needs to be introduced, originating from strong
interactions in the final state,

ADs− K + |A − + |
= Ds K ei(δs −γ) . (4.17)
ĀDs− K + |ĀDs− K + |

(This complication is exactly the reason why we will need the second pair of decays
to the final state Ds+ K − : to disentangle the two phases γ and δs .)

Combining these three points leads to the following expression:

 q   Ā − +   V ∗ V   V V ∗  A 
Ds K ts   ub  2 
λDs− K + = =  tb ∗   ∗ cs  ei(−2βs −γ+δs )) (4.18)

p Bs0 ADs− K + Vtb Vts Vcb Vus A1

In a similar way we obtain the expression for the other pair, where the Bs0 decays to the
CP-conjugated final state, Bs0 → Ds+ K − and Bs0 → B̄s0 → Ds+ K − (see Fig. 4.6):
 q   Ā + −   V ∗ V   V ∗ V  A 
Ds K ts   cb  1 
λDs+ K − = =  tb ∗   us ∗  ei(−2βs −γ−δs )) (4.19)

p Bs ADs+ K −
0 Vtb Vts Vcs Vub A2

where we used |ADs− K + | = |ĀDs+ K − | ∼ A1 .

c s

Vcs Vus
∗
Ds+ K−
Vub
∗ Vts Vcb
b W+ s s W− u
u, c, t
Bs0 Bs0 W W

u, c, t
s u b c
Vts
K− Ds+

s s

Figure 4.6: (a-b) The two interfering diagrams of the decay Bs0 → Ds+ K − , with phase
difference −γ + 2βs .
.
4.4 Direct CP violation: the B 0 → π − K + decay 53

4.4 Direct CP violation: the B 0 → π −K + decay

An example of direct CP violation is given by the decay B 0 → π − K + . A CP-asymmetry

has been observed in the processes B 0 → π − K + and its CP-conjugate B̄ 0 → π + K − ,
|Af | =
6 |Āf | [22]:

ΓB0 →π− K + − ΓB̄0 (t)→π+ K −

ACP = = −0.083 ± 0.004 (4.20)
ΓB0 (t)→π− K + + ΓB̄0 (t)→π+ K −

As before, a different magnitude of the total amplitude between a decay and its CP-
conjugate only appears if the total amplitude AB0 →π− K + consists of two interfering am-
plitudes with a phase difference. In addition, as before, this phase difference needs to
have two components of which one part is CP-odd and flips sign under the CP-operation,
and one part that is CP-even and does not change sign under the CP-operation (often
denoted as the strong phase, since this phase often arises from final state gluon exchange).
In the decays described in the previous sections, the second amplitude originated from the
possibility that the B-meson oscillated before its decay. That is not possible this time,
because the decay B 0 → B̄ 0 → π + K − results in a different final state.
The second amplitude is now given by the a so-called penguin-diagram, as shown in
Fig. 4.7. These penguin diagrams are notoriously difficult to calculate, and therefore it
is difficult to interpret this results in terms of the CKM-angles. However, from Fig. 4.7
it is clear that there is a weak phase difference between the tree (∼ Vub∗ Vus ) and penguin
amplitude (∼ Vtb∗ Vts ), and in general a different strong phase is expected. Intriguingly,
no CP-asymmetry has been observed in the analogous decay B + → π 0 K + , where the
spectator d-quark is “simply” replaced by a u-quark, ACP = 0.037 ± 0.021 [22].

u s
W
Vus
K+ K+
Vub
∗ Vts
b W+ s b u, c, t u
Vtb∗
g
B0 B0

d u d u

π− π−

d d

Figure 4.7: (a-b) The two interfering diagrams of the decay B 0 → π − K + .

.
54 Chapter 4 CP violation in the B-system

4.5 CP violation in mixing: the B 0 → l+νX decay

In the previous sections we always assumed for the B 0 -mesons that |q/p| = 1, originating
from calculations of the box diagram responsible for B 0 ↔ B̄ 0 oscillations. If |q/p| =
6 1
that would mean that the probability to oscillate differ for the B 0 and the B̄ 0 :

Prob(B 0 → B̄ 0 ) 6= Prob(B̄ 0 → B 0 )

The experimental confirmation has been measured using semi-leptonic decays. A semi-
leptonically decaying b-quark proceeds as b → l− ν̄X, whereas the anti-b quark decays as
b̄ → l+ νX. The charge of the lepton contains information whether the B-meson decayed
as a B 0 (containing a b̄-quark) or whether it oscillated and decayed as a B̄ 0 (containing
a b-quark). At the B-factories with the BaBar and Belle experiments both a B 0 and a
B̄ 0 -meson are produced simultaneously through the process e+ e− → Υ → B 0 B̄ 0 .
If the probability to oscillate would be larger for the B 0 than for the B̄ 0 , then the prob-
ability to observe two negatively charged leptons (B 0 → B̄ 0 → l− νX and B̄ 0 → l− νX)
would be larger than to observe two positively charged leptons!

P (B̄ 0 → B 0 ) − P (B 0 → B̄ 0 ) N ++ − N −− 1 − |q/p|4
ACP = = =
P (B̄ 0 → B 0 ) + P (B 0 → B̄ 0 ) N ++ + N −− 1 + |q/p|4

The combined value as measured at the B-factories and LHCb yields [22]:
q 
  0 = 1.0010 ± 0.0008 (4.21)
 
p B

In other words, no CP violation in mixing is observed in the B 0 -system.

The energy at the B-factories is not large enough to produce Bs0 -mesons. The measure-
ment of |q/p| for the Bs0 -system has been done at the Tevatron with the D0 and CDF
experiments, and at LHCb [22]:
q 
  0 = 1.0003 ± 0.0014. (4.22)
 
p Bs
 4.6 Penguin diagram: the B 0 → φKS0 decay 55

4.6 Penguin diagram: the B 0 → φKS0 decay

Already in the decay B 0 → π − K + we introduced the loop diagram that is known as

the penguin diagram, see Section 4.4. On the one hand these diagrams are difficult to
calculate, but on the other hand, these loop diagrams are very interesting because new,
heavy particles might run around in these loops, affecting the measurements. And because
the particles in the loop are virtual, even very heavy particles can contribute.
A particularly interesting example is the decay B 0 → φKS0 , which caused excitement
in recent years. The two interfering diagrams are shown in Fig. 4.8. The situation is
completely analogous as for the decay B 0 → J/ψKS0 from Section 4.1: the B 0 → B̄ 0
oscillation gives rise to the phase difference between the two diagrams (Vtd ∼ eiβ ) and the
time dependent CP-asymmetry is again given by

ACP, B0 →φKS0 (t) = − sin 2β sin(∆mt)

