The AMS-02 experiment: rst data and performances.ams.pg.infn.it/sites/default/files/TESI/Tesi_Palermo_AMS.pdf · AMS-02 has been designed to perform high precision measurements of

UNIVERSITA DEGLI STUDI DI PERUGIA

Facolta di scienze Matematiche, Fisiche e Naturali

Corso di Laurea Specialistica in Fisica delle Particelle Elementari

Tesi di Laurea

The AMS-02 experiment:

first data and performances.

Laureando

Matteo Palermo

Relatori

Prof.ssa Bruna Bertucci Dott. Jose Luis Bazo Alba

Anno accademico 2010-2011

Contents

Introduction 3

1 Physics motivations 5

1.1 Cosmic Rays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.1.1 CR composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

1.1.2 CR energy spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.1.3 The geomagnetic cutoff . . . . . . . . . . . . . . . . . . . . . . . . . 10

1.2 Matter-antimatter asymmetry . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.3 Dark matter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.3.1 Dark matter candidates . . . . . . . . . . . . . . . . . . . . . . . . . 16

1.3.2 Direct search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.3.3 Indirect search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2 The Alpha Magnetic Spectrometer-02 experiment 20

2.1 The Time-Of-Flight System . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.2 The Transition Radiation Detector . . . . . . . . . . . . . . . . . . . . . . . 24

2.3 The Ring Imaging Cherenkov Detector . . . . . . . . . . . . . . . . . . . . . 26

2.4 The Electromagnetic Calorimeter . . . . . . . . . . . . . . . . . . . . . . . . 27

2.5 The Anti-Coincidences Counters . . . . . . . . . . . . . . . . . . . . . . . . 29

2.6 The DAQ system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.7 The Triggering System of AMS-02 . . . . . . . . . . . . . . . . . . . . . . . 29

3 The Spectrometer: the Silicon Tracker and the Permanent Magnet 31

3.1 The Permanent Magnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.2 The Silicon Tracker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.2.1 Charge Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

3.2.2 Rigidity Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . 36

4 Tracker Efficiencies Study 39

4.1 Track Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

CONTENTS 2

4.1.1 Event Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.1.2 The TOF Road . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.1.3 Case A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

4.1.4 Case B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

4.1.5 Errors on the Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . 51

4.2 Track Reconstruction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

4.3 Event Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

4.4 Reconstruction Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

4.5 Intrinsic Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4.5.1 Noise Cluster Probability . . . . . . . . . . . . . . . . . . . . . . . . 71

4.5.2 Crosschecks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

4.6 Outcomes and Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . 80

4.7 Geometric Efficiencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

5 Preliminary Proton Flux 83

5.1 Event Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

5.2 Proton Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

5.3 Acceptance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

5.4 The Flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

5.5 Future Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

Bibliography 98

Ringraziamenti 101

Introduction

The Alpha Magnetic Spectrometer-02 (AMS-02) has been launched on May 16th 2011,

from the Kennedy Space Center-NASA, on board of the Space Shuttle Endeavour on its

last mission, the STS-134. On May 19th the detector has been successfully installed on

the International Space Station (ISS) where it is collecting data on cosmic rays (CR) in a

wide energy range: from GeV to TeV.

Among all scientific objectives of the AMS-02 experiment, the primary tasks are the search

for cosmic antimatter and the search for dark matter (DM), which represents ∼ 23% of the

mass-energy density of the observed Universe, besides to study the composition and energy

spectrum of the primary cosmic rays. AMS-02 has been designed to perform high precision

measurements of the CR fluxes, therefore fulfilling the requirements of large acceptance,

long exposure time and excellent particle identification necessary to achieve the intended

goals, AMS-02 will take data for at least 10 years.

In order to distinguish between matter and antimatter the capability to measure the charge

sign is mandatory. In this framework, the role played by the tracker is fundamental. This

subdetector has been developed to measure the charged particles path as well as their

momentum and the charge sign. The latter is measured by combining the information

from tracker and Time of Flight.

This thesis focuses on the study of Tracker performances, focusing in particular on the

efficiencies. After an initial run selection, a preliminary proton flux measurement has been

eventually preformed.

In Chapter 1 the designed physics motivations of AMS-02 are introduced. After a brief

review on Cosmic Rays the matter-antimatter asymmetry issue is presented. Eventually,

the Dark Matter topic has been introduced: from the first observations to its possible

candidates and the way to detect them.

In Chapter 2 the AMS-02 apparatus setup is reported. The description of all its sub-

detectors but the Silicon Tracker is included.

The Silicon Tracker is deeply described in Chapter 3.

Chapter 4 and 5 describe the core of this thesis. In Chapter 4 is reported the study of

the Tracker efficiencies. Firstly the track efficiency study will be presented. Afterwards the

CONTENTS 4

procedure adopted to study the reconstruction algorithm efficiency will be illustrated and

finally the intrinsic efficiency issue is described.

In Chapter 5 the procedure used to perform a preliminary proton flux measurement and

the obtained results are presented.

Chapter 1

Physics motivations

From the study of cosmic rays (CR) it is possible to learn several facts about the universe

such as the presence of astrophysical sources, propagation and acceleration mechanisms

of cosmic rays. In particular the antimatter component (positrons, anti-protons and anti-

nuclei) represents an extremely useful tool in order to:

• measure the CR’s energy spectrum;

• study the baryon asymmetry looking for the possible presence of anti-matter do-

mains;

• search for indirect signals of dark matter (non-baryonic particles beyond the Stan-

dard Model1 (SM));

• study the Solar System’s physics and the solar modulation;

• search for exotic2 (and not) astrophysical sources. This topic will not be stressed in

this thesis.

1.1 Cosmic Rays

The Earth is continuously hit by energetic particles which either interact with the atmo-

sphere or stop in the Earth or pass through; such energetic particles are called cosmic

rays. The study of cosmic radiation and of its interaction with the Earth atmosphere and

magnetosphere is the subject of a wide and lively experimental program since the begin-

ning of last century. Ground, underground, balloon borne and space based detectors have

1The Standard Model of particle physics is a theory that describes the electromagnetic, weak, and

strong nuclear interactions, which mediate the dynamics of the known subatomic particles [1].2One of the exotic astrophysical sources are the primordial black holes [2], [3], which represent DM

candidates of non-baryonic origin.

1.1 Cosmic Rays 6

been conceived along the years to pursue this research program which ranges over different

domains of fundamental physics.

The first hints for the presence of an ionizing radiation of unknown origin date to the early

years of 1900, when anomalous discharge rates were observed in electroscopes placed in

ground based laboratories. Evidence of the extraterrestrial origin of this radiation came

in 1912 with the first pioneering measurements on balloon (Hess [4], Kohlhorster [5]) of

the increasing ionization rate with the altitude. From the 1930s to the early 1950s, the

cosmic radiation provided a natural source of high energy particles, energetic enough to

penetrate into the nucleus and produce secondaries. With the advent of particle accelera-

tors, the interest in cosmic rays as a source of high energy particles rapidly fell off, leaving

to the astrophysical domain the open questions on their nature, origin and propagation

history from their sources to the Earth. Only few decades ago, with the development of

new experimental techniques (long duration balloon flights and satellites) a new interest

on cosmic ray propagation and sources has arisen, in order not only to answer to some fun-

damental questions on the Universe and Cosmology, but also to test new theories beyond

the Standard Model.

1.1.1 CR composition

Even if it is possible to classify cosmic rays in the three following categories, usually the

word CR refers to the charged components:

• charged particles: the cosmic radiation is dominated by light nuclei. Roughly it

is composed by ∼99% protons and nuclei and ∼1% electrons. Among the hadrons,

∼79% are protons, and about ∼70% of the rest are He nuclei [6]; the remaining

∼1% are heavier nuclei (see Fig.1.1 for an all nuclei spectrum of cosmic rays). Since

these particles are charged their main feature is that their path is changed by the

magnetic field (few microgauss) which is present in the interstellar medium, thus

they don’t carry any information about the direction of their source. In Fig. 1.2

there is a schematic view of the charged cosmic rays’ path, from their sources to the

Earth’s atmosphere.

• electromagnetic radiation: covering the whole electromagnetic spectrum. Differ-

ently from charged particles, they can travel for long distances without any deflection

due to the galactic and extragalactic magnetic field. This means that they point back

to their sources. By studying them we can get the source’s direction. However, at

energies larger than ∼TeV, they are absorbed in the extragalactic background light.

This limits the maximum observable distance. The gamma rays energy band extends

from 0.5 MeV up to 100 TeV.

1.1 Cosmic Rays 7

• neutrinos: these particles are unique messengers to explore the Universe because

they have no charge and can only interact via the weak force, thus their path is not

affected by the magnetic field and their flux practically remains unchanged along the

path from the source to the Earth. These important features make neutrinos very

interesting for astroparticle research. Nevertheless these particles are not detected

by experiments like AMS-02, so they will not be taken into account in this thesis.

Again, their energy range extends (in principle) several order of magnitude even if

at present no identifiable source of high energy (E > 10 GeV) neutrinos has been

observed.

Figure 1.1: Major nuclei components of the primary cosmic radiation [6]

It is customary to define primary cosmic rays the particles accelerated in astrophysical

sources, and secondary cosmic rays the particles produced by the interaction of primaries

with interstellar gas. Electrons, protons, helium and stellar nucleosynthesis nuclei (such as

carbon, oxygen, iron) are primaries. Nuclei such as lithium, beryllium, and boron (which

are not abundant end-products of stellar nucleosynthesis) are secondaries. Antiprotons

and positrons are also in large part secondaries, with a possible primary component.

1.1 Cosmic Rays 8

Figure 1.2: Sketch of the journey of a cosmic ray from the production site to the Earth’s

atmosphere. Figure from [7]

1.1.2 CR energy spectrum

The CR energy spectrum extends several orders of magnitude, from few 108 eV/nucleon,

reaching up to energies of ∼ 1020 eV/nucleon. As it is shown in figure 1.3 the flux drops

with energy following different power laws:

Φ(Ek) = KiEk−γi (1.1)

where Ek is the kinetic energy per nucleon, Ki is a normalization factor,and γi is the

spectral index. Values of the spectral index in the range 2.5-3.1 have been observed for

the different nuclear species composing the cosmic ray flux. The presence of power laws

is an indication of the non-thermal origin of these particles because the thermal radiation

would follow the black body distribution.

The deviation from the power law, observed below 10 GeV , is a consequence of the

influence of the solar wind called solar modulation[8].

Solar modulation

The solar wind, a steady flux of charged particles outgoing from the sun prevents extra

solar cosmic rays with energy below this limit to reach the Earth vicinity. This effect can

be better understood by studing the low energy (<10 GeV) CR, in particular the cosmic

radiation intensity variation. This solar modulation is due to the Sun activity which de-

pends both on the time (11 year cycle varying from a period of maximum activity and

1.1 Cosmic Rays 9

Figure 1.3: Energy spectrum of charged cosmic rays.

maximum effect on CR to a minimum) and on the particle charge sign since at each solar

activity maximum the sun magnetic field polarity inverts. The charge-dependent effect on

the CRs has already been observed by the BESS experiment[9].

Up to energies of the order ∼100 MeV /nucleon the observed flux of cosmic rays reach-

ing the Earth is of heliospheric origin; therefore these particles are called solar energetic

particles (SEP). The latter are fully ionized nuclei present in the solar wind and in the

solar system interplanetary medium, which are accelerated by shock waves in the solar

wind flux. These shock waves are usually related to energetic solar events but can be also

originated from the interaction of the solar wind with the magnetic fields of the planets.

Most energetic SEP are produced during solar flare eruption events that are able to ac-

celerate the particles up to the energy of few GeVs [10], [11].

Another component distinguished from the SEP is contributing to the fluxes in the range

of energies of 10÷50 MeV /nucleon: the so called Anomalous Cosmic Rays (ACR). ACR

are single ionized nuclei reaching the top of the Earth’s atmosphere and are believed to

represent interstellar neutral nuclei that have drifted into the heliosphere, become ionized

1.1 Cosmic Rays 10

by the solar wind or by the UV radiation, and then accelerated at energies > 10 MeV

/nucleon, probably at the solar wind termination shock.

Later on this thesis, another important factor which must be taken into account at low

energies will be described: the geomagnetic cutoff.

Extra solar cosmic rays, usually called Galactic Cosmic Rays (GCR), represent the

dominant component of the cosmic rays flux in the Earth vicinity, for energies above

∼ 100 MeV . They are particles accelerated at distant sources, which propagate in the

galaxy through relatively weak magnetic fields experiencing electromagnetic and nuclear

interactions with photons and nuclear matter in the interstellar medium. Approaching

the heliosphere they interact with the fields carried by the solar wind, which, as stated

already, effectively modify their flux intensity up to the energy of ∼10 GeV. Above the

latter threshold, as seen, the overall GCR spectrum follows a power law with a spectral

index γ = 2.7. At the energy of ∼ 1015 eV the spectrum dips and the slope change to γ

= 3.1 this feature is called the knee. At about 3 · 1018 eV , there is another slope change

but in opposite direction, the so called ankle. The spectrum becomes hard to quantify, but

can again approximately described by γ = 2.7.

Furthermore the presumptive cutoff, known as the GZK cutoff (predicted by Greisen,

Zatsepin and Kuzmin [12],[13]), at the upper end has been observed[14].

One of the main questions in astroparticle physics is how it is possible to have such en-

ergetic particles. In 1949 Enrico Fermi proposed a model for the acceleration mechanism

of charged particles which is still valid nowadays[15]. In this model magnetized plasma

clouds interact with the charged particles (mainly protons and electrons) in a way that

each particle, by hitting the irregular magnetic field, can gain energy.

1.1.3 The geomagnetic cutoff

The name geomagnetic cutoff refers to the geomagnetic field effect on the cosmic ray

energy spectrum. It modulates the intensity of the cosmic rays approaching to Earth and

constraints the motion of the secondary particles produced in the interactions between the

primary cosmic rays and the atmosphere [16].

At magnetic equator, the magnetic field lines run nearly parallel to the spherical surface,

while at the poles they encounter the surface at nearly normal incidence. In the polar

region, low energy charged particles can therefore reach the surface following the field

lines, while in the equatorial region the same particles would be deflected by the Lorentz

force. The rigorous description of this phenomenon is available at [17]; in particular the

effects on earth-orbiting spacecrafts like AMS-02 is available at [18].

1.1 Cosmic Rays 11

The geomagnetic field is originated in the mantle part of our planet. Great differences

in temperature and chemical composition generate strong density gradients that create

ascensions currents in the magma. Combined with the rotation of the Earth, these currents

generate a set of convection cells that through a dynamo effect are believed to produce the

geomagnetic field. This field can be described, at first order, as a magnetic dipole tilted

with respect the rotation axis of ∼ 11.5 , displaced by ≈ 400 km with respect the Earth’s

center and with a magnetic moment M = 8.1 · 1025Gcm3. The dipole orientation is such

that the magnetic south pole is located near the geographic north pole at a latitude 75 N

and longitude 291. The magnetic north pole is instead near the geographic south pole, on

the border of the Antarctica. The intensity at the Earth’s surface varies from a maximum

of ∼ 0.6 G near the magnetic poles to a minimum of ∼ 0.2 G in the region of the South

Atlantic Anomaly (SAA), in between Brazil and South Africa (see Fig. 1.4).