Any difference in the measurement of sin 2β between the decays B 0 → J/ψKS0 and B 0 →
φKS0 might be attributed to new particles in the loop adding an extra phase. The value
of sin 2βφKS was slighly low compared to the value of sin 2βJ/ψKS as measured with tree
diagrams, which generated considerable debate a few years ago.

s s
W W
φ φ
u, c, t Vtd u, c, t b u, c, t
b s d s
g g
B0 B0 W W
Vtd
d W s, d b W s, d
W u, c, t W

KS0 KS0

d, s d, s

Figure 4.8: The two interfering diagrams of the decay B 0 → φKS0 , with phase differ-
ence 2β.
.
56 Chapter 4 CP violation in the B-system
Chapter 5

CP violation in the K-system

CP violation was first discovered in the kaon system and struck the community with large
surprise. CP violation was discovered almost 10 years before the CKM-mechanism was
invented, at the time that only the three quarks (u, d and s) were known. We will discuss
CP violation in the K-system because of its large historical importance.
The nomenclature used in the K-system has some small differences compared to the
B-system, which we will introduce in this chapter. The connection in the K-system
between CP violation and our well known Lagrangian and its short range couplings is less
straightforward.

5.1 CP and pions

Before we dive into the K-system, we give the CP properties of the pion, which will be
relevant when we will discuss the K-decay into pions.
The π 0 is a pseudoscalar meson consisting of a quark and an antiquark. The total wave-
function of the π 0 must be symmetric as it has spin 0. It must however be antisymmetric
under the interchange of the spin of quark and anti-quark as these are fermions. Therefore
the wave function must also be antisymmetric under interchange of the positions of the
quark and antiquark.

|π 0 i = |q ↑ q̄ ↓ i − |q ↓ q̄ ↑ i + |q̄ ↑ q ↓ i − |q̄ ↓ q ↑ i

Performing the parity transformation then yields

P |π 0i = |q̄ ↓ q ↑ i − |q̄ ↑ q ↓ i + |q ↓ q̄ ↑ i − |q ↑ q̄ ↓ i = −1 |π 0 i

The π 0 is thus an eigenstate of the Parity operation with eigenvalue -1. One says it has
negative intrinsic parity.

57
58 Chapter 5 CP violation in the K-system

Performing the C-operation yields (check)

C|π 0 i = |π 0 i
This can also be deduced from the fact that it decays into two photons. As a photon is
nothing more than a combination of electric and magnetic fields and the C operation will
invert both components (why), so that
C|γi = −1 |γi
from which it follows that
C|π 0 i = C|γγi = (−1)2 |γγi = |π 0 i
The combined transformation yields:

CP |π 0 i = −1 |π 0 i (5.1)

and so it is a CP eigenstate with eigenvalue -1 or it “has CP=-1” or “is CP-odd”.

The system |π 0 π 0 i must be symmetric under interchange of the two particles as they
are identical bosons. The CP operation will therefore be merely the product of the CP
operation on the two π 0 s separately
CP |π 0π 0 i = (−1)2 |π 0 π 0 i = +1 |π 0 π 0 i
For the |π + π − i system the C operation interchanges π + and π − and the P operation
changes them back again so that the full CP operation is equivalent to the identity trans-
formation:
CP |π + π − i = 1|π + π − i = +1 |π + π − i
All systems of two pions are eigenstates of CP with eigenvalue +1: they are thus “CP-
even”.
The |π 0 π 0 π 0 i system is again simple because we are dealing with identical bosons the CP
operation is the product of the operation on the three pions separately:
CP |π 0π 0 π 0 i = (−1)3 |π 0 π 0 π 0 i = −1 |π 0 π 0 π 0 i
It is therefore a CP-odd system.
For the |π + π − π 0 i system the relative angular momenta come into play. Let us consider the
situation where the |π + π − i system is produced with angular momentum L = l then if the
total angular momentum of the |π + π − π 0 i system is zero (we are heading for K 0 decay)
the relative angular momentum of π 0 ↔ (π + π − ) will also be L = l. Now performing
the CP operation will for the |π + π − i be the identity operation again, performing it on
the |π 0 i will give -1 and for the part of the wavefunction describing the relative angular
momentum L(π 0 ↔ (π + π − )) = l one gets (−1)l (the wave function is proportional to
Pl (cos θ)). So for l even the system is CP-odd, and for l odd the system is CP-even.
Summarizing the results sofar we have:
5.2 Description of the K-system 59

Pion state CP eigenvalue

π0 -1
π+π− +1
π0π0 +1
π0π0π0 -1
π+π−π0 -1 (L(π+ π− )↔π0 = 0, 2, ..)
+1 (L(π+ π− )↔π0 = 1, 3, ..)

So if CP-symmetry holds then a particle will only be able to decay into a two pion system
if it is a CP eigenstate with eigenvalue +1.

5.2 Description of the K-system

As was introduced in chapter 3 we express the CP-eigenstates as follows:

1 
|K+0 i = √ |K 0 i + |K̄ 0 i

2
1 
|K−0 i = √ |K 0 i − |K̄ 0 i

2

The |K+0 i and |K−0 i states have definite CP-eigenvalues

CP |K+0 i = +1 |K+0 i
CP |K−0 i = −1 |K−0 i

If CP is conserved, the state |K+0 i will only decay into π + π − or π 0 π 0 ( or with a higher
angular momentum to π + π − π 0 ) whereas the the state |K−0 i is strictly forbidden to decay
0
into a two pion final state. Because the mass of the KL/S is approximately 497.6 MeV and
the mass of a pion is about 139.6 MeV the available phasespace for the two pion decay
is almost a factor 1000 larger than that available for the three pion decay. As a result,
the lifetime of the CP-odd eigenstate of the K-system is very large, much larger than the
lifetime of the CP-even eigenstate. This is the reason that the CP-eigenstates are referred
to as the KS0 and KL0 , where the subscripts stand for short and long, respectively, and not
referred to as heavy and light as is done in the B-system 1 .