Figure 1.4: Earth’s magnetic field intensity in 2000 from the Danish Orsted satellite. Fields

value are expressed in nT

The latter region refers to the area where the Earth’s inner Van Allen radiation belt comes

closest to the Earth’s surface. The effect is caused by the non-concentricity of the Earth

and its magnetic dipole.

The corrected geomagnetic coordinates

The corrected geomagnetic coordinates (CGM), latitude and longitude, have been intro-

duced by Gaustafsson et al. [19] in order to have a coordinate system so that the dipole

field theory can be applied to realistic distorted fields by still using the corrected longitude

and latitude. CGM coordinates of a point A in space, by definition are calculated (see Fig.

1.5) tracing the geomagnetic field-line passing in A to the dipole geomagnetic equator

(B), then returning to the same altitude along the purely dipole field line and assigning

the obtained dipole longitude and latitude (AM ) as the CGM coordinates of the starting

point.

1.1 Cosmic Rays 12

Figure 1.5: Definition of corrected geomagnetic coordinates.

Since it might happens that at near-equatorial regions the magnetic field line does not

reach the dipole equator, in order to define the CGM coordinate an approach based on a

B minimum value along the given magnetic field line is applied [19].

The important feature of these coordinates is that the relation of altitude and latitude of

a purely dipolar case (see Fig. 1.6) are preserved. This means that the field lines, defined

by (purely dipolar case):

R = R0 cos2 λdipole (1.2)

where λdipole is the dipole latitude, R0 is the altitude at the dipole’s equator in units of

Earth radii and R is the distance from the dipole center (in units of Earth radii), can be

repalced by the more realistic one:

R = LR0 cos2 λM (1.3)

where λM is the CGM-latitude and L is the McIllwain L-parameter [20].

The CGM-longitude is shifted from geographic longitude by approximately 71 depend-

ing on CGM-latitude. Furthermore, in the CGM coordinates, the magnetic field lines are

parallel to the CGM meridians.

1.2 Matter-antimatter asymmetry 13

Figure 1.6: Dipolar field line structure.

1.2 Matter-antimatter asymmetry

One of the open questions concerning the modern Big Bang theory is related to the matter-

antimatter asymmetry. Assuming that during the Big Bang matter and antimatter were

produced in equal quantities, why the universe appears to be made basically of matter?

The latter question can be formulated as follows: how is it possible to obtain, from symmet-

ric interactions between particles and anti-particles at the Big Bang epoch, an asymmetric

universe today?

The Russian physicist Andrej Sakharov answered this question pointing out three neces-

sary and sufficient conditions to obtain such asymmetry[29]:

• baryon number violation;

• decoupling from thermal equilibrium;

• CP and C symmetry violation.

While the second condition is largely satisfied during the universe’s evolutive phases, the

first one and the third one still represent a problem. The baryon number violation has

never been observed (only upper limits on the proton mean live time have been set) and

the CP-violation, even if it has been experimentally observed, is not large enough to satisfy

the third Sakharov’s condition3.

Another scenario is then possible: the one which foresees a completely symmetric universe

between matter and antimatter with the existance of antimatter domains ([31], [32]) ,

far enough from the matter ones in order to avoid the annihilation. In addition there ex-

ist theoretical arguments based on limits coming from gamma-ray observations([33],[34]),

which state that the distance from an hypothetical antimatter domain must be of the

3Direct CP violation measured at Babar experiment [30] Acp = −0.133± 0.030± 0.009

1.3 Dark matter 14

same order of magnitude of the particle horizon4. Even though the irregularity on matter

and antimatter distribution should come out in the cosmic microwave background (CMB)

structure [32], this is instead extremely isotropic. This apparent asymmetry can be inves-

tigated by searching Helium anti-nuclei in the cosmic radiation because these anti-nuclei

can not be produced in secondary processes. Since the anti-Helium expected contribution

to the CR flux is less than 10−12 [35], the observation of an anti-Helium would indicate

the antimatter domain presence in a symmetric universe. In general, the observation of

anti-nuclei with Z > 2 in the CR flux would represent an unambiguous proof of anti-star

nucleosynthesis existence in antimatter domains.

So far several balloon-borne and satellite-borne experiments have been performed in order

to find such signals like BESS[36], AMS-01[37] and PAMELA[38].

1.3 Dark matter

The existence of a non-luminous matter in the Universe is a well established result of

observational cosmology and astrophysics. Indeed, in the first half of the past century,

many experiments[21] have been performed to measure the relative contribution of Dark

Matter (DM) to the total energy density of the Universe and a lot of theoretical effort was

spent to solve the problem of its nature. The ordinary matter which is described by the

SM of particle physics represents only ∼4.6% of the Universe energy density, the ∼22.7%

is due to DM, while the rest (∼72.8%) is represented by dark energy5. The DM problem

is today one of the most interesting research topics since it implies very close connections

between supposedly distant fields of research, such as cosmology and particle physics, and

hints to a more comprehensive picture of Nature. Indeed, almost all of the proposed DM

models involve the introduction of new physics at the fundamental level, with extensions

to the Standard Model of particle physics.

The first historical evidence of DM existence is the measurement of galactic rotation

curves[22]: observation of gas clouds in galaxies shows that the rotation velocity tends to

a constant as the distance from the galaxy center increases. A rotation curve is just the

measurement of the orbital velocity of objects belonging to a gravitational system as a

function of their distance from the centre of the system itself. Since the binding force is

gravitation, a measure of orbital velocity (v) amounts to a measure of the binding mass

4The particle horizon is the maximum distance from which particles could have traveled to the observer

in the age of the Universe. It represents the boundary between the observable and the unobservable regions

of the universe [28], so its distance at the present epoch defines the size of the observable universe.5In physical cosmology, astronomy and celestial mechanics, dark energy is a hypothetical form of energy

that permeates all of space and tends to increase the rate of expansion of the universe.

1.3 Dark matter 15

(Mb) according to the formula one can easily derive from Newton’s law:

v2 =Mb ·Gr

. (1.4)

If the visible matter (i.e. stars) would have been the unique contribution to the mass of the

galaxy, then the rotation velocity outside the luminous disk would decrease proportionally

to 1/√r. Instead, in most galaxies, the rotational velocity tends to a constant for large

values of r (see Fig.1.7).

Figure 1.7: Rotation curve of the spiral galaxy NGC 6503. The dashed lines show the

rotation curve expected from the disk material (stars) alone, the dot-dashed is the one

from the dark matter halo alone [23]

This disagreement implies the existence of a dark halo6 with M(r) ∝ r.

Another method to determine the mass of a galaxy through its gravitational properties is

to measure the lensing effect of its gravitational field on the light coming from sources in

its background[24] which substantially confirm the previous result.

More recently, a strong confirmation of DM existence came from the studies about the Uni-

verse baryonic matter. The estimation of baryonic matter density in the Universe involves

different methods, all of which give the same results. The most accurate of this meth-

6A dark matter halo is a hypothetical component of a galaxy, which extends beyond the edge of the

visible galaxy and dominates the total mass. Since they consist of dark matter, halos cannot be observed

directly.

1.3 Dark matter 16

ods is based on the formation of light nuclei during Big Bang nucleosynthesis7. Recent

measurements of D/H ratio (where D stays for deuterium and H for hydrogen), together

with nucleosynthesis predictions, estimate for the barionic matter contribution ΩB ∼ 0.03,

which is also in agreement for different nuclei abundances like He-3, He-4 and Li-7 [25].

The cosmic density of (optically) luminous matter is Ωlum ∼ 0.003 ΩB, so most baryons

are optically dark, probably in the form of a diffuse intergalactic medium. The comparison

of these results with the measured matter contribution ΩM shows that most of the matter

in the Universe in not only invisible, but it also has a non-nucleonic nature.

Dark Matter has played a relevant role in the evolution of the Universe, in particularly on

the structure formation.[26] The simplest model for the generation of cosmological struc-

tures is gravitational instability acting on some small initial fluctuations, whose origin can

be explained by the theory of inflation8. Immediately after the Big Bang, all matter is rel-

ativistic (hot); during the expansion, the Universe cools down until it reaches the temper-

ature at which DM particles decouple from the rest. DM, being heavy and non-relativistic,

starts to arrange in gravitational structures: the galactic halos. When baryons decouple,

they are gravitationally attracted inside DM aggregations to form galaxies. Therefore DM

forms the seed of galaxies. A scenario with dominant relativistic (hot) particles is dis-

favoured both by observation and numerical simulations, since it implies a number of

small scale structure inferior to the observed ones. Indeed, relativistic particles tend to

diffuse from one concentration of matter to another, thus transferring energy among them

and preventing the growth of structure on small scales.

1.3.1 Dark matter candidates

The nature of DM is yet to be understood, and there are many theoretical speculations

and experimental efforts on the subject.

As already stated, the widely accepted hypothesis is that DM is a Big Bang Relic, i.e. a

specie of particle observable today, predicted by Big Bang cosmology and bringing infor-

mations about the very first epoch of our Universe. In this framework ordinary matter

with low emission can take account for DM, and indeed the presence of cold hydrogen gas

in the halo has been suggested. However such gas should have reached hydrostatic equilib-

rium during the age of the galaxies and the equation of state combined to the gravitational

7Nucleosynthesis is the process of creating new atomic nuclei from pre-existing nucleons (protons and

neutrons). It is thought that the primordial nucleons themselves were formed from the quark-gluon plasma

from the Big Bang as it cooled below two trillion degrees.8Universe epoch during which it grew exponentially.

1.3 Dark matter 17

potential, gives for the temperature [27]:

T =GMPM(r)

4kπ' 1.3 · 106K , (1.5)

where M(r) is the mass contained within the distance r from the centre of the gas cloud,

G is the Newton constant of gravitation, k the Boltzmann constant and MP is the Planck

mass9. This is not cold gas and would be detectable through X-ray emission.

Other possible sources of ordinary (i.e. baryonic) DM, are the so called MACHOs (Massive

Compact Halo Objects), essentially remnants of late stages in star evolution, like white

dwarfs, neutron stars and black holes, or forming stars that have not enough mass to ignite

the nuclear reactions (brown dwarf/Jupiter-like objects).

Several candidates to the role of non baryonic DM have been proposed over time, including

primordial black holes, (i.e. formed before BBN), massive neutrinos, axions and Weakly

Interacting Massive Particles (WIMPs). The latters are the favourite candidates for the

non-baryonic component of universe matter-energy density. Under the name of WIMPs

are classified the particles other than neutrinos and axions that have the characteristics

to make a good DM candidate; namely they are:

• non-baryonic,

• long lived with respect to the Universe age or stable (otherwise they would have

decayed already),

• present as a relic population,

• massive (typically mχ = 10 GeV ÷ 1 TeV [39],[40]) hence,

• non relativistic at decoupling,

• their cross sections with ordinary matter are approximately of order of the weak

strength,

• electrically neutral,

• colorless.

Such WIMPs are predicted by several theories beyond the Standar Model of elementary

particles, like the supersymmetric model (SUSY) which represent a SM extention assum-

ing the existence of a symmetry which transforms bosons into fermion and viceversa. It

introduces a supersymmetric partner (with larger mass) for each particle of the SM.

9The Planck mass is defined such that the Gravitational potential energy between two masses MP of

separation r is equal to the energy of a photon (or graviton) of angular wavelength r, or that their ratio

equals one. Its value is MP =√

hcG' 1.2209 · 1019GeV/c2 .

1.3 Dark matter 18

The most important WIMP candidate in dark matter studies is the neutralino, the lat-

ter is a supersymmetric chargeless fermion obtained from the combination of the SUSY

partners of the neutral bosons:

χ = n1γ + n2Z + n3H01 + n4H0

2

where γ is the photon SUSY partner, Z is the Z boson one and H0i are the Higgs boson

SUSY partners, while ni are the combination coefficients.

From this combination, it is possible to obtain four different states with increasing

mass. The most interesting as dark matter candidate is the lightest one (LSP) since,

thanks to the R-parity10, it cannot decay into standard particles.

The techniques to search for WIMPs fall essentially in two categories: direct detection,

based on the measurement of the interaction of the WIMP with ordinary matter, and

indirect detection that looks for the products of WIMP pair annihilation.

A further interesting candidate for the dark matter research is the lightest Kaluza-Klein

particle[42].

1.3.2 Direct search

WIMPs are gravitationally trapped inside galaxies, with a rotational velocity relative to

the galactic center similar to that of the stars (∼220km/s at the Solar System). De-

spite they are extremely difficult to be directly detected because of their weak interaction

with matter, at these velocities WIMPs can undergo elastic scattering with nuclei. Direct

searches essentially look for the recoil of target nuclei due to interaction with the WIMP.

Recoil energies are in the range from 1keV to 100 keV, depending on WIMP mass, with

an expected rate of the order of 1 event per day per kg of detector. In order to detect these

recoils, detectors must be sensitive to keV energies. Given the latter energy range, natural

radioactivity is a major noise source so the typical direct search experiment is performed

in underground laboratories and requires the use of materials free of radioactive isotopes

to a high degree and the weakness of the interaction forces to use large amounts of target

material.

10New quantum number introduced by SUSY models defined as R = (−1)2s+3B+L with s=spin,

B=baryon number, L=lepton number. Its conservation implies the impossibility of the decay of LSP into

ordinary particles

1.3 Dark matter 19

1.3.3 Indirect search

The antiparticle component of CR represents a very important tool to study the dark

matter annihilation. Indeed, though WIMPs must be stable, nothing prevents them from

annihilating with their antiparticle; indirect searches look for the annihilation products in

the CR in order to detect an excess with respect to the abundance predicted by known

secondary production processes. Such excess of antiparticles in our galaxy with respect to

the expected flux (from secondary processes) may represents the signature of dark matter

annihilation. Nevertheless, it should be pointed out that these contributions to antiparti-

cles flux is mixed with the secondary particles, wich represent the main background for

the DM research. These secondaries are produced by CRs interactions with the interstellar

medium (ISM), so the excess will appear as distortions of the secondary antiparticles (like

positrons and anti-protons) energy spectrum. For example, the neutralinos are Majorana

fermions11, so they can annihilate with each other in the galactic halo producing standard

particles and antiparticles symmetrically[41]. Such antiparticles can be detected by exper-

iments like AMS-02.

11The property of Majorana’s particles is that they are equal to their antiparticles.

Chapter 2

The Alpha Magnetic

Spectrometer-02 experiment

In this chapter the AMS-02 experiment will be presented including its sub-detectors. The

spectrometer, which is the combination of a permanent magnet and a silicon tracking

system, will be discussed in detail in the next chapter since is the main matter of discussion

of this thesis work.

AMS-02 is a large acceptance magnetic spectrometer which is continuosly taking data on

board of the International Space Station since May 19th, 2011, along a 52 orbit at ∼ 400

km of altitude. It has been launched in May 16th from the Kennedy Space Center-NASA

in Florida (USA) on board of the Space Shuttle Endeavour being the main experiment of

its last mission (STS-134). In June 1998 another experiment, AMS-01 which was thought

as the prototype of AMS-02, flew for 10 days on board of the Space Shuttle Discovery

(NASA mission STS-91). AMS-01 measured the spectrum of protons, helia, electron and

positrons and provided new limits for the cosmic antimatter search [43]. Rather than the

prototype of AMS-02 the AMS-01 can be actually considered as a different experiment

since new components have been introduced to better recognize the particles and to ensure

redundant measurements of the particle characteristics. Indeed, this redundancy is the

main concept to be realized in order to perform such a challenging experiment. The most

important scientific goals are, as already stated, the search for antimatter of primordial

origin by looking for the presence of antinuclei, the indirect Dark Matter search by means

of antimatter channels like positrons and anti-protons. To achieve these tasks an excellent

particle identification and a strong e/p separation power (∼ 106) are required; therefore

multiple independent measurements of the particle properties are accomplished by means

of the following sub-detectors:

The Alpha Magnetic Spectrometer-02 experiment 21

• a Transition Radiation Detector (TRD) to distinguish leptons from hadrons;

• a Time of Flight (TOF) to measure the velocity of the particles and their charge;

• a spectrometer made by the composition of a permanent magnet and a silicon track-

ing system (Tracker) to measure the rigidity1 and the sign of the charge in addiction

to its absolute value Z;

• An AntiCoincidence Counter scintillator system (ACC), used as veto, to reject the

particles with a high incident angle that does not pass through the magnet aperture;

• a Ring Imaging Cherenkov detector (RICH) to measure the speed and the Z of the

particles;

• a 3-D sampling Electromagnetic Calorimeter (ECAL) to measure the particles’ en-

ergy, contributing to e/p rejection power of ∼ 104.

In addiction there is a Star Tracker that gives the orientation of the detector with respect

to the fixed stars with an accuracy of few arc seconds2 [44].

In Fig.2.1 the whole detector is shown.

Figure 2.1: Schematic view of the AMS-02 detector.

1the rigidity is defined as R = c·pZe

where c is the speed of light, p is the particle momentum, Z is the

charge absolute value and e is the electron charge.2A unit of angular measure equal to 1/3600 of a degree.

The Alpha Magnetic Spectrometer-02 experiment 22

If we consider a particle coming from the top, during its path along AMS-02 it first

encounters the 1st Tracker layer (see Fig. 2.2) then it passes into the TRD. After the

TRD, the particle comes accross the first two planes of the TOF system and enters in the

spectrometer itself passing through seven double-sided silicon layers (the so called inner

tracker) dipped into a quasi-uniform magnetic field. At the end of the spectrometer the

particle meets the last two TOF planes and goes into the RICH. Eventually it enters in

the ECAL having passed through the last Tracker layer.

Figure 2.2: Example of particle crossing AMS-02

Given this path the Z of the particle is indipendently measured four times in the TOF, up to

nine times in the Tracker and also by the RICH. While the velocity is measured in the TOF

and RICH detectors and also in TRD since it is sensitive to the boost factor γ. Nevertheless,

the particle charge sign, which is the crucial quantity in the matter/antimatter separation,

is determined only by the combination of the TOF and the bending property of the

spectrometer. From this point of view the spectrometer is the real core of AMS-02.

It must be noticed that AMS-02 can also detect photons in two different ways: either

the photon converts (γ → e+e−) in the material right before the inner tracker so we

can measure the corresponding tracks to the positron and electron, or it produces an

electromagnetic shower in the ECAL. Note that in the latter case ECAL is used as a

stand-alone detector. This represent the so called photon trigger.

2.1 The Time-Of-Flight System 23

Space operation conditions

The space operation condition represent a non negligible aspect for space-borne experi-

ments like AMS-02. Indeed several constraints are imposed by the transport on the Space

Shuttle and by the ISS environment as the weight limit of 7 tons. More important are the

very low power consumption (1.5 kW against 300 thousands electronic channels) and the

data rate transfer which is limited to 6 Mbps.

Furthermore AMS-02 has to deal with vibrations up to 150 dB during shuttle launch and

temperature variations between -30 C and +50 C in vacuum.

For all the above reasons all the sub-detectors have been tested in order to comply with

the space safety imposed by NASA and to achieve the designed physics performances.

2.1 The Time-Of-Flight System

Charged particle ionization in a scintillating medium causes molecular exitation/disexcitation

processes with the fast emission of fluorescence light (the disexcitation live time is ∼ 10−8

s). The photon collection provides a very precise timing measurement and an estimation

of the particle energy deposit. Scintillation counters placed at both ends of a particle de-

tection experiment measure the time of flight of the traversing particle, i.e. the particle

velocity, and provide the trigger for other detectors. The AMS-02 TOF system is com-

posed of 4 planes of segmented scintillation counters 2 above and 2 below the magnet (as

seen in Fig.2.1). The four planes contain, beginning from top, 8, 8, 10 and 8 scintillator

paddles. The upper and the lower TOF planes are alternatively positioned along the x

and y coordinates providing a granularity of 12x12 cm2 cells. The TOF single counter

consists of 1 cm thick polyvinyltoluene scintillator of trapezoidal (18.5-26.9x117-134 cm2 )

or rectangular (12x117-134 cm2 ) shape, at both ends coupled, via plexiglass light guides,

to 4 (rectangular) or 6 (trapezoidal) photo-multipliers (PMTs) as shown in Fig.2.3.The

main modification with respec to AMS-01 concerns the readout light guides, that had to

be curved in order to align the PMTs to the stray magnetic field, which in the proximity

of the TOF system is still intense enough to influence the PMT performance significantly.

This arrangement has been chosen in order to optimise background rejection at trigger

level and to help in offline track reconstruction, providing an estimate of the positions

where the particle enters and leaves the volume occupied by the inner Tracker.

The timing resolution on the time-of-flight is ' 160 ps for protons and ' 100 ps for

particles with Z≥2. The resulting resolution on velocity β is σβ/β ' 3% for protons and

σβ/β ' 1% for ions. Moreover this system has a capability of discrimination between

downward/upward going particles at the level of 10−9.

2.2 The Transition Radiation Detector 24

Figure 2.3: Top panel: schematic design of the upper (left) and lower (right) TOF planes.

Bottom panel:the upper (left) and the lower (right) TOF planes

The energy deposition in a single TOF counter gives also an evaluation of the particle

charge exploiting the relation between the energy loss by ionization (∝ Z2 ) and the pro-

duced scintillation light. Therefore the TOF system will provide four independent charge

measurements.

Furthermore the TOF provides the charged particles fast trigger signal for the whole ap-

paratus. This important feature of the TOF will be used for the track efficiency anlaysis

as will be described in Chapter 4.

2.2 The Transition Radiation Detector

The TRD (see Fig. 2.4) use the transition radiation, or rather the electromagnetic radi-

ation produced when charged particles traverse the boundary between two media with

different dielectric constants. The energy of the emitted photos is proportional to the

Lorentz gamma factor for the particles which cross the dielectric surfaces.

The main feature of the transition radiation is the presence of a threshold with respect to

the Lorentz factor. Protons and electrons with the same energy can be than distinguished.

Indeed, while electrons and positrons of high energy are relativistic, protons are under

threshold up to energy ∼300 GeV.

Since the probability for a particle to emit one photon in a single transition is partic-

ularly small (∼ 10−2) the TRD is made by 20 layers, 20 mm thick, of fleece radiator

2.2 The Transition Radiation Detector 25

Figure 2.4: The TRD detector: it has the shape of truncated octagonal pyramid.

(polypropylene/polyethylene) interleaved with straw tubes, 6 mm think (see Fig. 2.5) ,

which detect the emitted photons. The straw tubes are filled with a Xe:CO2 (80%:20%)

gas mixture operating in full-avalanche mode (∼1500V)[45]. These straw tubes, which are

homogeneously distribuited among the radiator, are actually the peculiarity of the TRD

since their presence removes the necessity of external detectors for photons. In total there

are 5248 straw tubes, arranged alternatively along the X and Y axis of AMS; the lower

and upper four layers measure the Y coordinate, while the 12 inner layers measure the

X-coordinate. This structure provides a 3D tracking capability.

Figure 2.5: Layout and operating principle scheme of one TRD layer.

Given that one layer radiator thickness corresponds to 0.06 g/cm3 density, the multi layer

structure enhances the photon yield significantly, up to 50% for 5 GeV electrons.

By using the TRD it is possible to obtain an electron/proton rejection power of ∼ 102 for

protons up to 500 GeV, with 90% electron efficiency.

2.3 The Ring Imaging Cherenkov Detector 26

2.3 The Ring Imaging Cherenkov Detector

The RICH detects the Cherenkov radiation produced when charged particles traverse a

dielectric medium with velocity greater than the light one in the same medium. The effect

of these particles results in a radiation cone (the so called Cherenkov cone) along the

particle trajectory. The properties of the cone depend both on the particles velocity and

the refractive index n(ω) of the dielectric material. The relation between these quantities

is represented by the Cherenkov relation 2.1:

cos θC =1

n(ω)β(2.1)

where θC represents the photons’ emission angle with respect to the particle path, called

the Cherenkov angle, and ω is the frequency of the emitted radiation.

The measurement of the particle velocity β is performed by measuring the radius, which

is related to θC of the ring produced on the photon detector. Furthermore the RICH

provides the charge measurement of the incoming particle by measuring the number of

emitted photons Nγ in a certain frequency range, once the particle has crossed a thickness

dx inside the medium:d2Nγ

dxdω= αZ2 sin2 θC . (2.2)

The RICH of AMS-02 is placed under the spectrometer, between the lower TOF and

the ECAL (see Fig.2.1). In figure 2.6 the main structure components of RICH are shown. It

has a truncated conical shape with 60 cm and 67 cm of upper and lower radius respectively

and a height of 47 cm.

Figure 2.6: The RICH structure.

It is composed by three main parts: the radiator, the conical mirror and the photon

detection plane. The radiator is 3 cm thick and it is placed on the upper part of the

2.4 The Electromagnetic Calorimeter 27

detector. It is made by two different materials: 80 blocks of silica aerogel 3 (11.5x11.5x3

cm3 each) for the external part and 16 blocks of NaF in the central zone (8.5x8.5x0.5 cm3

each). The choice of two materials with different refractive indexes (n=1.336 for NaF, 1.04

for aerogel) is led by the fact that in the detection plane there is a hole in correspondance of

the ECAL to let particles go unaffected into it, and the NaF which has a bigger refractive

index guarantees large θC , thus the cone is wider and thus easier to be detected (see

Fig.2.7).

Figure 2.7: The RICH photon detection.

The detection plane is made by 680 photomultipliers (PMT) with 16 pixel (4x4 mm2)

each one. From the spatial coordinates of PMTs and from the direction of the incoming

particle, the Cherenkov cone is reconstructed and therefore the velocity is determined.

Note that upgoing particles do not leave signals in the RICH.

The conical mirror is around the radiator. It reflects the photons which do not go directly

to the photon detector plane.

The velocity resolution σβ/β is around 0.1%/Z. Furthermore the RICH can measure the

charge up to Z=26 with a charge confusion smaller than 20%.

2.4 The Electromagnetic Calorimeter

The ECAL is a fine grained lead-scintillating fiber sampling calorimeter conceived to

perform an accurate 3-D imaging of the longitudinal and lateral shower development.

Its average density is 6.8 g/cm3, for a total thickness of 16.65 cm which correspond to

approximately 17 radiation lenghts and to a total weight of 496 kg.

3The aerogel is a mixture of m(SiO2) and 2m(H2O), where m is an integer. It has a porous structure

with bubbles, most of them of small size compared to optical wavelengths.

2.4 The Electromagnetic Calorimeter 28

It is arranged in superlayers, 18.5 mm thick, each one made by 9 grooved lead foils of 11

mm of thickness interleaved with scintillating fibers, 1 mm thick. The resulting structure

Figure 2.8: The assembled superlayers.

has an active area 648x648 mm2. The detector imaging capability is obtained by stacking

superlayers with fibers alternatively parallel to X-axis (4 layers) and Y-axis (5 layers) as

shown in Fig.2.8.

Each superlayer is readout by 36 PMTs, arranged alternatively on the two opposite ends.

Each PMT covers an active area of 9x9mm2, corresponding to∼ 35 fibers; this area is called

cell which represents the minimum detection unit (see Fig.2.9). The whole calorimeter is

then made up of 1296 cells for 324 PMTs providing an accurate 3-D imaging of the

longitudinal shower profile. As already stated ECAL is used to measure the energy of

Figure 2.9: Cross section of the ECAL lead-fiber structure.

particles and to contribute for an e/p separation of ∼ 104. The energy resolution has been

measured:σEE

=9.9%√E(GeV )

⊕ 1.5% . (2.3)

The calorimeter also provides a stand-alone photon trigger. The trigger efficiency is 90%

at 2 GeV and more than 10% for energy larger than 10 GeV.

2.5 The Anti-Coincidences Counters 29

2.5 The Anti-Coincidences Counters

The ACC system is composed by 16 scintillation paddles of 8 mm thickness arranged on

a cylinder that covers completely the inner Tracker. Each paddle is 220 mm wide and is

readout by PMTs from both sides. Since the ACC is located inside the magnetic field, the

PMTs cannot be placed in direct contact with the paddles. Therefore the light coming

from the scintillation panels is collected in wavelenght-shifting fibers of 1 mm diameter

and then routed through clear fibers to the PMTs which are than placed near the outer

end of the magnet.

This subdetector is necessary to ensure that no particles enter inside the spectrometer from

the sides. Therefore the ACC provides a fast veto trigger to suppress triggers originating

by secondaries particles produced by the interaction with the detector support.

2.6 The DAQ system

All the AMS electronic boards are constructed and installed in multiple copies. In the

case of failure of the primary board the system will automatically switch to the use of

a secondary board. One of the most important basic boards is the Power Distribution

System (PDS), the board that feeds with the power coming from the ISS photovoltaic

arrays all the AMS-02 electronics. The JMDC (Main DAQ Computer) sends commands

and receives replies from all boards. In the case of the special importance of the JMDC

board there are a four copy redundancy. Despite the specific requirements imposed by

physics are different for each sub-detector, an unified approach has been adopted for their

DAQ electronics (see Fig.2.10). Analog signals from the detectors are digitized, typically

through an Analog-to-Digtal Converter (ADC), and compressed in Data Reduction boards

(TDR for Tracker, RDR for RICH, EDR for ECAL, etc.). The next node in this tree, the

JINF, receives data from up to 24 xDR. In the JINF data from the xDRs are collated,

buffered and sent to the top level JINJ boards. The JINJ collates, buffers and passes data

to a JMDC. The JMDC receives the complete event and analyses it to ensure that it might

contains interesting physics, monitoring also the detector performance. The selected events

are then buffered and sent out the HRDL (High Rate Dynamic Link) when they become

available.