1 
|KS0 i = √ |K 0 i + |K̄ 0 i

2
1 
|KL0 i = √ |K 0 i − |K̄ 0 i

(5.2)
2
1
The KL0 corresponds to the heavy eigenstate, so could also have been named the KH ...
60 Chapter 5 CP violation in the K-system

5.3 The Cronin-Fitch experiment

Until 1964 all measurements were consistent with the notion of CP-symmetry, even those
which involve the weak interaction. In fact CP-symmetry was invoked to explain the
large difference in lifetime between the KL0 and KS0 . The experiment which unexpectedly
changed this situation was performed by Christensen, Cronin, Fitch and Turlay [23] in
1964.
The experimental apparatus is shown in Fig. 5.1. It consisted of a Be-target placed in
a π − beam. All particles produced in the interactions, including any K 0 s were allowed
to decay in a low pressure He-tank. Decay products were detected in two magnetic
spectrometers placed roughly 20 m from the target. The distance of 20 m corresponds to
approximately 300 lifetimes for the KS0 . All decay products must therefore come from the
KL0 . All opposite charge combinations of particles, which had a reconstructed decay vertex
within the He-volume were analysed and their invariant mass was determined under the
assumption that both detected particles were pions. Obviously one expects to observe
invariant mass combinations with a mass smaller than the K 0 mass emanating from the
KL0 → π + π − π 0 decay (M(π + π − ) < M(KL0 ) − M(π 0 )). However some background was
produced in the experiment from the decays KL0 → πµν and KL0 → πeν where the µ and
the e are misidentified as pions (check the mass limit for these decays). Fig. 5.2a) shows
the measured spectrum. The figure shows a Montecarlo prediction from all known decays
of the KL0 (e.g. the peak at about 350 MeV is from the KL0 → π + π − π 0 decay. At first

Figure 5.1: The experimental apparatus with which CP violation was first measured.
.
5.3 The Cronin-Fitch experiment 61

glance there is no real discrepancy between the measurements and the MC prediction.
Certainly there is no indication of an excess of events at around 500 MeV. If we however
plot the cosine of the angle between the flightpath of the K 0 and the direction of the
momentum sum of the two particles for 490 < M(π + π−) < 510 MeV we start to see an
excess appear for cos θ ∼ 1, see Fig. 5.2b). This is of course exactly what one expects for
the decay KL0 → π + π − . Fig. 5.2d) shows this in a little more detail. The forward peak
is only present for 494 < M(π + π−) < 504 MeV. Outside this mass interval there is no
indication for a forward enhancement. The enhancement contains 49 ± 9 events. This was
after many consistency checks finally taken as proof that the decay KL0 → π + π − occurs
in nature. After acceptance correction the experiment gave a branching ratio of:
Γ(KL0 → π + π − )
BR(KL0 → π + π − ) = 0
= 2.0 ± 0.4 × 10−3
Γ(KL → all charged decay modes)
This result proves then that CP-symmetry is violated in the decay of the KL0 , of course

(c)

(a)

(d)

(b) (e)

Figure 5.2: (a) The measured two “pion” mass spectrum. (b) The distribution of the cosine
of the angle between the summed momentum vector of the two pions and the direction of
the K 0 beam. (c-e) The angular distribution for different ranges in the invariant mass.
.
62 Chapter 5 CP violation in the K-system

one has to be careful that the effect seen is indeed the decay of the KL0 , as there are some
subtle effects that could affect the result.

5.3.1 Regeneration

Here we will discuss the effects of the passage through matter of a state which is a
superposition of |K 0 i and |K̄ 0 i. In the Hamiltonian we will now also have to take into
account the strong interactions of the state with the matter it is passing through. We
will neglect any inelastic interactions as these will merely decrease the intensity. We
know from experiment that the strong interactions of |K 0 i and |K̄ 0 i are different. The
¯ contains an s-quark and can, in its interaction with matter, produce strange
|K̄ 0 i (sd)
baryon resonances, like K̄ 0 + n → Λ + π 0 , whereas the K 0 (s̄d) can not. So the total cross
section for K 0 scattering will be smaller than for K̄ 0 .
Suppose that a pure KL0 beam would incident on matter where all K̄ 0 would be absorbed,
then the outgoing beam would be pure K 0 . Similar to a Stern-Gerlach filter, half of the
outgoing kaons would then decay as a KS0 and half as KL0 , see Eq. 5.2:
1 
|K 0 i = √ |KS0 i + |KL0 i

2

In principle the effect seen in the Cronin experiment could have been due to regeneration
of the KL0 beam. If this would be the case then clearly by introducing more material in the
path of the KL0 beam the effect would increase. The experiment was therefore repeated
with liquid hydrogen instead of He in the decay path. The density and so the size of the
regeneration then grows by a factor of 1000. The growth of the signal was found to be
the equivalent of 10 events. The experiment was also repeated with the He replaced by
vacuum. The signal persisted, so that regeneration could be ruled out as the cause.
Finally one has to prove that the particle which decays into the π + π − state is in fact the
KL0 state. To prove this one determined that there existed interference between the state
decaying into π + π − and a regenerated KS0 .
The only remaining conclusion was therefore that CP-symmetry is violated in weak
interactions.
5.4 Master Equations in the Kaon System 63

5.4 Master Equations in the Kaon System

In Section 3.7 a classification of the various types of CP violation was made. In the
following these various types will be examined in the kaon system. First, let us introduce
the quantity η+− by starting from the familiar master equations.

ΓK 0→f (t) = |Af |2 |g+ (t)|2 + |λf |2 |g− (t)|2 + 2ℜ[λf g+

∗


(t)g− (t)]
 2
2 p
ΓK̄ 0→f (t) = |Af |   |g− (t)|2 + |λf |2 |g+ (t)|2 + 2ℜ[λf g+ (t)g−
∗
 

(t)]
q
(5.3)

with
1 −ΓS t
|g± (t)|2 = + e−ΓL t ± e−Γt (e−i∆mt + e+i∆mt )


e
4
∗ λf −ΓS t
− e−ΓL t + e−Γt (e−i∆mt − e+i∆mt )


λf g + (t)g− (t) = e (5.4)
4
yields

|Af |2 
ΓK 0 →f (t) = e−ΓS t (1 + |λf |2 + 2ℜλf ) + e−ΓL t (1 + |λf |2 − 2ℜλf ) +
4 
e ((1 − |λf |2 )(e−i∆mt + e+i∆mt ) + 2ℜ(λf (e−i∆mt − e+i∆mt )))
−Γt

|Af |2 
= e−ΓS t (1 + λf )(1 + λ∗f ) + e−ΓL t (1 − λf )(1 − λ∗f ) +
4 
e ((1 − |λf |2 )(cos ∆mt) + 2ℑ(λf ) sin ∆mt))
−Γt

 (1 − λf )(1 − λf )∗
∼ e−ΓS t + e−ΓL t +
(1 + λf )(1 + λf )∗
(1 − |λf |2 ) ℑ(λf ) 
e−Γt ( (2 cos ∆mt) − 4 sin ∆mt))
(1 + λf )(1 + λf )∗ (1 + λf )(1 + λf )∗

and finally

ΓK 0 →f (t) = N e−ΓS t + e−ΓL t |η+− |2 + 2e−Γt |η+− | cos(∆mt + φ+− )

ΓK̄ 0 →f (t) = N e−ΓS t + e−ΓL t |η+− |2 − 2e−Γt |η+− | cos(∆mt + φ+− )