2.7 The Triggering System of AMS-02

AMS-02 is equipped of triggers for charged particles thanks to the coincidence of signals

coming from the TOF. ACC system is used as veto counters for particles out of AMS

field of view. The ECAL provides the stand alone trigger for photons. Combining the

2.7 The Triggering System of AMS-02 30

Figure 2.10: The DAQ system of AMS-02.

signals from the three above sub-detectors, Fast Trigger FT and Level1 LVL1 triggers are

generated. The estimated LVL1 triggers rate runs from 200 Hz to 2000 Hz, depending on

the geomagnetic latitude.

Chapter 3

The Spectrometer: the Silicon

Tracker and the Permanent

Magnet

The core of AMS-02 is composed by the combination of the silicon Tracker and the per-

manent magnet, namely the spectrometer. The capability of the Tracker to describe the

particles path along AMS and the capability of the magnet to deflect the particle trajec-

tory by means of a quasi-uniform magnetic field lead AMS to measure the momentum,

the charge absolute value and its sign. The faculty of measuring the sign of the particle

charge is crucial for AMS to achieve its scientific goals since the main difference between

matter and antimatter is actually the charge sign. Therefore this feature of the experiment

requires to be described with accuracy.

3.1 The Permanent Magnet

The magnet has a cylindrical shape, a length of ∼ 1 m, an inner diameter of 1115 mm and

an outer diameter 1298 mm. It is made of 6400 Nd-Fe-B blocks, 5x5x2.5 cm3, arranged

in 64 sections (see Fig.3.1). This configuration produced a dipole quasi-uniform magnetic

field of 1.5 T along the X axis and a negligible dipole moment in order to avoid mechanical

torques and interferences with elctronics: the external residual field is below 2·10−2 T. The

resulting bending power for the inner Tracker region (namely from layer 2 to layer 8) is

BL2=1.5 Tm2. The total weight of the magnet including the support structure is 2.2 tons.

3.2 The Silicon Tracker 32

Figure 3.1: Permanent magnet scheme.

3.2 The Silicon Tracker

The AMS-02 Silicon Tracker is made by 9 layers of double sided micro strips silicon sen-

sors, arranged in 6 planes as shown in figure 3.2. The first layer is placed on top of the

TRD, the layers from 3 to 8 are placed inside the magnet volume and thus are dipped into

the magnetic field, while layer number 2 is placed at the border of the magnet volume.

The last layer (number 9) is placed in between the RICH and ECAL. Each plane located

Figure 3.2: The Silicon Tracker layout.

inside (outside) the magnet volume has a composite structure with two 220 (700) µm thick


layers of Carbon fiber surrounding a 12 (40) µm thick, low density Aluminum honeycomb

interior of density ρ = 16.02 (32.0) kg/m3. Each one of the 9 layers has a diameter of ∼1 m and contains approximately 20 ladders. The ladder represents the basic unit of the

Tracker (see Fig. 3.3). Each ladder is composed of double sided silicon sensors (from 7 to

15) grouped along the X-direction and coupled to a read out chain characterized by a low

power consumption (∼0.7 mW per channel), a low noise and a large dynamic range. Each

Figure 3.3: Exploded view of a ladder.

double sided silicon sensor has dimensions of 72.045 x 41.360 x 0.300 mm3. It consists of

a high resistivity n-type bulk, with p+ and n+ strip implantations running in orthogonal

directions on the opposite faces of the sensor, with an inter-strip implantation (readout)

pitch of 27.5 (110) µm and 104 (208) µm for the p and n side respectively. The finer pitch

p-side strips are used to measure the bending coordinate corresponding to the Y-axis in

the AMS reference frame, while the orthogonal n-side strips measure the X-coordinate.

There are, in total, ∼ 2300 silicon sensors arranged on 192 ladders distributed along the

nine layers, from 1 to 9 as follows: 26, 22, 22, 22, 20, 20, 22, 22 and 16, resulting in a global

sensitive active area of 6.4 m2.

To minimize the amount of material in the sensitive region of the detector and reduce

the geometric inefficiencies, the front-end electronics (the hybrids) is located at the end

of the ladder, tilted of 90 with respect to the silicon plane. A metalized kapton cable is

used to route the n-side signals to the same ladder end of the p-side. In order to read

out all of the sensors of the n-side the bonding scheme shown in Fig.3.4 (bottom right)

was adopted. This kind of connection introduces an ambiguity on the determination of


Figure 3.4: The ladder sides. Top left:the ladder p-side view. Top right: the ladder n-side

view. Bottom left: the p-side bonding scheme. Bottom right: the n-side bonding scheme

the X-coordinate that must be solved during the reconstruction phase using the spatial

information provided by the TOF system. The silicon sensors of each ladder are held by

a 5 mm thick foam support that is glued to the n-side kapton cable. The exposed surface

of the foam is covered with a 100 µm thick layer of carbon fibre. Small aluminium frames

are glued to the carbon fibre surface and are provided with screw fixation holes to attach

the ladder on the mechanical structure of a plane.

Each ladder is then readout by 1024 high dynamic range, AC coupled readout channels,

640 for the p-side and 384 for the n-side, for a total of 196k readout channels, correspond-

ing to ∼ 3 Mbit raw data per event. Data compression is therefore mandatory in order

to keep manageable the event size for the acquisition. This task is performed by a set of

Tracker Data Reduction (TRD) boards, each of whom processes the signal sequentially

read from one ladder. The 1024 channels are read by chips, called VA, which are basically

pre-amplifiers. Each VA has 64 channel, therefore there are 10 VA and 6 VA for X and Y

side respectively. Calibration runs are performed every 46 minutes to monitor the average

signal level of each channel in absence of energy deposit (pedestal) and its spread (noise).

A compression algorithm running on the TRD Digital Signal Processor is used to select

channels where the pedestal subtracted signal level is greater than three times its char-

acteristic noise. Neighbour channels are also readout in order to allow a more accurate

position measurement (more details are available at [46]).

When a ionizing particle crosses a silicon detector electron/hole pairs (∼104) are pro-

duced. Charges are rapidly drifted (∼ 10 ns) by the sensor electric field, generated by the

inverse bias regime, toward the segmented electrode strips. The ionisation loss of singly

charged particles traversing the fully depleted, reverse-biased 300 µm sensor, is described


by a Landau distribution, with the peak signal given by the specific dE/dx of the particle

in silicon as resulting from the Bethe-Bloch parametrisation. This results in a peak en-

ergy loss for a minimum ionising particle which corresponds to the generation of ∼ 22000

hole/electron pairs in the silicon. Due to the reverse bias applied to the sensor using the

punch-through mechanism, the holes (electrons) drift rapidly to the p (n) surfaces where

the accumulated charges on the readout strips is fed to the front-end electronics (see Fig.

3.5). The obtained signal is proportional to the energy deposit and identifies the coordinate

Figure 3.5: The double sided silicon strip sensors: schematic view

of the traversing particle. At the single sensor level, the position resolution is determined

by the sampling pitch and the signal-to-noise performance.

While the three outermost planes (namely layer 1,2 and 9) are equipped with a single layer

of ladders, the three innermost planes are equipped with double layers of ladders. Such

configuration minimizes the uncertainties on rigidity measurements due to interactions in

the detector material.

The Tracker system is also composed by the so called Tracker Thermal Control System

(TTCS), a two phases (liquid, gas) CO2 heat exchanging system that cools down the

Tracker electronics. The CO2 at about 80 bar pressure, is circulated by a pump. It enters

the Tracker volume at a temperature just below the boiling point, and passes by thermal

bars only on the innermost planes, where the heat from front-end electronics is collected

in series. At each input, a small fraction of the fluid is evaporated. The presence of two

loops, on the upper and lower plane of the inner Tracker, allows an homogeneous cooling

of the Tracker with a minimum amount of material in the tracking volume.


3.2.1 Charge Measurement

In AMS-02 the absolute value of the charge is measured independently in each of the 9

layers (besides in the four planes of TOF and with the RICH). This Tracker capability to

measure Z through its ionisation energy loss, allows the identification of the various nuclei

species in the CR.

According to the Bethe-Block formula:

−dEdx

= kZ2 z

Aβ2

[1

2ln

2mec2β2γ2TmaxI2

− β2 − δ(βγ)

2

](3.1)

where:

• x is the amount of traversed material in g/cm2;

• k ' 0.307 MeV g−1 cm2;

• z and A are the atomic number and the atomic weight of the absorbing material

respectively;

• I is the mean excitation potential of the material;

• Z,β and Tmax are the incoming particle charge, speed and maximum energy trans-

ferable in a collision with an electron;

• δ(βγ)/2 is the so called density effect1 correction;

due to the quadratic dependence on the charge of dE/dx, the readout electronics must

have an appropriate dynamic range to measure the charge of nuclei beyond carbon, while

keeping a good signal-to-noise ratio for minimum ionizing particles.

3.2.2 Rigidity Measurement

A usual technique used to measure the charged particles momentum is to measure the

deflection of the particle trajectory by means of a magnetic field as follows. When a

particle, with charge Ze, enters inside a region with a uniform magnetic field B it suffers

of the Lorentz force2

F = Zeγv ∧B .

1The density effect arises from the fact that electric field of the particles also tends to polarize the atoms

along its path. Because of this polarization, electrons far from the path of the particle will be shielded from

the full electric field intensity. Collisions with these outer lying electrons will therefore contribute less to

the total energy loss. This effect becomes more important as the particle energy increases.2Expressed in units where c=1


Since the momentum is defined as p = γmv than we have:

F =Zep

m∧B . (3.2)

Because of the properties of the vector product the only momentum component which

should be taken into account is the one perpendicular to the magnetic field B, namely p⊥.

The Lorentz force acts as a centripetal force, therefore from Eq. 3.2 we have:

ZepB

m= m

(γv)2

r(3.3)

where r is the curvature radius. The latter tells us that in order to measure the momentum

the curvature radius must be measured:

p =r

ZeB. (3.4)

The magnetic rigidity R which is the quantity measured by the Tracker in AMS-02 is

defined as

R =p

Ze= Br from Eq. 3.4 . (3.5)

To measure the particle trajectory at least three points measurements are needed and

using the sagitta method the rigidity can be obtained. Following conventions presented in

figure 3.6 where S is the sagitta, L the distance between the first and the third position

measurement plane and θ the angle of circular sector, in the small angle approximation

we have:

L = 2r sinθ

2≈ rθ (3.6)

S = r(1− cosθ

2) ≈ rθ

2

8=BL2

8R. (3.7)

In the minimal case of only three positions measurements, the sagitta is:

S = x2 −x1 + x3

2(3.8)

and assuming that all the three points are measured independently with the same precision

σx the error on the sagitta measurement is

σS =

√3

2σx . (3.9)

From the relation 3.7 sagitta and rigidity have the same relative error, resulting:

σRR

=σSS

=

√3

2

8R

BL2σx . (3.10)

This tells us that the resolution is getting worse with the increasing of rigidity.

In the case of AMS-02 it has been decided that there should be at least four position

mesurements up to a maximum of nine in order to reconstruct a track and consequently

the magnetic rigidity.


Figure 3.6: Scheme of the sagitta method

Tracker performances

As already stated the spatial resolution for the non-bending coordinate, namely X, is

∼ 30µm while for the bending coordinate, namely Y, is ∼ 10µm. The rigidity resolution

has been evaluated at a beam test to be 28% for R=400 GV. The Maximum Detectable

Rigidity3 (MDR) is ∼ 220 GeV for the inner Tracker configuration and ∼ 2.2 TeV for

the full span configuration, thus with the all nine positions measuraments available. The

geometric acceptance for the inner Tracker is 0.5 m2 sr and for the full span configuration is

0.04 m2 sr. The inner Tracker bending power, given a magnetic field magnitude of B=0.15

T, is BL2=0.15 T m2.

3The Maximum Detectable Rigidity is the rigidity for which the relative error on it is 100%.

Chapter 4

Tracker Efficiencies Study

In this chapter the study of the Tracker system efficiency will be presented. Three aspects,

in particular, will be stressed: the track efficiency, the reconstruction algorithm efficiency

and the intrinsic efficiency.

In order to achieve its scientific goals, AMS-02 has to measure the particles fluxes. To

measure the flux of a certain kind of particle an experiment like AMS-02 measures the

number of events that have been identified to belong to that particle type. The relation

between the counting events Ndet and the incident flux Φ(E,Ω) (from [47]), in the case of

a stationary and isotropic flux, is:

Ndet =

∫ t0+T

t0dt

∫ ∞0

dEΦ(E)

∫Ωdω

∫Σd−→σ · r · ε(ω) (4.1)

where:

• E is the energy;

• t is the exposure time (see Chapter 5 for details);

• t0 is the time at the start of observation;

• d−→σ is the element of surface areaof the detector

• Σ is the total area of the detector;

• ω is the solid angle;

• r is the unit vector in direction ω;

• ε (ω) is the detection efficiency of the detector.

The latter equation 4.1 is usually adopted as the definition of flux. Indeed, from 4.1, we

have that the flux, for a certain energy range (E1,E2), is given by:

Φ(E1 < E < E2) =∆Ndet

Acceptance ·∆t ·∆E(4.2)

4.1 Track Efficiency 40

where:

• Acceptance=∫

Ω dω∫

Σ d−→σ · r · ε(ω)

• ∆Ndet∆t·∆E is the particles counting rate in the energy range dE.

Therefore the Acceptance represents the proportionality factor between the counting rate

and the incident flux. It depends both on geometrical factors and the detection efficiency

of the experiment (for more details see Chapter 5).

While the counting rate for a certain energy interval dE is estimated from data, the

Acceptance of the experiment is estimated from the MonteCarlo (MC). Therefore, in order

to measure properly the fluxes, we need a reliable MC or at least we should know how it

relates to the data. In this theis the track efficiency of the Tracker system has been studied

in order to tune and calibrate the MC.

In this chapter the study of the track efficiency, thus the efficiency of the Tracker system

as a whole, is first described. Afterwards, since it is affected both on the reconstruction

algorithm and by the properties of the detector itself. The study of the reconstrusction

algorithm efficiency and the intrinsic efficiency of the detector will be discussed.

4.1 Track Efficiency

The strategy used to measure the track efficiency is the following: first of all the sample

definition is required. Since the primary and simplest signal available in AMS is the charged

particle trigger, hence the information from the latter has been used to define the sample

for the track efficiency. As already stated this trigger is provided by the TOF system. With

this information a TOF track has been reconstructed, called the TOF road (see section

4.1.2).

Two different samples have been defined by using the TOF roads:

• sample A: presence of a TOF road;

• sample B: presence of a TOF road with a geometric match with respect to the

seven Tracker inner layers.

Given these two samples, two different track efficiencies have been defined as the number

of Tracker tracks over the chosen sample.

There are two reasons to build a TOF road instead of taking just the purely trigger events:

• first of all the TOF road is required to make a geometric match with the Traker

layers (sample B);

• the TOF road allows to study the track efficiency behaviour with respect to the

polar angle θTOF and the azimuth angle φTOF , as it will be described.


4.1.1 Event Selection

The event selection criteria adopted to study the track efficiency are the following:

• No SAA: the events taken inside the SAA have been excluded, since in this region

the expected trigger rate is high and consequently the DAQ efficiency significantly

decreases (see Fig. 4.1), moreover the large number of low energy particles imping-

ing in the detector are origin of multi track events which spoil the reconstruction

performance;

• TOF 4/4: this stronger trigger request was adopted since to build the TOF road

we want to have signal from all the four TOF planes in order to have a better TOF

track χ2;

• Zero ACC: the particles coming from the side of the magnet have been rejected;

• No more than one Tracker track: this is a quality request to have clean events;

• Only One Particle: the AMS particle object is defined once there is a β reconstructed

using the TOF system;

• β ≥ 0.9: only downgoing relativistic particles are used.

Figure 4.1: Left: the expected trigger rate map. Right: the DAQ efficiency map.


4.1.2 The TOF Road

As already stated the TOF is composed of four planes of segmented scintillator paddles,

alternatively oriented with respect to AMS reference frame. Each paddle is 12 cm wide

and just from the position of the paddles with respect to the AMS reference frame the

TOF provides one coordinate measurement for each plane. This means that with the two

upper planes it gives only one position (X,Y) measurement of the particle impact point.

Therefore we end up with two points (X,Y), then there is only one possible straight line

as a TOF track. Moreover the position measurements is uniformly distributed along the

hit paddle (see Fig. 4.2). The nominal value is assigned to the center of the paddle with a

Figure 4.2: Position measurement uniform distribution for the position of the paddle. The

nominal value is set in the center of the paddle position with a σ = L/√

12, where L is

the paddle width.

σ of 12/√

12 cm, according to the uniform distribution, that is much larger than the error

position given by the Tracker system. We have to take this large uncertainty into account

since we want to use some geometric properties of the TOF road, as already explained.

Due to the discrete structure of TOF, the number of possible TOF roads that we can

obtain is finite and rather small (combinatory calculation) which means that we would

not be very sensitive to the angular properties. Moreover, if each plane measures only

one of the two coordinates X and Y, then the risk of accidentals is high. Therefore in

order to minimize the accidental risk and to smear the discretization effect due to the

TOF structure we use also the time information from each hit paddle (which is readout

on both ends by PMTs). The time information provides then the measurement of the


coordinate which runs parallel to the paddle. Hence in this way we have a complete X

and Y position measurement from each TOF plane (see Fig. 4.3). This means that we

Figure 4.3: X and Y position measurement from one paddle, combining the time informa-

tion and the spatial information.

have 4 points instead of two, hence we can perform a fit to have a more realistic particle

trajectory. The fit minimizes the χ2 of the possible tracks.

Furthermore, since the position measurement given by the time information is normally

distributed around the real value (instead of the nominal value), this allows to smear the

discretization effect, although it is not completely removed as it is shown in figure 4.4.

Position Measurements with the TOF

Once the DAQ system is on, thus a common time reference is set for all the PMTs of the

TOF, by measuring the arrival time of a signal on both the TOF paddle sides (t1 and t2)

with a TDC it is possible to measure the impact position of the incoming particle, with

respect to one coordinate. Indeed we have:

t1 = t0 +x

vand t2 = t0 +

L− xv

(4.3)

where

• t0 is the particle arrival time on the TOF paddle;

• L is the paddle lenght;

• x is the distance between the impact point and one of the two sides of the paddle,

which we want to measure;


Figure 4.4: TOF road impact point position map on layer 7 after the smearing of the

discretization effect. Nevertheless the discrete structure is still visible.

• v is the speed of light inside the paddle.

Therefore, taking the difference between t1 and t2 (∆t) we have from 4.3:

x =L

2− v∆t . (4.4)


4.1.3 Case A

The sample of case A is simply the existence of a TOF road, namely a straight track built

with TOF information. Given this sample the track efficiency has been defined as follows:

• εA = number of Tracker tracksnumber of TOF roads .

The obtained results is:

εA = (73.263± 0.008)% .

Thanks to the existence of the TOF road the track efficiency behaviour with respect to the

cosine of the polar angle, cos(θTOF ), has been studied (see Fig. 4.5). The angular range is

compatible with the TOF shape and dimensions.

Furthermore the study of the track efficiency with respect to the azimuth angle, φTOF ,

has been also performed (see Fig. 4.6). As we can see from figure 4.6 the TOF structure

is clearly represented.

Figure 4.5: Track efficiency vs cos(θTOF ) with respect to the sample A. The values of

cos(θTOF ) are due to the TOF structure.

Since the X and Y dimensions of the TOF system are larger than the inner Tracker

ones, the choice of this sample A includes a geometric inefficiency contribution which we

want to avoid to better characterize the track efficiency. This contribution is particularly

visible in figure 4.5 in the drop at small angles. Indeed this inefficiency is due to those

vertical tracks which pass near the TOF edges. The same effect is responsible for the drops

on the efficiency in figure 4.6. In order to get rid of this geometric effect the sample B has

been introduced.


Figure 4.6: Track efficiency vs φTOF with respect to the sample A. The behavior reflects

the TOF structure (right).

4.1.4 Case B

The reason for a geometric match requirement of the TOF road with the inner Tracker

layers is to remove geometric inefficiencies. Indeed the TOF road has to pass through

the Tracker in order to detect a track. Note that, as it will be explained in the next

section, the minimal number of hits (see next section) to build a track is 4 and 3 of them

must belong to the double inner layers, one for each couple. From this algorithm feature

arises the requirement to match only with the inner Traker layers. In order to perform the

match the layers geometry has been studied using MC. In figure 4.7 two examples of the

layers geometry are shown. From this information the layers borders have been accurately

described.

As previously explained, the TOF road spatial resolution is not as accurate as the

Tracker system one due to the TOF paddle width. Therefore in order to have a reliable

geometric match with the inner Traker layers the distributions of the difference between the

TOF road and the Tracker track impact point prediction on those layers have been studied

(see Fig. 4.8 and 4.9). The geometric match requires that the TOF road impact point

prediction on a certain inner layer must be inside three RMS of the latter distributions

from the layer edges.

Given the sample B, the track efficiency has been defined as follows:


Figure 4.7: The geometric description of layer 6 (left) and 2 (right). The ladders structure

is visible.

• εB = number of Tracker tracksnumber of TOF roads with geometric match with the inner Tracker.

The obtained results is:

εB = (82.286± 0.014)% .

As expected, the latter track efficiency εB is larger than in case A because we removed

the geometric inefficiency contribution. This enhancement of about 11% is also visible in

figures 4.10 and 4.11. Moreover the efficiency drop at small polar angles, θTOF has been

reduced as well as the TOF structure effect with respect to the azimuth angle φTOF . The

same result arises from the study of the track efficiency behaviour with respect to geo-

magnetic latitude (see Fig. 4.12).

Figure 4.8: Difference between the TOF road and Tracker track impact point prediction on

layer 4 vs rigidity, for X-side (left) and Y-side (right). In the Y-side the effect of curvature

due to the presence of magnetic field is visible.


Figure 4.9: Difference between the TOF road and Tracker track impact point prediction

on layer 4 distributions, for X-side (left) and Y-side (right).

Figure 4.10: Track efficiency vs cosθTOF with respect to the sample A(blue) and B(orange).

The range of cos(θTOF ) for case B is smaller because of the geometric matching requests

with the inner Tracker, but the matching has also effect on the track efficiency, which

results higher.


Figure 4.11: Track efficiency vs φTOF with respect to the sample A(blue) and B(orange).

The behavior of case B is less sensitive to the TOF border structure.

Figure 4.12: Track efficiency vs geomagnetic latitude for case A (blue) and B (orange).

The track efficiency results higher close to the geomagnetic poles.

The comparison between data and MC has been also performed for the track efficiency

εB. The result obtained for MC is:

εMC = (87.59± 0.04)%

therefore, for the ratio data/MC, we have:

εDATA/MC = (0.9395± 0.0005) . (4.5)


The latter value will be used to evaluate a preliminary proton flux as described in

Chapter 5.

In figures 4.13 and 4.14 the comparison of the track efficiency behaviour against both

cosθTOF and φTOF is shown. The ratios between data and MC are substantially stable in

both cases.

Figure 4.13: Data and MC comparison of track efficiency with respect to cosθTOF (left)

and φTOF (right).

Figure 4.14: Data/MC ratio for track efficiency with respect to cosθTOF (left) and φTOF

(right).


4.1.5 Errors on the Efficiency

The errors on the efficiency reported in this analysis are purely statistical.

The ratio

ε =k

N

, where N is the total sample events and k is the number of succeses, i.e. the accepted

events, has been taken as the best estimator for the efficiency. The total sample events N

is assumed to be constant, hence the number of successes k is distributed according to the

Binomial distribution. Therefore the sigma of k is:

σ(k) =√N · ε · (1− ε)

. The resulting error on ε is then:

σ(ε) =

√ε · (1− ε)

N.

Note that the total sample events used in this anaysis is about ∼ 107.

4.2 Track Reconstruction 52

4.2 Track Reconstruction

The basic signal quantities for each side (X and Y) are the clusters. The clusters are then

combined together to form hits which are used to build the tracks.

Clusterization and hit creation

A clusterization algorithm searches for a seed strip whose signal is above the so called High

Threshold with respect to the Signal-to-Noise ratio (S/N). Then it expands the cluster

in the surroundings of the seed, stopping when reaching the so called Low Threshold on

S/N. Each cluster holds the position information on one coordinate. The cluster X and Y

must be combined into a 2-D cluster, namely the hit. Since all the possible pairs of clusters

within a ladder are made, many hits per ladder (so per layer) are created, but at most one

will be associated to the resulting track.

Track Finding

The hits are divided per ladder and per layer. The first pattern search is made on per

ladder scan, searching among all the combinations of ladders on different layers. Those

that may likely correspond to a track on the basis of a χ2-method. The search is then

made using the hits on the selected ladder set. Note that this track finding procedure is

performed on the inner Tracker only.

Track Fitting Procedure

Tracks are searched within the inner Tracker only. This means that the fit on the inner

layers (layer 2 not included) is only performed. The resulting track is then extrapolated

first to layer 2 and then to layers 1 and 9. If a hit is found either in layer 2 or layers 1 or

9, the inner track plus the new hit(s) is fitted.

Given the large distance between the innermost layers and the external layers 1 and 9, the

inaccuracy of the entrapolation on these layers, as we will see later on this chapter, can

be a source of inefficency for the tracking system. Furthermore, the multiple scattering

(MS) can also play an important role for the reconstruction efficiency of layers 1 and 9.

The latter is also affected by the backsplash1 because of its proximity to the ECAL.

Fitting Method

The fitting method is based on a path integral calculation along the expected trajectory

with Runge-Kutta tracking. For details see [48].

1The backsplash regards the extra clusters due to the particles backscattering in the ECAL.

4.3 Event Selection 53

4.3 Event Selection

Additional event selection criteria with respect to the track efficiency study have been

applied for the reconstruction and for the intrinsic efficiency in order to clean up the

sample. Only the request on the β of the particle is slightly different: only β ≥ 0 was

applied, thus down going particles.

• Events with only one Tracker track have been selected since there might be events

with more than one track due to δ-rays, fragmentation, etc.

• If there is a track it should be associated to the AMS particle object.

• Events with no more than one TRD track.

• The βTOF must be reconstructed using at least one upper and one lower TOF plane.

• Clean TOF clusters:

– only clusters whose signal come from both sides of the hit paddle are kept;

– only clusters which passed both the low and the high threshold;

– if multiple events have occurred within the TDC gate, the clusters are rejected.

• The absolute charge value measured by the Tracker should be equal to 1 or 2 for

protons and helia respectively.

4.4 Reconstruction Efficiency

In this section the study of the effect on the efficiency of the track reconstrusction algo-

rithm is reported. Since it is one of the main contributions to the track efficiency, a good

understanding and potentially a tuning of such an effect is certainly important to perform

data analysis.

The reconstruction efficiency has been studied for each layer separately and then the study

has been extended to each single ladder to find out the ladders, for each layer, which are

particularly problematic. The first and more trivial definition for the reconstruction effi-

ciency for a given layer j-th is:

• εR:the number of events with a hit associated to the track on layer j-th over the total

number of events with track.

Because of the adopted fitting procedure it is ensured that the reconstructed track passes

through at least the three double layer planes of the inner Tracker. However this is not

ensured for the external layers 1 and 9. This fact is a source of geometric inefficiency for

4.4 Reconstruction Efficiency 54

these layers. In order to remove this effect, a second and more consistent definition for the

reconstruction efficiency has been used:

• εRA:the number of events with a hit associated to the track on layer j-th over the

total number of events for which the interpolation (or extrapolation in the case of

layer 1 or 9) prediction of a refitted track is inside the so called Active Area for that

layer.

The refit is performed without the hit on the layer under study to avoid possible bias on

the resulting distribution, especially for the inner layers. The Active Area of a ladder is

defined as the area inside of a sensor, within 300 µm from the sensor edge (see Fig. 4.15

and Fig. 4.16).

Figure 4.15: Scheme of the Active Area of a ladder.

Figure 4.16: Left: Active Area for a sensor in X-side. Right:Active Area for a sensor in

Y-side.


The results of using this second efficiency definition, εRA obtained for each layer using

protons and helia respectively are listed in Table 4.17 and summarized in figure 4.18.

Layer

ε_RA Pr (%) ε_RA He (%)

1

88.1 ± 0.1

89.6 ± 0.2

2

95.94 ±0.01

97.27 ± 0.02

3

95.59 ± 0.01

96.650 ± 0.02

4

94.50 ± 0.01

96.18 ± 0.02

5

94.99 ± 0.01

96.09 ± 0.02

6

81.25 ± 0.02

84.68 ± 0.04

7

93.40 ± 0.01

95.29 ± 0.02

8

89.39 ± 0.01

92.00 ± 0.03

9

83.1 ± 0.1

84.0 ± 0.2

Figure 4.17: Reconstruction efficiency results, layer by layer

Figure 4.18: Reconstruction efficiency for each layer for protons (left) and for helia (right).