(5.5)
1−λf
with η+− = 1+λf
= |η+− eiφ+− |.
64 Chapter 5 CP violation in the K-system

5.5 CP violation in mixing: ǫ

It is clear that we can no longer identify the KL0 with the K−0 and the KS0 with the K+0 ,
as they are clearly no longer eigenstates of the full Hamiltonian, and therefore we write:

|KS0 i = p|K 0 i + q|K̄ 0 i

|KL0 i = p|K 0 i − q|K̄ 0 i

Historically, the CP violation was parameterized by introducing an arbitrary complex

number ǫ, because the KS0 and KL0 were almost CP-eigenstates:
1
|KL0 i = p |K−0 i + ǫ|K+0 i


(5.6)
1 + |ǫ|2

and
1
|KS0 i = p |K+0 i − ǫ|K−0 i


(5.7)
1 + |ǫ|2
p p
where p = (1 + ǫ)/ 2(1 + |ǫ|2 ), q = (1 − ǫ)/ 2(1 + |ǫ|2 ) and thus q/p = (1 − ǫ)/(1 + ǫ).
The parameters p and q are normalized such that |p|2 + |q|2 = 1.
Let us consider the decays K 0 → π + π − and K̄ 0 → π + π − and define the parameter λf , as
in Eq. (3.17) [2]:
 q  Ā + −
π π
λπ + π − = (5.8)
p K Aπ+ π−
The amount of CP violation is measured by determining the relative branching ratio of
BR(KL0 → π + π − ) over BR(KS0 → π + π − ):

hπ + π − |T |KL0 i pAπ+ π− − q Āπ+ π− 1 − λπ + π −

η+− ≡ + − 0
= =
hπ π |T |KS i pAπ+ π− + q Āπ+ π− 1 + λπ + π −

If η+− 6= 0 then that means |λπ+ π− | =

6 1. In this way we have reduced the CP violation to
CP violation in the mixing of the K 0 and K̄ 0 , whereas the interaction that describes the
decay is still CP-invariant. Only the part of the wavefunction that is CP-even will decay
to the CP-even two pion states.
Similarly, for the decay to two neutral pions the parameter η00 is introduced. Their
measured values are:

|η+− | = (2.285 ± 0.019) × 10−3

|η00 | = (2.275 ± 0.019) × 10−3

The value of ǫ is related to ǫ ≈ η+− and ǫ ≈ η00 (see next section) and amounts to
|ǫ| = (2.228 ± 0.011) × 10−3 [24], yielding (compare to Eqs. (4.21-4.22)):
q  1−ǫ
  0= = 0.995552 ± 0.000024. (5.9)
 
p K 1+ǫ
5.6 CP violation in decay: ǫ′ 65

If this is true this will have consequences for the semi-leptonic decay of the KL0 . We
rewrite Eq. (5.6) as:
1
|KL0 i = p |K−0 i + ǫ|K+0 i


1 + |ǫ|2
1
(1 + ǫ)|K 0 i − (1 − ǫ)|K̄ 0 i
 
= p
2(1 + |ǫ|2 )

The charge asymmetry in the decay of the KL0 will then be

Γ(KL0 → e+ π − νe ) − Γ(KL0 → e− π + ν̄e )
A+− =
Γ(KL0 → e+ π − νe ) + Γ(KL0 → e− π + ν̄e )
|1 + ǫ|2 − |1 − ǫ|2
= (5.10)
|1 + ǫ|2 + |1 − ǫ|2
∼ 2ℜ ǫ

If the wavefunction of the KL0 is indeed a superposition of the two CP-eigenstates then
there will be a difference in the rates. The measured asymmetry [24] is

A+− = 3.32 ± 0.06 × 10−3

confirming the above assumption (ǫ = |ǫ|eiφǫ , with φǫ = 43o ). The size of the effect is
consistent with the two pion decay rates.
There is of course still the possibility that the decay of the CP-eigenstates, from which
the KL0 is built, also violates CP-symmetry: i.e. the |K−0 i part of the wavefunction decays
into π + π − in that case we speak of direct CP violation.

5.6 CP violation in decay: ǫ′

If the amount of CP violation would depend on the final state, then that obviously implies
that the decay contributes to the CP violation. In other words, η+− 6= η00 implies direct
CP violation. We will see that this difference is expressed with the parameter ǫ′ [2]:
η00 ǫ − 2ǫ′ ǫ′
≈ ≈ 1 − 3
η+− ǫ + ǫ′ ǫ

To investigate the possibility of direct CP violation in the K 0 system we consider the

transition from the |K 0 i state to an eigenstate of the strong interaction and perform the
CP transformation:
CP
hA|H|K 0 i −→ hA|(CP )−1H(CP )|K 0i = hA|H|K̄ 0 i

Here we have used the arbitrary phase of the CP-transformation CP |K 0i = +|K̄ 0 i. For
one transition amplitude this is always possible. However if a second transition can take
66 Chapter 5 CP violation in the K-system

place then this will have to follow the same phase-convention, and so if a transition is
found that has a non-zero phase with respect to the first then we will have CP violation.
The two pion system can occur in two distinct eigenstates of the strong interaction, namely
I = 0 and I = 2. So we can decompose the two-pion states emanating from the KL0 and
KS0 decay into the Isospin eigenstates:
+ − 1 √ 
|π π i = √ 2|2π, I = 0i + |2π, I = 2i
3
1  √ 
|π 0 π 0 i = √ |2π, I = 0i − 2|2π, I = 2i
3
We can now define the amplitude for the transitions into the I = 0 state as:
h2π, I = 0|T |K 0i = h2π, I = 0|T |K̄ 0i = A0 eiδ0
where we have added a phase shift due to the final state strong interactions in the I = 0
state, δ0 . For the I = 2 state we will in general then not have a real amplitude:
h2π, I = 2|T |K 0i = A2 eiδ2
h2π, I = 2|T |K̄ 0i = A∗2 eiδ2
(δ2 is the final state strong interaction phaseshift for the I = 2 state.) Introducing the
following variables:
F = ei(δ2 −δ0 )
F ℜA2
∆ = √
2 A0
iF ℑA2
ǫ′ = √
2 A0
we find for the amplitudes
hπ + π − |H|KL0 i
η+− = = ǫ + ǫ′ (1 + ∆)−1
hπ + π − |H|KS0 i
and
hπ 0 π 0 |H|KL0 i
η00 = = ǫ − 2ǫ′ (1 − 2∆)−1
hπ 0 π 0 |H|KS0 i
so that, assuming that |∆| ≪ 1 and |ǫ′ | ≪ 1 will be small, we find for the rate asymmetry
 η00 2 R(KL0 → π 0 π 0 ) R(KS0 → π + π − ) ǫ′
 