Note that the listed values refer to the second definition of the reconstructed efficiency

with a rigidity cut: events which reconstructed rigidity larger than 20 GV for layer 1 and

9 and larger than 10 GV for the inner layers. The latter choice arises from the study of

the reconstruction efficiency with respect to the rigidity which has also been performed.

Indeed, if we look at figure 4.19 and 4.20 it is clear that the mentioned rigidity cuts


have been chosen in order to take the efficiency plateau. The drop on the efficiencies at

low rigidities, except for layer 2, is due to the effect of multiple scattering. The effect of

multiple scattering is described by:

θ =

√4Lπ

αX0

Zmc

βp(4.6)

where L is the thickness of the material, X0 is the radiation lenght, p is the momentum of

the particle, β is the velocity, m is the mass, c is the speed of light, α is the electromagnetic

coupling costant and Z the charge.

Figure 4.19: The reconstruction efficiency εR (empty squares) and εRA (full triangles) vs

Rigidity for protons for layers 1(top left), 2(top right), 3(bottom right), 9(bottom right).

Due to the 1βp dependence, the MS effect is larger at low rigidities. The efficiencies then

drop because the hit might be too far from the impact point prediction to be associated

to the track. On the contrary, in case of layer 2, below 1 GV the reconstruction efficiency

is basically 100%. This is because the reconstruction algorithm forces the search for a hit

in layer 2 by expanding the search area because for low rigidities the curvature is larger.

Furthermore, from figures 4.19 and 4.20 top left and bottom right, it can be also seen


Figure 4.20: The reconstruction efficiency εR (empty squares) and εRA (full triangles) vs

Rigidity for Helia for layers 1(top left), 2(top right), 3(bottom left), 9(bottom right).

the difference between the two definitions due to the geometric effects of the track fitting

procedure. As already stated the latter effect is much more important for external layers

than for inner ones.

It should be pointed out that the rigidity which we are taking into account is evaluated

using only the inner Tracker and moreover is not evaluated with the refitted track but

with the original one. The assumption made on the use of the refitted track is that the

reconstructed rigidity with the refitted track is substantially the same of the original one.

This assumption is actually verified as shown in figure 4.21.

Taking the ratio between data and MonteCarlo (MC) we find a pretty stable behaviour,

with small statistical fluctuations. An example for protons and helia is given in figures

4.22 and 4.23. From this study arises that the MC for helia does not properly simulate

the behaviour at low rigidities, while there are no important issues regarding protons.

The behaviour of the second definition of the reconstruction efficiency (εRA) with

respect to the cos θTK (Fig. 4.24) and φTK (Fig. 4.25) has also been studied and compared

to the MC. Regarding these two variables, the MC simulates properly the data both for

protons and helia.


Figure 4.21: Refitted rigidity vs rigidity for protons (left) and for helia(right).

Figure 4.22: Reconstruction efficiency for Data (empty squares) and MC (full triangles)

vs rigidity for protons (left) and for helia(right) for layer 3.

Figure 4.23: Data/MC reconstruction efficiency ratio vs rigidity for protons (left) and for

helia (right) for layer 3.


Figure 4.24: Top panel: Data-MC comparison of εRA vs cosθTK for protons (left) and

helia (right). Bottom panel: Data/MC ratio of εRA vs cosθTK for protons (left) and helia

(right). Both panels refer to layer 8.


Figure 4.25: Top panel: Data-MC comparison of εRA vs φTK for protons (left) and helia

(right). Bottom panel: Data/MC ratio of εRA vs φTK for protons (left) and helia (right).

Both panels refer to layer 8.


A further feature that arises from the results listed in table 4.17 is that for each layer

the measured efficiency is larger for helia than for protons (see Fig. 4.26). The explanation

Figure 4.26: Difference between the reconstruction efficiency for helia and protons for each

layer.

of this effect is straightforward; given the thickness of the silicon sensors (300 µm), the

expected ADC counts corresponding to the ionization energy loss are respectively:

• Protons: around 30 ADC counts;

• Helia: around 120 ADC counts.

This is clearly visible in figure 4.27.

Figure 4.27: Signal amplitude distributions for protons (left) and helia (right).

The protons value is very close to the S/N threshold of the readout electronics, there-

fore it may happen that the protons signal is lost, so no clusters (and so no hits) are

reconstructed.


After the study layer by layer , the ladder by ladder study has been performed. The

efficiency definition used in the case of ladders is the following:

• the number of events with a hit associated to the track on the ladder i-th of the layer

j-th over the total number of events for which the interpolation prediction point of a

refitted track is inside the so called Active Area for that particular ladder.

The Tracker has in total 192 ladders distributed in the nine layers as explained in Chapter

3. The resulting efficiencies, arranged layer by layer, are shown in figures 4.28. In the latter

each point corresponds to a single ladder while in figure 4.29 the results are summarized

both for protons and helia.

Figure 4.28: Ladder reconstruction efficiencies for protons (blue) and helia (red).

The comparison with the MC was performed also for the ladder study (see Fig. 4.30).

It is then clear that the behaviour of certain ladder are not properly described by the

MC, especially those with smaller efficiency. Nevertheless, to better understand if these

ladders behaviours are due to the reconstruction algorithm or are features of the ladders

themselves we should first study the intrinsic efficiency which is described in the following

section.


Figure 4.29: Reconstruction efficiency distribution of ladders for protons (blue) and helia

(red).

Figure 4.30: Top panel: Data-MC comparison of εRA for each single ladder for protons

(left) and helia (right). Bottom panel: Data-MC ladder distributions comparison of εRA

for protons (left) and helia (right).

4.5 Intrinsic Efficiency 64

4.5 Intrinsic Efficiency

The intrinsic efficiency describes the capability of the device itself to properly detect

a particle. In order to study this efficiency we then need to get rid of the reconstruction

algorithm effects. Therefore we looked directly to the clusters instead of the hits associated

to the track. Since the clusters are quantities related to only one coordinate, a separated

study of the X and Y coordinates has been done.

Moreover, since clusters are basic signal elements, there may be also noise-like clusters

inside the Active Area previously defined. Therefore a region of confidence where we look

for the signal-like clusters (see Fig. 4.31) needs to be defined.

Figure 4.31: Schematic view of the confidence window.

For the definition of these regions the following procedure was adopted. The unsigned

residuals distributions (X and Y side) between the interpolated impact point prediction of

the refitted track inside the Active Area and the closest cluster position versus the rigidity

have been studied. Note that the use of the refitted track has been considered necessary

to have ubiased residuals.

For each layer the integral (W) from 0 to a certain value of the residual has been

evaluated as the integral (T) from 0 to infinite2 bin per bin of rigidity (see Fig.4.32):

W (j) =

∫ Residual

0(BinContent(j, Residual)) · dy (4.7)

T (j) =

∫ ∞0

(BinContent(j, Residual)) · dy (4.8)

where j is the bin index along the X-axis.

The ratio F(j)=W(j)/T(j) between these two quantities represents the fraction of clos-

est clusters which fall inside a certain confidence window with respect to all closest clusters.

2in this case infinite means the border of the Active Area


Figure 4.32: Residual distributions for X-side (left) and Y-side (right) for protons, layer 3.

In the Z-axis for each point (R,Residual) of the plot the fraction F(j) values are reported.

Therefore F(j) assumes values from 0 to 1 and is represented by the Z-axis content of figure

4.32. This fraction is therefore the efficiency for having the closest cluster inside a window

of confidence. These windows of confidence for each layer for each side have been defined

as the plateau of the latter efficiency with respect to the rigidity (see Fig. 4.33) since it

is expectet to be rigidity independent. In the definition of the window a rigidity cut is

included to avoid the enlarging at low rigidities due to the MS.

Figure 4.33: Efficiency profile with respect to the rigidity for a certain window of confidence

for X(red) and Y(blue) side respectively.

The multiple scattering effect is clearly visible in figure 4.32. The rigidity cuts are set

at 20 GV for layer 1 and 9, while at 10 GV for the others.

The chosen windows are listed in table 4.34. The windows for the external layers are larger


because the MS effect is larger, expecially for layer 1 since the particle must travel across

the TRD but also the RICH presence for layer 9 is relevant.

A further reason for these bigger windows is that the extrapolation of the refitted track

can be much less accurate, especially at low rigidities. Indeed the external layer have also

bigger rigidity cuts. Another important issue for these layers is the alignment, but we will

not deepen it. The layer 9 is also affected by the backsplash of the particle given by the

proximity to the ECAL. This contribution will be discussed later on this chapter.

Layer

X-side [μm]

Y-side [μm]

1

2000

2000

2

500

300

3

300

300

4

300

300

5

300

300

6

500

300

7

300

300

8

400

300

9

2000

2000

Figure 4.34: List of the chosen windows of confidence, for X and Y side respectively.

After the windows have been chosen the intrinsic efficiency have been defined as follows:

• the number of events for which the closest cluster falls inside the window of confidence

for layer j-th over the events for which the refitted track impact point prediction is

inside the Active Area of a ladder of the layer under study (see Fig. 4.35).

Figure 4.35: Scheme of accepted and rejected events for the intrinsic efficiency.

The assumption made on defining the intrinsic efficiency in this way is that the closest

cluster is the one that belongs to the real particle, namely is the signal-like cluster. The

reliability of this assumption has been investigated and it will be discussed later on this


section. With the given intrinsic efficiency definition the results obtained layer per layer

are listed in table 4.36 and shown in figure 4.37.

Layer

Efficiency Protons (%)

Efficiency Helia (%)

x_side

y_side

x_side

y_side

1

80.8 ± 0.1

89.0 ± 0.1

87.7 ± 0.2

92.04 ± 0.2

2

84.10 ± 0.02

89.78 ± 0.01

89.05 ± 0.03

90.95 ± 0.03

3

83.26 ± 0.02

92.27 ± 0.01

87.45 ± 0.03

93.20 ± 0.03

4

87.87 ± 0.01

91.54 ± 0.01

91.49 ± 0.03

93.03 ± 0.03

5

79.57 ± 0.02

92.83 ± 0.01

82.69 ± 0.04

93.67 ± 0.03

6

67.32 ± 0.02

92.70 ± 0.01

73.11 ± 0.05

93.70 ± 0.03

7

85.63 ± 0.02

90.52 ± 0.01

89.39 ± 0.03

92.30 ± 0.03

8

88.60 ± 0.01

91.38 ± 0.01

91.90 ± 0.03

92.82 ± 0.03

9

80.3 ± 0.1

85.0 ± 0.1

85.2± 0.2

88.0 ± 0.2

Figure 4.36: Intrinsic efficiency results X and Y sides, layer by layer.

Figure 4.37: Intrinsic efficiency, X and Y side, for each layer for protons (left) and for helia

(right).

Again, as expected, the efficiency is larger in case of helia than protons due to the


higher energy deposit. In figure 4.38 it is shown the efficiency difference between helia and

protons for each side.

Figure 4.38: Efficiency difference between helia and protons for each side respectively.

As we can see the difference is larger for X-side. This feature was actually expected

because in this side the loss of signal, from the signal generation point to the readout

electronics, is larger ( about 30%) due to the strips readout scheme. This means that in

this side it is easier to loose signal especially for proton-like signals, which are close to the

S/N threshold of the readout electronics. The comparison between data and MC has been

done. In figure 4.39 the efficiencies for MC and data together for protons (left) and for

helia (right) are shown, while in figure 4.40 the previous comparison is summarized. The

MC results less representative for X-side than Y-side


Figure 4.39: Data-MC intrinsic efficiencies comparison for each ladder for X-side (top

panel) and Y-side (bottom panel) for protons (left panel) and for helia (right panel).


Figure 4.40: Data-MC intrinsic efficiencies ladder distribution comparison for X-side (top

panel) and Y-side (bottom panel) for protons (left panel) and for helia (right panel).


4.5.1 Noise Cluster Probability

In order to investigate the reliability of the assumption made on the definition of the

intrinsic efficiency, namely that the closest cluster is the signal-like one, two methods have

been used to estimate the noise cluster probability inside the window of confidence. In

figure 4.41 the ADC counts distributions for the closest cluster (blue) and for all clusters

(red) inside the window are shown, for protons and helia respectively.

Figure 4.41: ADC counts distributions for the closest clusters (blue) and for the all clusters

(red) inside the window of confidence for protons (left) and for helia (right), layer 3.

From these figures it is clear that there might be more than one cluster inside the

window. This aspect is even more true for the external layers than for inner ones (see Fig.

4.42).

Figure 4.42: ADC counts distributions difference between the closest clusters and the all

clusters inside the window of confidence for protons, layer 1 (left) and layer 5 (right).

Because of this multi-cluster occurrence the noise cluster probability inside the widow,

thus the probability to have a noise-like cluster inside the window, has been estimated in

two different ways.


• 1st method: the noise cluster probability inside the window is defined as the number

of events with more than one cluster over the total number of events with at least

one cluster inside the window. In figure 4.43 the cluster occurrence distributions for

each layer, normalized for the number of events in that layer, is shown for X and

Y coordinate respectively. The values obtained with this method are listed in table

4.44 .

Since the windows for external layers are larger, then also the noise cluster proba-

bilities are so. This calculation gives a feeling of how probable we could have got a

noise-like cluster inside the window. Moreover an event by event correlation study

has also been done. For each layer for each side the amplitudes of all clusters inside

the window against the closest clusters’ amplitude have been studied (see Fig. 4.45).

From this plot the noise cluster probablity can be inferred as well. Indeed, by taking

the ratio between the events off-diagonal and the total events integral we got the

same values of table 4.44. Furthermore by taking the ratio between the events of the

upper part and the total events integral we got a rough estimation of the probability

that the closest cluster is noise-like (see Fig. 4.46). Since this probability is small, it

makes our assumption reliable. The obtained values are listed in table 4.47.

• 2nd method: the noise cluster probability inside the window has been evaluated

by calculating the mean noise clusters number per event per layer per ladder per

side per strip times the strip density inside the window. In practice, the mean noise

clusters number per layer per side per event (< µ >) has been estimated from

data. Afterwards, since the number of ladders per layer (nL) is known as well as the

readout channels per side (ncX = 348,ncY = 640) and the readout strip density per

side (dSX = 1/0.208 mm, dSX = 1/0.110 mm) the noise cluster probability inside

the window has been evaluated as follows:

Probability =< µ >

nLncdS · window .

In figure 4.49 the obtained < µ > values are reported. Note that these numbers are

larger for the external layers 1 and 9 because the S/N threshold are smaller. While

for layer 2 there is a contribution of δ-rays from TRD. The results for the probability

are listed in table 4.48. In Fig. 4.50 the probabilities X and Y side for both methods

are shown. With both methods the noise cluster probability inside the window is

small enough to consider the assumption reliable.


Figure 4.43: Normalized clusters occurrence distributions for X-side (left) and Y-side

(right).

Layer

Probability Protons (%)

Probability Helia (%)

x_side

y_side

x_side

y_side

1

1.00 ± 0.03

1.55 ± 0.03

2.9 ± 0.1

4.0 ± 0.1

2

(6.3 ± 0.4) 10-3

(122 ± 2) 10-3

(2.1 ± 0.2) 10-2

(303 ± 6) 10-3

3

(11 ± 5) 10-5

(8 ± 1) 10-4

(2.4 ± 1.7) 10-4

(43 ± 7) 10-4

4

(11 ± 5) 10-5

(8 ± 1) 10-4

(5 ± 2) 10-4

(39 ± 7) 10-4

5

(9 ± 5) 10-5

(8 ± 1) 10-4

(7 ± 3) 10-4

(5 ± 1) 10-3

6

(31 ± 9) 10-5

(9 ± 1) 10-4

(19.0 ± 0.5) 10-3

(4 ± 1) 10-3

7

(7 ± 4) 10-5

(26 ± 2) 10-4

(4 ± 2) 10-4

(13 ± 1) 10-3

8

(18 ± 6) 10-5

(44 ± 3) 10-4

(11 ± 3) 10-4

(19 ± 1) 10-3

9

2.47 ± 0.04

3.4 ± 0.1

6.6 ± 0.2

9.2 ± 0.2

Figure 4.44: Noise clusters probability for each layer for each side, first method.


Figure 4.45: All clusters amplitudes vs closest cluster amplitude inside the window of

confidence for protons (left) and helia (right), layer 2 Y-side.

Figure 4.46: All clusters amplitudes vs closest cluster amplitude inside the window of

confidence for protons: scheme of noise cluster probability alternative evaluation.


Layer



x_side

y_side

x_side

y_side

1

0.40 ± 0.02

0.63 ± 0.02

0.3 ± 0.04

0.46 ± 0.04

2

0.63 ± 0.02

( 9 ± 5) 10-5

(48 ± 8) 10-4

(36 ± 2) 10-3

3

(9 ± 5) 10-5

(42 ± 9) 10-5

(3 ± 1) 10-4

(19 ± 5) 10-4

4

(4 ± 3) 10-5

(37 ± 9) 10-5

(2 ± 1 ) 10-4

(13 ± 3) 10-4

5

(7 ± 4) 10-5

(5 ± 1) 10-4

(2 ± 1 ) 10-4

(24 ± 5) 10-4

6

(11 ± 6) 10-5

(5 ± 1) 10-4

(7 ± 3) 10-4

(13 ± 4) 10-4

7

0

(13.0 ± 2) 10-4

0

(48 ± 7) 10-4

8

(8 ± 4) 10-5

(21 ± 2) 10-4

(5 ±2) 10-4

(58 ±8) 10-4

9

1.28 ± 0.03

1.93 ± 0.03

1.8 ± 0.1

2.9 ± 0.1

Figure 4.47: Probability for the closest cluster to be noise-like, for each layer for each side,

for protons and helia.

Layer



x_side

y_side

x_side

y_side

1

(4468± 5) 10-4

(6479 ± 7) 10-4

(502 ± 1) 10-3

(658 ± 2) 10-3

2

(10976 ± 7) 10-5

(1594 ± 1) 10-4

(1304 ± 2) 10-4

(1702 ± 3) 10-4

3

(2180 ± 2) 10-5

(2662 ± 2) 10-5

(2691 ± 4) 10-5

(3015 ± 5) 10-5

4

(2133 ± 2) 10-5

(2677 ± 2) 10-5

(2722 ± 4) 10-5

(3123 ± 5) 10-5

5

(1909 ± 1) 10-5

(2636 ± 2) 10-5

(2332 ± 4) 10-5

(2960 ± 5) 10-5

6

(4428 ± 3) 10-5

(2936 ± 2 )10-5

(5419 ± 9) 10-5

(3375 ± 6) 10-5

7

(2526 ± 2) 10-5

(2631 ± 2) 10-5

(2969 ± 5) 10-5

(2866 ± 5) 10-5

8

(3055 ± 2) 10-5

(2637 ± 2) 10-5

(3809 ± 6) 10-5

(3009 ± 5) 10-5

9

(4508 ± 5) 10-4

(6778 ± 7) 10-4

(538 ± 1) 10-3

(724 ± 2) 10-3

Figure 4.48: Noise clusters probability for each layer for each side, second method.


Figure 4.49: Mean noise cluster numbers per layer per side per event for protons (left) and

helia (right).

Figure 4.50: Noise Cluster Probability summary for protons (left) and helia (right) whit

respect the two different methods.


By focusing on each single ladder the same procedure was applied. The obtained results

for protons (left) and helia (right) are shown in figure 4.51.

Figure 4.51: Intrinsic efficiency ladder by ladder, X and Y side, for protons (left) and helia

(right).

It is clear that there are few problematic ladders, especially for what concern the X-

side. Most of them, however, were already known not to work properly before the launch.

In figure 4.52 previous results are summarized.

Figure 4.52: Summary of the intrinsic efficiency ladder by ladder, X and Y side, for protons

(left) and helia (right).

Again, as already stated the efficiency is larger for the Y coordinate than for X, for the

same reasons given above. As we can see, most of the ladders have an intrinsic efficiency

larger than 80% and 90% for X and Y coordinate respectively. A data/MC comparison

has also been done (see Fig. 4.53).

Since, in some cases, the difference is not negligible this means that those ladders are

not properly discribed by the MC. In particular the MC is overestimating the efficiency,

especially for those ladders with lower efficiency.


Figure 4.53: The intrinsic efficiency difference between MC and DATA ladder by ladder,

X and Y side, for protons (left) and helia (right).

In order to take into account the extrapolation effect for the external layers 1 and 9, which

can be important, the windows have been set to 2500 µm with a rigidity cut at 30 GV.

Moreover, to further reduce this effect the active areas are reduced with a sensor edge

cut of 3 mm instead of 300 µm. To reduce the backsplash effect on layer 9 the following

additional requests were applied. A MIP is requested using the ECAL information. Thus

the number of hits in ECAL should be lower than 25 and the energy deposit of the shower

smaller than 0.5 GeV for protons, while for helia nECALhits≤ 100 and Edep≤ 2 GeV. In

figure 4.54 the differences between the intrinsic efficiencies after and before these additional

requests and changes are shown.

A significant improvement is then clear for layers 1 and 9. Note that, besides the

backsplash, the extrapolation effect is larger for layer 9 than layer 1 because the distance

from the closest inner Traker layer is bigger, namely the distance between layer 8 and 9 is

larger than between layer 1 and 2.


Layer0 1 2 3 4 5 6 7 8 9 10La

d. e

ff. d

iff A

fter-

Bef

ore

(%)

0

2

4

6

8

10

12

14

x-sidey-side

Layer0 1 2 3 4 5 6 7 8 9 10La

d. e

ff. d

iff A

fter-

Bef

ore

(%)

0

2

4

6

8

10

12

14

x-sidey-side

Figure 4.54: Intrinsic efficiency difference after and before the further requests of the MIP

and the Active Area, X and Y side, for protons (top) and helia (bottom).

4.5.2 Crosschecks

To further validate the reliability of the refitted track and the reconstruction algorithm

the following quantities have been estimated from data. For the refitted track reliability:

• number of tracks outside the ladder (thus in the dead interspace between ladders in

a certain layer) with the hit associated to the track for the layer under study over

the events for which the refit could be done;

• number of tracks outside the Active Area with the hit associated to the track over

the events for which the refit could be done;

and for what concern the reconstruction algorithm:

4.6 Outcomes and Improvements 80

• number of events for which the closest cluster X and Y side respectively, does not

belong to the hit associated to the track over the events for which the refitted impact

point prediction is inside the Active Area.

All these quantities are ∼ 10−1% or even less. It is possible then to claim that both the

refitted track and the reconstruction algorithm are sufficiently reliable.

4.6 Outcomes and Improvements

From the study presented above arises that the track efficiency of AMS-02 is larger than

80%. This efficiency depends both on the reconstruction algorithm efficiency and the in-

trinsic efficiency.

The reconstruction efficiency has been studied for each single layer and then the study was

extended for each single ladder, using protons and helia respectively. For most of ladders

the reconstruction efficiency is larger than 90% .

Afterwards the intrinsic efficiency has been studied as well, after excluding the reconstruc-

tion algorithm effects. The strategy used to evaluate the intrinsic efficiency allowed to

study the relative behaviour of the ladders relatively to the layer efficiency. It ends up

that the intrinsic efficiency is larger than 80% and 90% for X and Y coordinate respec-

tively.

This study allowed to understand the effect of the reconstruction algorithm, especially

of its behaviour with respect to the rigidity. Moreover we can claim that the differences

between the reconstruction efficiencies, at high rigidities, are basically due to intrinsic ef-

ficiencies of the ladder. For example the smaller value of the reconstruction efficiency for

layer 6 with respect to the inner ones is actually due to a lower intrinsic efficiency which

arises in particular from the X side. The confirmation that the X-side is less efficient than

Y-side because of the different readout scheme was obtained.

An important improvement for this work is to study first the efficiency of each single sil-

icon wafer of a ladder and then of each single VA, namely the front-end readout chip, in

order to check if the intrinsic efficiency of certain ladders is due to front-end electronics

effects or if it is a feature of the detector itself. Furthermore, since one VA read the strips

of different sensors, it is possible to understand if the inefficiencies of those sensors are

correlated because of the VA response.

Another parameter which should be taken into account is the disalignment.

4.7 Geometric Efficiencies 81

4.7 Geometric Efficiencies

The geometric efficiencies for each layer have been also studied.

• The ladder geometric efficiency: events for which the refitted track is inside a ladder

over the events for which the refit could be done.

• The sensor geometric efficiency: events for which the refitted track is inside a sensor

of a ladder over the events for which the refit could be done.

• The sensor Active Area geometric efficiency: events for which the refitted track is

inside the sensor Active Area in a ladder over the events for which the refit could be

done.

The obtained results are listed in table 4.56 and plotted in figure 4.55.

Figure 4.55: Geometric efficiencies.

Since these efficiencies are geometric the same values for protons and helia are obtained.

Note that for layer 1 and 9 the effect of the geometric acceptance lowers their efficiencies.

4.7 Geometric Efficiencies 82

Layer

εladder (%) εsensor (%)

εactive (%)

εladder (%)

εsensor (%)

εactive (%)

1

54.9 ±0.2

54.6 ±0.2

53.7 ± 0.2

55.6 ± 0.4

55.3 ±0.4

54.2 ± 0.4

2

91.45 ±0.02

88.75 ±0.02

86.93 ± 0.02

91.19 ± 0.05

88.31 ± 0.06

86.46 ± 0.06

3

97.30 ±0.01

94.69 ± 0.02

92.86 ± 0.02

97.19 ± 0.03

94.41 ± 0.04

92.46 ± 0.05

4

97.60 ± 0.01

94.51 ± 0.02

92.57 ± 0.02

97.39 ± 0.03

94.24 ± 0.03

92.26 ± 0.03

5

97.00 ± 0.01

93.78 ± 0.02

91.85 ± 0.02

96.96 ± 0.03

93.64 ± 0.03

93.43 ± 0.05

6

96.72 ± 0.01

93.70 ± 0.02

91.87 ± 0.02

96.69 ± 0.03

93.50 ± 0.04

91.62 ± 0.05

7

97.03 ± 0.01

96.06 ± 0.02

91.99 ± 0.01

96.97 ± 0.03

93.75 ± 0.04

91.83 ± 0.05

8

96.38 ± 0.01

96.06 ± 0.01

95.89 ± 0.01

96.31 ± 0.03

96.02 ± 0.03

95.86 ± 0.04

9

33.32 ± 0.01

32.96 ±0.01

32.96 ± 0.01

29.7 ± 0.4

29.3 ± 0.3

29.3 ± 0.3

Protons Helia

Figure 4.56: List of the geometric efficiencies.

Chapter 5

Preliminary Proton Flux

In this chapter the measurement technique used to perform a preliminary proton flux

estimation for different geomagnetic latitudes (θM ) is described.

The flux is defined, in general, by the following equation:

Ndet =

∫T

∫Ω

∫Σ

∫ ∞0

T (E,ω, t) ε(E,ω) Φ(E,ω, t) d−→σ · r dω dE dt (5.1)

where:

• E is the energy;

• t is the time;

• d−→σ is the element of surface area of the detector

• Σ is the total area of the detector;

• ω is the solid angle;

• r is the unit vector in direction ω;

• T(E,ω,t) is the exposure time;

• ε (E,ω) is the detection efficiency of the detector for a certain particle;

• Φ(E,ω, t) is the incident particle flux.

If the flux is stationary and isotropic we can separate the time dependent part from the

other, thus, from equation 5.1:

Ndet =

∫TT (E,ω, t)

∫Ω

∫Σ

∫ ∞0

ε(E,ω) Φ(E) d−→σ · r dω dE dt (5.2)

where we can define: ∫TT (E,ω, t) dt = ∆T (E,ω) . (5.3)

Preliminary Proton Flux 84

The latter equation 5.3 represents the integrated exposure time, thus the total effective

time that the detector was able to detect particles. Under the reasonable assumption that

the exposure time has no dependence on E and ω we can split in two contributions,

namely:

∆T = LT ·∆t (5.4)

where:

• LT represents the Live Time, namely the fraction of time for which the experiment

was actually able to detect the particles. LT values can go from 0 to 1.

• ∆t is the data acquisition time.

Therefore, under the assumption that the detection efficiency is energy independent, from

equation 5.5, for a given energy range (E1,E2), we have that:

Ndet = LT ∆t

∫ E2

E1

Φ(E) dE

∫Ω

∫Σε(ω) d−→σ · r dω (5.5)

The latter equation implies that the flux over the energy range (E1,E2) is given by:

Φ(E1 < E < E2) =∆Ndet

A(E) · LT ·∆t ·∆E(5.6)

where:

• A=∫

Ω dω∫

Σ d−→σ · r · ε(ω) is the so called acceptance.

• ∆NdetLT ·∆t·∆E is the particles counting rate in the enegy range dE.

As already stated in the beginning of the previous chapter the acceptance is the propor-

tionality factor between the incident flux and the counting rate for a given energy range

dE. It depends both on the detector geometry and efficiency. Even in the case of an unitary

efficiency the analytical calculation for a complex detector as AMS-02 is not praticable,

therefore the acceptance has been evaluated using MC data.

5.1 Event Selection 85

5.1 Event Selection

As a first step of our analysis, a global selection was applied at the run1 level in order

to reject samples corresponding to data taking periods with abnormal conditions. Typical

quantities used in this context were:

• the quality of calibration, tagging runs with abnormal numbers of bad tracker chan-

nels;

• the average livetime and event size as a function of the geomagnetic latitude, pointing

to a critical behaviour of the data acquisition chain.