  = = 1 − 6ℜ
 η+−  R(KS0 → π 0 π 0 ) R(KL0 → π + π − ) ǫ
So if ǫ′ 6= 0 then ℑA2 6= 0 and so the phase of the transition to the I = 2 state is not equal
to the phase of the transition to the I = 0 state and we will have direct CP violation.
The experimental result is
ǫ′ ǫ′
ℜ ≈ = 1.65 ± 0.26 × 10−3 (5.11)
ǫ ǫ
5.7 CP violation in interference 67

5.7 CP violation in interference

In Section 3.7 a classification of the various types of CP violation was made. We just saw
how CP is violated in the kaon system in decay, and in mixing:

1) CP violation in decay: ǫ′

2) CP violation in mixing: ǫ

3) CP violation in interference between a decay with and without mixing,

The CP violation in the interference between a decay with and without mixing obviously
depends on the neutral meson mixing and is therefore time-dependent. Often this is thus
referred to as time-dependent CP-asymmetry. Interference occurs when there are two
amplitudes for a transition from a given initial state to a given final state. For this to
happen, we now need a final state that is a CP eigenstate in order to obtain the two
amplitudes K 0 → fCP and K 0 → K̄ 0 → fCP . An example of such a final state is simply
K 0 → π+π−.
The time dependent CP asymmetry of K 0 → π + π − and K̄ 0 → π + π − is shown in Fig 5.3b)
and is compared to the time dependent CP asymmetry as measured in the B-system with
B 0 → J/ψKS .

CPLEAR

•: K̄ 0
◦: K 0

(a) (b)
0
Figure 5.3: Number of B → J/ψKS0 candidates with a B 0 tag NB0 and with a B̄ 0 tag
NB̄0 , and below the asymmetry ACP (t) = (NB0 − NB̄0 )/(NB0 + NB̄0 ), as functions of ∆t.
From [21]. (b) Number of K 0 → π + π − candidates. Open circles ◦ correspond to kaons
that started as K 0 , wheras closed circles • correspond to K̄ 0 tags. From [25]
.
68 Chapter 5 CP violation in the K-system
Chapter 6

Experimental Aspects and Present

Knowledge of Unitarity Triangle

6.1 B-meson production

In principle b and b̄ quarks are always made in pairs, the way they dress up into hadrons
is however dependent on the specific production. At present there are three accelerator
types, where significant results can be expected for CP violation.

• e+ e− colliding beam machines at a CM energy of the Υ(4S)

• e+ e− colliding beam machine at a large CM energy (p.e. LEP)

• Hadron colliders such as Tevatron (pp̄) or LHC (pp)

e+ e− colliding beam machines at a CM energy of the Υ(4S)

The Υ(4S) resonance is the first bb̄ resonance which can decay into “open” b. It decays
into B 0 B̄ 0 or B + B − -mesons, see Fig. 6.1. The CM energy (mass of the Υ(4S)) is such
that only the B-mesons are produced. Most notably the mass of the Bs0 is such that it
can not be produced in these collisions. Also additional hadrons (pions and kaons) are
kinematically forbidden. In the CM system the B-mesons are produced essentially at
rest. This means that the decay-length cannot be measured as the velocity of the meson
is to good approximation zero. This means that if a decay-length measurement is to be
made one must ensure that the CM system is in motion. For this reason the accelerators
at Stanford (BaBaR) and KEK (Belle) employ asymmetric beam energies for the e+ and
e− beams. It has a disadvantage for the machine as one needs two separate accelerators.
BaBaR for instance has a 3.1 GeV e+ beam and a 9 GeV e− beam to produce a CM
energy of 10.56 GeV. The CM system thus has a βγ = 0.56 so that the mean decay length

69
70 Chapter 6 Experimental Aspects and Present Knowledge of Unitarity Triangle

b
B̄ 0
e− b d¯
γ
Υ

e+ b̄ d

B0
b̄

Figure 6.1: B 0 B̄ 0 production via Υ(4S) decay.

.

of a B-meson (τ = 1.5 ps) produced at rest in the CM system is βγcτ ≈ 250 µm. There
is also a disadvantage for the analysis as the actual production vertex is not known, so
that all CP asymmetries must be rewritten in terms of the difference of the lifetimes of
the two B-mesons.

Hadron colliders such as Tevatron (pp̄) or LHC (pp)

The production mechanism for b and b̄ quarks is the same in both pp̄ and pp collisions.
They are formed when a gluon from the proton fuses with a gluon of the (anti-)proton
(see Fig. 6.2). From measurements in deep-inelastic scattering we know that the gluon

p hadrons
_
b

( )_
p b

hadrons

Figure 6.2: B B̄ production in high energy hadron machines through gluon-gluon fusion

density in the (anti-)proton is largest at very small fractional momentum (x). In fact the
gluon density behaves approximately as
g(x) ∝ x−3/2
6.2 Flavour Tagging 71

√
This means that to produce a mass M(≈ 2mb ) = x1 x2 s, where s is the centre-of-mass
energy squared, the most probable situation is that either x1 or x2 is very small (and the
other large). The b and the b̄ are thus produced in the same hemisphere at small angles
to the beam and at quite high momentum. The momenta are in the range of 30 to 100
GeV giving a mean decay distance of βγcτ ≈ 3 − 10 mm. In addition to the B-mesons
many other particles are produced. A similar mixture of B-meson flavours occurs as in
high energy e+ e− collisions.
At these machines the interaction rate of events with no b quarks is so high that a selection
must be performed in order not to be swamped by unwanted events. In fact at the LHC
the production rate of b quark containing events is so high that a large fraction have to be
filtered away because they do not contain interesting decays. This question is obviously
a very subtle one and is too involved for the present lecture.

6.2 Flavour Tagging

As we saw in chapter 4 many measurements depend on the knowledge whether the B-

meson oscillated or not. In order to determine whether the B-meson was produced as a
B or a B̄, the flavour at production needs to be tagged. In principle tagging is simple.
All one has to do is identify the flavour of the B meson accompanying the one decaying
into a CP eigenstate. There are several methods which are or will be used.

• Complete reconstruction of the decay of a charged B-meson. This is the gold plated
tagging method. It suffers however from efficiencies and branching ratios. Typically
the decay in which one is interested has a branching ratio of less than 10−3 . Com-
bining this with a similar branching ratio for the tagging decay gives too small a
fraction of events. At the Υ this method is anyway excluded.

• Determination of the charge of the secondary vertex of the accompanying B-meson.

Also not usable at the Υ.

• Semi-leptonic decay of the accompanying B-meson. The b-quark will decay into
a negatively charged lepton whereas the b̄-quark decays into a positively charged
lepton. If these semi-leptonic decays are of charged B-mesons detection of the lepton
will provide an unambiguous tag. If however the accompanying B-meson is neutral
then the tag only indicates the flavour at the time of the accompanying B-meson’s
decay and it can have oscillated. It is interesting to calculate that the time integrated
CP asymmetry actually becomes zero at the Υ if only the difference in lifetimes and
the semi-leptonic tagging are used. In the high energy e+ e− and hadron machines
this method works, be it with some tagging inefficiency. This inefficiency has to be
measured using other channels (p.e. double semi-leptonic decays) or estimated from
Monte Carlo. This method also suffers from misidentification due to semi-leptonic
charm decay in the decay of the accompanying B-meson. The lepton from this decay
72 Chapter 6 Experimental Aspects and Present Knowledge of Unitarity Triangle

has the opposite charge to the one from the original b decay and so will further wash
out the tagging information.

• Other methods include reconstruction of charm particles (the charge of the kaon
in the decay provides an unambiguous tag, apart from B 0 ↔ B̄ 0 oscillations) and
charge of the total opposite jet (at high energy e+ e− )
 6.3 Present Knowledge on Unitarity Triangle 73

6.3 Present Knowledge on Unitarity Triangle

At present there are several measurements which constrain the CKM unitarity triangle.
Combining all these measurements in a global fit is a stringent test of the internal consis-
tency of the Standard Model. The two best known groups that perform these global fits
are CKMfitter [26] and UTfit [27].
In this section we will present the input to the fit, of which the following four measurements
provide the strongest constraints:

I sin2β The measurement of sin 2β constrains one of the three angles of the triangle.

II ǫK The measurement of ǫK provides a constraint that follows a hyperbola in the

(ρ, η) plane.

III |Vub | The measurement

p of |Vub /Vcb| constrains one side of the triangle as it is
2 2
proportional to ρ + η .

IV ∆m The measurements of ∆md and ∆ms for the B 0 and Bs0 systems constrain
another side, as it is proportional to ((1 − ρ)2 + η 2 ).

The first two measurements are direct proofs of CP violation, in the B and K-system
respectively. The last two measurements however, provide strong constraints in the ρ̄, η̄-
plane, but are no signs of CP violation on their own, since they allow for vanishing
imaginary part of the CKM elements, η̄ = 0, see Fig. 6.3.

1.5
η
exc

excluded area has CL > 0.95 γ

lud
ed

γ 1
at C

summer18
L>

1.0
∆md & ∆ms
0.9

∆md
5

β ∆md
sin 2β ∆ms
0.5
0.5
∆md
εK α εK Vub
γ β Vcb
η

0.0 0
α
Vub α
-0.5
−0.5 α

γ εK
-1.0
CKM sol. w/ cos 2β < 0
fitter
Summer 19 (excl. at CL > 0.95)
−1

-1.5
-1.0 -0.5 0.0 0.5 1.0 1.5 2.0 −1 −0.5 0 0.5 1
ρ ρ
Figure 6.3: Global fits to the unitarity triangle, by (a) CKMfitter [26] and (b) UTfit [27].
74 Chapter 6 Experimental Aspects and Present Knowledge of Unitarity Triangle

6.3.1 Measurement of sin 2β

The determination of the angle β as measured at the B factories with the BaBar and
Belle experiments, has been extensively discussed in Section 4.1, which resulted in the
present world average when combined with the value from LHCb [28]:
sin 2β = 0.70 ± 0.02

sin(2β) ≡ sin(2φ1) HFLAV

Moriond 2018
PRELIMINARY
BaBar 0.69 ± 0.03 ± 0.01
PRD 79 (2009) 072009
BaBar χc0 KS 0.69 ± 0.52 ± 0.04 ± 0.07
PRD 80 (2009) 112001
BaBar J/ψ (hadronic) KS 1.56 ± 0.42 ± 0.21
PRD 69 (2004) 052001
Belle 0.67 ± 0.02 ± 0.01
PRL 108 (2012) 171802
2018

ALEPH 0.84 +-01..8024 ± 0.16

HFLAV

PLB 492, 259 (2000)

Moriond

OPAL 3.20 +-12..8000 ± 0.50

EPJ C5, 379 (1998)
CDF 0.79 +-00..4414
PRD 61, 072005 (2000)
LHCb 0.76 ± 0.03
JHEP 11 (2017) 170
Belle5S 0.57 ± 0.58 ± 0.06
PRL 108 (2012) 171801
Average 0.70 ± 0.02
HFLAV

-2 -1 0 1 2 3

Figure 6.4: Average value of the measurement of sin 2β.

.

6.3.2 Measurement of ǫK

The measurement of |ǫK | = (2.228 ± 0.011) × 10−3 provides a constraint on the position of
the apex of the universality triangle in the (ρ̄, η̄) plane. The value of |ǫK | is a measure of
the CP violation in K-mixing, and is thus related to the imaginary part of the box diagram
contribution that is responsible for the short range mixing of K 0 ↔ K̄ 0 : ǫ ∼ ℑM12 /∆mK .
Using the box diagram to calculate |ǫK |, see Eq. (3.15), one arrives at the following
form [29]:
G2 m2 m f 2 B
|ǫK | = 12F√2πW2 K∆mKK K ℑ ηc S(xc )(Vcs∗ Vcd )2 + ηt S(xt )(Vts∗ Vtd )2 + 2ηct S(xc , xt )(Vcs∗ Vcd Vts∗ Vtd )
 
2 B
G2F m2W mK fK
= √
12 2π 2 ∆mK
K
[ηc S(xc )2ℜ(Vcs∗ Vcd)ℑ(Vcs∗ Vcd ) + ηt S(xt )2ℜ(Vts∗ Vtd )ℑ(Vts∗ Vtd ) −
ηct S(xc , xt )ℜ(Vcs∗ Vcd )ℑ(Vcs∗ Vcd )]
here again the functions S(xq ) have been derived by Inami and Lim [17] and quantify
the loop contributions from quark q, depending on xq = m2q /m2W . The ηq include the
6.3 Present Knowledge on Unitarity Triangle 75

NLO QCD corrections for each function. The factors fK2 BK again parameterize the non-
perturbative strong corrections. The value of fK is well known from measurements of
charged K decay, so that the most uncertain value is that of the bag-factor BK . The
third term is evaluated as follows: (Vcs∗ Vcd Vts∗ Vtd ) ≡ (λc λt ) = (ℜλc + iℑλc )(ℜλt + iℑλt ).
Using ℑλc ≈ −ℑλt and ℜλt ≪ ℜλc , we then find ℑ(Vcs∗ Vcd Vts∗ Vtd ) ≈ −ℜ(Vcs∗ Vcd )ℑ(Vcs∗ Vcd ).
Using the Wolfenstein parameterization we find [2]:

G2F m2W mK fK2 BK 2 6 

A λ η ηc S(xc ) − ηt S(xt )A2 λ4 (1 − ρ) − ηct S(xc , xt )

|ǫK | = √
12 2π 2 ∆mK
≈ 104 A2 λ6 η ηc S(xc ) − ηt S(xt )A2 λ4 (1 − ρ) − ηct S(xc , xt ) .
 

With |Vcb | = Aλ2 and |Vus | = λ and the evaluation of the Inami-Lim functions S(xc ) ≈
2.4 × 10−4 , S(xt ) ≈ 2.6 and S(xc , xt ) = 2.2 × 10−3 [30] we can rewrite as:

|ǫK | ≈ 104 η|Vcb|2 |Vus |2 2.4 × 10−4 + 2.6|Vcb|2 (1 − ρ) − 2.2 × 10−3

 

≈ 10−3 η [(1 − ρ)]

(6.1)

which becomes a hyperbolic band in the (ρ̄, η̄) plane, given in Fig. 6.3.

6.3.3 |Vub/Vcb |

The ratio |Vub /Vcb| provides a strong constraint on the unitarity triangle. The present
measurement is given by
|Vub | = (3.94 ± 0.36) × 10−3
In the Standard Model in the Wolfenstein parameterization this quantity is given by

λ p 2
|Vub/Vcb | = (ρ̄ + η̄ 2 )
1 − λ2

where ρ̄ = (1 − λ2 /2)ρ and η̄ = (1 − λ2 /2)η. This constraint is shown in Fig. 6.3 by the
circular band with its origin at (0,0) in the (ρ̄, η̄) plane. This band is a strong constraint
in the (ρ̄, η̄) plane, but it is on its own not a measurement of CP violation: a solution
with the apex of the unitarity triangle at η̄ = 0 would be perfectly consistent with this
constraint.

6.3.4 Measurement of ∆m

The mass difference ∆m of the two mass-eigenstates of a neutral meson system results in
an oscillatory behaviour between the meson and anti-meson, B 0 ↔ B̄ 0 , as explained in
chapter 3.
76 Chapter 6 Experimental Aspects and Present Knowledge of Unitarity Triangle

The oscillations of neutral B-mesons were first observed at the PETRA collider at DESY
with the ARGUS experiment in 1987 [31]. The oscillations were more rapid than expected
at that time, because the mass of the top quark was not expected to be that heavy. In
fact, fortunately it turned out that the B 0 -meson has a fair chance to oscillate before she
decays. As a consequence of the determination of ∆md , a lower limit on the top quark
mass could be set, mt > 40 GeV 1
The mass difference ∆md has been very accurately determined by the B factories.
These decays can, as with semi-leptonic decays, only proceed from the B 0 or B̄ 0 part of
the wavefunction. In this case the tagging is done using muons detected in the opposite
hemisphere to the particle under study. A clear oscillation is seen and the extracted mass
difference is obtained
∆md = (0.5065 ± 0.0019)ps−1

In the Standard Model the calculation of the box-diagram yields the following expression
for this mass difference, see Eq. (3.16):

G2F 2
mW ηC S(xt )A2 λ6 (1 − ρ̄)2 + η̄ 2 mBd fB2 d BBd
 
∆md = 2
6π
where S(xt ) is again the Inami-Lim function [17], xt = m2t /m2W , mt and mW are the top
quark and W masses, ηc = 0.55±0.01 is the NLO QCD correction to the box-diagram am-
plitude and the most uncertain factor fB2 d BBd parameterizes the non-perturbative strong
corrections. Using the best estimates of all the parameters this translates into a limiting
region in the (ρ̄, η̄) plane. It is shown as the circular shaded band centered around (1, 0)
in Fig. 6.3.
Recently the CDF and D0 experiments at the pp̄ collider Tevatron at Fermilab have
measured the mass difference in the Bs0 system:

∆ms = (17.757 ± 0.021)ps−1

The ratio ∆md /∆ms will allow a determination of this radius which is theoretically less
uncertain, as this quantity is given by:

∆md mBd fB2 d BBd 2  2 2


= λ (1 − ρ̄) + η̄
∆ms mBs fB2 s BBs

Here almost all corrections have cancelled and the ratio of the non-perturbative factors is
much better under control, hence the narrow circular band inside the circle coming from
∆md alone, see Fig. 6.3.

This happened at the time that the TOPAZ e+ e− -collider in Japan was about to become operational
1

with the aim to discover the top quark up to a mass of about 40 GeV...
6.3 Present Knowledge on Unitarity Triangle 77

0.7

0.6 ∆md
sin2β
0.5
0.4

0.3 εK

0.2

0.1
|Vub|
0
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1
ρ

Figure 6.5: Sketch of the four measurements ǫK , |Vub |, ∆md and sin 2β in the (ρ, η) plane.
In the Standard Model, all four curves should be consistent with one value of (ρ, η).
.
78 Chapter 6 Experimental Aspects and Present Knowledge of Unitarity Triangle

6.4 Outlook: the LHCb experiment

The B-factories at the SLAC (USA) and KEK (Japan) with the BaBar and Belle ex-
periments have been extremely succesful in measuring CP violation in the B 0 system,
resulting in a very accurate determination of the angle β. However, the uncertainty on
the angle γ is still large. The angle βs has not been measured at all yet, although some
claims of new physics have been made [32, 33], based on the measurements at the CDF
and D0 experiments with the Tevatron collider at Fermilab, Chicago. The Bs0 system is to
date very poorly constrained, and might hide interesting new physics effects in the b ↔ s
transition.
The LHCb detector aims at determining γ and βs at unprecedented precision. Two prime
examples are given in Sections 4.3 and 4.2 were γ and βs are extracted from the decays
Bs0 → Ds± K ∓ and Bs0 → J/ψφ, respectively. The B-factories run at the Υ(4S) resonance
which does not provide enough energy to produce Bs0 -mesons. On the other hand, Bs0 -
mesons are produced at the pp̄ collider at Fermilab. But a relatively low bb̄ cross section
of 50µb at the center-of-mass energy of 2 TeV and a modest yearly collected luminosity
of ∼ 1fb−1 only yields ∼ 3, 000 Bs0 → J/ψφ events in the period from 2002-2008.
The LHCb experiment operates at the LHC collider running at a center-of-mass energy
of 7(8) TeV in 2011 (2012) (with a bb̄ cross section of 230µb) and a yearly luminosity of
2 fb−1 . In 2015-2018 the LHC operated at 13 TeV, and as a result, the LHCb experiment
collected about 100,000 Bs0 → J/ψφ events per year. Not only due to the large amount
of collected Bs0 -mesons, but also due to the optimized (forward) detector design, see
Fig. 6.6, an accuracy of σφs = 0.03 was reached in 2018, combined with results from
ATLAS and CMS. The optimized detector design also comprises particle identification

Figure 6.6: Schematic view of the LHCb detector.

.
6.4 Outlook: the LHCb experiment 79

(to distinguish kaons from pions) and an efficienct hadron trigger, which places LHCb
in a special situation compared to the other LHC experiments ATLAS and CMS. Due
to these two features, LHCb collected 6000 Bs0 → Ds± K ∓ events in 2011 and 2012, with
which γ was determined with σγ = 20o , showing the feasibility of this analysis with high
precision.
These new precision measurements will scrutinize the Standard Model and her CKM-
mechanism. Together with the determination of angular distributions and branching
ratios of rare decays such as B 0 → K ∗ µ+ µ− and Bs0 → µ+ µ− the measurements at LHCb
might reveal new particles inside virtual loops, complementary to the possible direct
production of new particles at ATLAS and CMS.
80 References
References

[1] I.I. Bigi and A.I. Sanda, CP Violation. Cambridge University Press, Cambridge,
(England), 2000.

[2] G.C. Branco, L. Lavoura and J.P. Silva, CP Violation. Clarendon Press, Oxford,
(England), 1999.

[3] C.S. Wu et al., Experimental test of parity conservation in beta decay. Phys. Rev.
105, 1413 (1957).

[4] R.L.Garwin et al., Observations of the failure of conservation of parity and charge
conjugation in meson decays: The magnetic moment of the free muon. Phys. Rev.
105, 1415 (1957).

[5] C.Jarlskog (editor), CP Violation, 1989, available on

http://books.google.com/books?id=U5TC5DSWOmIC&vq=cp violation&pg=PR1.
Published by World Scientific.

[6] Y. Nir, CP violation in meson decays. Preprint hep-ph/0510413, 2005.

[7] M.Kobayashi and K.Maskawa, CP violation in the renormalizable theory of weak

interaction. Prog. Theor. Phys. 49, 652 (1973).

[8] L. Wolfenstein, Parametrization of the kobayashi-maskawa matrix. Phys. Rev. Lett.

51, 1945 (1983).

[9] R. Fleischer, Flavour physics and CP violation: Expecting the LHC. Preprint
arXiv:0802.2882, 2008.

[10] I.Y. Bigi, Flavour dynamics & CP violation in the standard model: A crucial past
and an essential future. Preprint hep-ph/0701273, 2007.

[11] Fukuda, Y. and others, Evidence for oscillation of atmospheric neutrinos.

Phys.Rev.Lett. 81, 1562 (1998), hep-ex/9807003.

[12] Maki, Ziro and Nakagawa, Masami and Sakata, Shoichi, Remarks on the unified
model of elementary particles. Prog.Theor.Phys. 28, 870 (1962).

81
82 References

[13] I. Esteban, M. C. Gonzalez-Garcia, M. Maltoni, I. Martinez-Soler and Th. Schwetz,

Updated fit to three neutrino mixing: exploring the accelerator-reactor
complementarity. JHEP 01, 087 (2017).

[14] Harrison, P.F. and Perkins, D.H. and Scott, W.G., Tri-bimaximal mixing and the
neutrino oscillation data. Phys.Lett. B530, 167 (2002), hep-ph/0202074.

[15] Lenz, A. and others, Anatomy of New Physics in B-Bbar mixing. Phys. Rev.
D83, 036004 (2011), 1008.1593.

[16] W.E. Burcham and M. Jobes, Nuclear and Particle Physics. Longman House,
Essex, (England), 1995.

[17] T. Inami and C. S. Lim, Effects of superheavy quarks and leptons in low-energy
weak processes KL → µ+ νµ , KL → µ− ν̄µ , K + → π + ν ν̄ and K 0 ↔ K̄ 0 . Prog. Theor.
Phys. 65, 297 (1981).

[18] C. Gay, B mixing. Ann. Rev. Nucl. Part. Sci. 50, 577 (2000).

[19] M. Baak, Measurement of CKM-angle γ with charmed B 0 meson decays. Ph.D.

Thesis, Free University Amsterdam, 2007.

[20] I.Y. Bigi and A.I. Sanda, Notes on the observability of CP violations in B decays.
Nucl. Phys. B193, 85 (1981).

[21] BaBar Coll., Aubert et al., Improved measurement of CP asymmetries in

B 0 → (cc̄)K 0∗ decays. Phys. Rev. Lett. 94, 161803 (2005).

[22] Amhis, Yasmine Sara and others, Averages of b-hadron, c-hadron, and τ -lepton
properties as of 2018 (2019), 1909.12524.

[23] J.H. Christenson et al., Evidence for the 2-π decay of the K20 meson. Phys. Rev.
Lett. 13, 138 (1964).

[24] Particle Data Group, C. Amsler et al., Review of particle physics. Phys. Lett.
B667, 1 (2008).

[25] CPLEAR Coll., A.Angelopoulos, A determination of the cp violation parameter

eta+- from the decay of strangeness-tagged neutral kaons.

[26] CKMfitter Coll., J. Charles et al., CP violation and the CKM matrix: Assessing the
impact of the asymmetric B factories. Eur. Phys. J. C41, 1 (2005).

[27] UTfit Coll., M. Bona et al., The unitarity triangle fit in the standard model and
hadronic parameters from lattice QCD: A reappraisal after the measurements of
∆ms and BR(B → τ ν). JHEP 0610, 81 (2006).

[28] Heavy Flavor Averaging Group (HFAG), E. Barberio, et al., Averages of b-hadron
and c-hadron properties at the end of 2007. Preprint arXiv:0808.1297v3, 2008.
References 83

[29] G. Buchalla, A.J. Buras and M.E. Lautenbacher, Weak decays beyond leading
logarithms. Rev. Mod. Phys. 68, 1125 (1996).

[30] G. Colin, B mixing. Ann. Rev. Nucl. Part. Sci. 50, 577 (2000).

[31] ARGUS Coll., H. Albrecht et al., Observation of B 0 - anti-B 0 mixing. Phys. Rev.
Lett. B192, 245 (1987).

[32] UTfit Coll., M. Bona et al., First evidence of new physics in b↔s transitions.
Preprint arXiv:0803.0659, 2008.

[33] L. Silvestrini, First evidence of new physics in Bs mixing and its implications.
Nucl. Phys. Proc. Suppl. 185, 41 (2008).
84 References

Acknowledgments

Special thanks to Tristan du Pree, Serena Oggero, Daan van Eijk, Jeroen van Leerdam,
Kristof De Bruyn, Meike de With, Rob Knegjens and Lera Lukashenko for spotting mis-
takes in these lecture notes, and for suggesting many improvements!