We then applied an event selection in order to define the proton sample used for the

flux measurement based on the criteria listed in the following:

• No SAA: the events taken inside the SAA have been excluded (see Fig. 5.1), since

in this region the expected trigger rate is high and consequently the DAQ efficiency

significantly decreases (see Fig. 4.1), moreover the large number of low energy par-

ticles impinging in the detector are origin of multi track events which spoil the

reconstruction performance;

• TOF 4/4: signal from all the four TOF planes should be available;

• Zero ACC: the particles coming from the side of the magnet have been rejected;

• Only One Particle: the AMS particle object is defined once there is a β reconstructed

using the TOF system;

• β > 0: only downgoing particles are considered.

• Events with only one Tracker track have been selected since there might be events

with more than one track due to δ-rays, fragmentation, etc.

• If there is a track it should be associated to the AMS particle object.

• Events with no more than one TRD track.

• The βTOF must be reconstructed using at least one upper and one lower TOF plane.

• Clean TOF clusters:

– only clusters whose signal come from both sides of the hit paddle are kept;

– only clusters which passed both the low and the high threshold;

1The runs are the consecutive data taking periods.

5.2 Proton Rate 86

– if multiple events have occurred within the TDC gate, the clusters are rejected.

• The absolute charge value measured by the Tracker should be equal to 1.

• The presence of the hit on layer 2 is required to measure the rigidity in order to

improve the rigidity resolution, thus only events with a hit on layer 2 associated to

the track are analyzed.

Figure 5.1: Geomagnetic latitude vs geomagnetic longitude. The cut on the SAA is clearly

visible.

5.2 Proton Rate

As already stated, the counting rate for a given energy range is:

Rate =∆Ndet

LT ·∆t ·∆E

and it has been evaluated in ten different geomagnetic intervals defined starting from the

equatorial region (|θM | < 0.2 rad) up to the most polar region covered by the ISS orbit

(|θM | > 1 rad and equispatiated of 0.1 rad in between. The number of events Ndet is

represented by the remaining events after the selection criteria described above.

The exposure time evaluation, instead, deserves a more complete description.

5.3 Acceptance 87

The Exposure Time

The main issue on the evaluation of the exposure time is the Live Time estimation. In

AMS-02 there is a counter which looks for the status (busy or 0 and ready or 1) of the DAQ

system every 20 ns. Every second the average of this counter is evaluated and transmitted

to the DAQ chain such that the AMS offline reconstruction can uniquely associate to each

event the corresponding LT. The status of the DAQ is set to 0 (busy) between the runs.

The LT corresponding to the entire data taking interval ∆t is defined to be the mean

value of the LT distribution. Note that the latter distribution is made by only 1 event

every second, not by all the events since the LT is evaluated only every 1 second. Without

this expedient the mean LT would have been otherwise biased by the 1 second intervals

with more incoming events.

The resulting exposure time for each run is therefore < LT > ·∆t. Note that the time

spent inside the SAA, tSAA, is removed from the total ∆t. In order to be sure that all the

time spent inside the SAA has been removed, 2 seconds are subtracted from the arrival

time of the first event that belongs to the SAA region and 2 seconds are added to the last

one. In figure 5.2 and 5.3 the LT distributions for the ten geomagnetic latitude intervals

are shown.

In figure 5.4 the resulting proton rates for the different geomagnetic latitude intervals,

and the corresponding geomagnetic cutoff effect, are shown. Note that the rates are ex-

pressed with respect to the rigidity (R) instead of the energy because the Traker measures

directly R not E.

5.3 Acceptance

As already stated, because of the complexity of the shape of the AMS-02 detector and

because in the definition of acceptance given in equation 5.6 the detection efficiency is

included, the acceptance has been evaluated using the following MC approach [49]. The

evaluation is performed as the efficiency with respect to a known-acceptance shape, namely

the injection plane:

• events are generated by an injection plane with a known geometric acceptance of

Γplane = π ·Area; the plane should be big enough to cover the field of view of AMS.

• The MC is generated with a spectrum compatible with the angular distribution of

the flux that we want to measure, in this case isotropic2. Moreover also the rigidity

dependence of the generated flux is known, therefore we know the Ngen(R).

2The definition of the acceptance given in equation 5.6 actually represents the integrated acceptance

and is valid only under the assumption of an isotropic flux. Otherwise the differential acceptance A(E, θ, φ)

should be evaluated.

5.3 Acceptance 88

Figure 5.2: Live Time distributions for different θM intervals. From top to bottom, from

left to right (in radians): |θM | < 0.2 , 0.2 < |θM | < 0.3 , 0.3 < |θM | < 0.4 , 0.4 < |θM | < 0.5

, 0.5 < |θM | < 0.6 , 0.6 < |θM | < 0.7.

The resulting acceptance A(R) for a given rigidity interval dR is then given by:

A(R1 < R < R2) = Γplane ·Ndet(R)

Ngen(R)(5.7)

i.e. the geometric acceptance of the injection plane times the global detection efficiency of

AMS.

5.3 Acceptance 89

Figure 5.3: Live Time distributions for different θM intervals. From top to bottom, from

left to right (in radians): 0.7 < |θM | < 0.8 , 0.8 < |θM | < 0.9 , 0.9 < |θM | < 1 , |θM | > 1.

5.3 Acceptance 90

Figure 5.4: Proton rates for the different geomagnetic latitude unsigned intervals.

The generated flux in the MC is isotropic and has a rigidity spectrum ∼ R−1. This

means that between R1 and R2 we have:

Ngen(R1 < R < R2) =

∫ R2

R1

R−1 dR = ∆ ln R . (5.8)

By using a logarithmic binning in rigidity, therefore the generated flux appears flat (see

Fig. 5.5), thus:Ngen

∆ logR= k = cost. (5.9)

In figure 5.5 the generated rigidity (red) and the reconstructed rigidity (blue) distributions

are shown. It is clear that the reconstructed rigidity deviates from the generated one

approaching the MDR, which is around 220 GeV since we are considering only the inner

Tracker. This deviation will affect the proton flux in that rigidity region, therefore an

unfolding procedure [50] is required to a correct proton flux description. Nevertheless it

has not been applied in this analysis.

In figure 5.6 the acceptances A(R) for different rigidity intervals for different event

selection criteria configurations are reported. The black curve is related to the requirements

5.3 Acceptance 91

Figure 5.5: The generated momentum (red) and the reconstructed momentum (blue) dis-

tributions. The deviation of the reconstructed momentum from the generated approaching

the MDR is clearly visible.

of TOF4/4 and Zero Anti, the blue one has been evaluated after the request to have

only one track, associated to the AMS particle object. The difference is due not only to

geometric factors but also to the detection efficiency, which is also included in the definition

of the acceptance 5.6. The red curve represents the acceptance evaluated after the entire

selection criteria chain and it is the one that have been used to evaluate the proton flux.

The MC data are produced in two different rigidity ranges:

• low range from 500 MV to 5 GV;

• high range form 5 GV to 4 TV.

For practical reasons this splitting requires to chose a logarithmic binning that falls exactly

at 5 GV. Nevertheless there is a small step in correspondence of the boundary bin between

the two ranges (right after 5 GV). This is because the MC generation goes a bit beyond

the nominal range values and since in the low range there are less bins involved, hence

more generated events per each bin, compared to the high range the resulting acceptance

in the last bin of the low range is underestimated. Furthermore the drop of the acceptances

at low momentum has two contributions: either the particle can be absorbed or has small

radius of curvature that makes the particle to exit from the inner Tracker without passing

trhough the lower TOF planes.

5.4 The Flux 92

Figure 5.6: The acceptances for different selection criteria configurations. Black: after the

TOF4/4 and Zero Anti requirements. Blue: after the requirement to have only one track,

associated to the AMS particle object. Violet: after the request for an absolute value of

the charge equal 1. Red: after the entire selection criteria chain.

5.4 The Flux

After the evaluation of the acceptance, bin per bin in rigidity, the proton flux has been

evaluated for the different geomagnetic latitude intervals using the equation 5.6. The

obtained results are shown in figure 5.7. Note that only the statistical errors have been

taken into account.

The geomagnetic cutoff effect at low rigidities is clearly visible. As expected, moving

from the equator (θM = 0) to the geomagnetic pole (θM > 1) this effect decreases. More-

over no systematic effects are visible for the cosmic (i.e. over cutoff) proton flux, thus the

fluxes over cutoff converge to the same values as shown in figure 5.8 that reports the ratio

between the fluxes over cutoff for the different geomagnetic latitude intervals and the equa-

torial bin. This tell us that, even if each bin in θM is characterized by different < LT > the

normalization ends up to be the same. Therefore the calculation of the livetime is correct.

Since in the previous chapter we measured the track efficiency from data and we eval-

uated the ratio between the efficiency from data and MC (see equation 4.5) we corrected

the cosmic proton flux dividing bin per bin in rigidity by the obtained result. In figure 5.9

5.4 The Flux 93

Figure 5.7: Proton flux for the different geomagnetic latitude intervals. The geomagnetic

cutoff effect is clearly visible.

the comparison between the cosmic proton flux (namely the flux over cutoff) measured by

AMS-01 [51] and AMS-02 (after the latter correction) and other experiments is shown.

5.4 The Flux 94

Figure 5.8: Proton flux ratios over cutoff.

5.5 Future Improvements 95

Figure 5.9: Cosmic Proton flux measured by AMS-01 (red) and AMS-02 (blue).

5.5 Future Improvements

The primary proton flux has been measured by taking into account only the statistical

errors. A future improvement for this measurement will be the study of the systematic

uncertainties. First of all, as already said, an unfolding procedure should be applied near

the MDR because the mistag of the reconstructed momentum with respect to the real

one plays an important role as shown in figure 5.5. A systematic effect is also present

at low rigidities (under few GV) due to the fact that the reconstructed momentum is

systematically smaller than the real one because of the huge particle energy losses which

are, at these rigidities, no more negligible with respect to the total energy of the particle.

Indeed, as shown in figure 5.10, the 1Rrec− 1

Rgendistribution is highly asymmetric with

large tails on the positive side, corresponding to reconstructed rigidities systematically

lower than the generated ones. This asymmetry arises at low rigidities as it can be seen

in figure 5.11.

Other possible sources of systematic errors that should be taken into account are the

efficiency of the various event selection criteria that have been applied and the trigger

efficiency of the experiment.


Figure 5.10: 1Rrec− 1

Rgendistribution. It is not symmetric.

Figure 5.11: 1Rrec− 1

Rgenvs Pgen. The contribution of the low momenta to the asymmetry

is clear.


A further improvement will be the conversion from the geomagnetic latitude to the

corrected geomagnetic latitudes (CGM). The latter correction takes into account the fact

that the Earth magnetic field is not perfectly dipolar. This effect is visible in figure 5.12

that shows the rates for specular geomagnetic intervals with respect to the geomagnetic

equator. The full and empty dots of same colour and shape would overlap in CGM based

intervals, whereas they clearly do not.

Figure 5.12: Proton rates for the different geomagnetic latitude signed intervals.

.

Bibliography

[1] S. Weinberg, Phys. Rev. Lett. 19, 1264 (1967).

[2] Maki K, Mitsui T and Otrio S, 1996, Phys. Rev. Lett. 76 3474.

[3] B. J. Carr. Primordial black holes. Prepared for Workshop on Conference on the

Future of Theoretical Physics and Cosmology in Honor of Steven Hawking’s 60th

Birthday, Cambridge, England, 7-10 Jan 2002.

[4] V. F. Hess, Phys. Z. 13(1912) 1084

[5] W. Kohlhorster, Phys. Z. 14(1913) 1153

[6] K.Nakamura et al, (Particle Data Group), J. Phys. G 37 (2010)

[7] S. J. Stochaj, Proc. of th 27th ICRC, Hamburg(Germany) (2001),Rapp. Papers p.

136

[8] L. J. Gleeson and W. I. Axford, Astroph. J. Lett. 149 (1967) 115

[9] Clem J and Evenson P, 2007, Proc. 30th Int. Conf. on Cosmic Rays (Merida)

[10] H. Debrunner et al., Geopyhs. Res. 93 (1988) 7206;

[11] J. Steinacker et al., Astron. Astrophys. 224(1989) 259

[12] K. Greisen, Phys. Rev. Lett. 16(1966) 748

[13] G. T. Zatsepin, V. A. Kuzmin, Pis’ma Zh. Eksp. Teor. Fiz. 4(1966) 114 [JETP. Lett.

4 (1966) 78]

[14] R. U. Abbasi et al. High Resolution Fly’s Eye Collaboration, First Observation of the

Greisen-Zatsepin-Kuzmin Suppression, Phys. Rev. Lett. 100, 101101 (2008)

[15] Malcom S. Longair, High Energy Astrophysics: Volume 2, Stars, the Galaxy and the

Interstellar Medium

BIBLIOGRAPHY 99

[16] P.Zuccon, Monte Carlo simulation of the cosmic rays interactions with the near Earth

environment, PhD Thesis 2002

[17] D. J. Cooke et al., Nuovo Cimento 14 (1991) 213

[18] D.F. Smart, M.A. Shea, A review of geomagnetic cutoff rigidities for earth-orbiting

spacecraft, Advances in Space Research 36 (2005) 2012-2020

[19] G. Gustafsson et al., J. Atmos. Terr. Phys. 54 (1992) 1609

[20] H. Hilton, J. Geophys. L parameter, a new approximation Res. 28 (1971) 6952

[21] N. Jarosik et al, Seven-Year Wilkinson Microwave Anisotropy Probe (WMAP) Ob-

servations: Sky Maps, Systematic Errors, and Basic Results

[22] F. Zwicky. Spectral displacement of extra galactic nebulae. Helv. Phys. Acta, 6:110,

1933.

[23] A.H.Broeils et al, Mon. Not. R. Astr. Soc., 249:523 (1991)

[24] Y. Mellier. Cosmological applications of gravitational lensing. astro-ph/9901116, 1999.

[25] Schramm D.N, Turner N.S, Rev. Mod. Phys., 70:303-318 (1998)

[26] D. Caraffini, Anti-proton Flux Detection and Indirect Search for Dark Matter with

the AMS-02 Experiment. PhD Thesis, 2004

[27] Olive Keith A. TASI lectures on dark matter. astro-ph/0301505, 2003.

[28] Harrison E.R. (2000). Cosmology: the science of the universe. Cambridge University

Press.

[29] A. Goobar, L. Bergstrom, Cosmology and Particle Astrophysics, Springer-Praxis

Books, 2nd ed. (2004).

[30] B. Aubert et al. BABAR Collaboration Direct CP Violating Asymmetry in B0 →K+π− Decays. Phys. rev. Lett. 93, 131801 (2004)

[31] Galaktionov Yu V., Antimatter in cosmic rays, 2002 Rep. Prog. Phys. 65 1243.

[32] Cohen et. al 1998, CP Violation and the Origins of Matter, ApJ 495 539-49 15, 27-th

SLAC Summer Institute.

[33] Adams F C et al., 1997, Astrophys. J., 491, 6

[34] Cohen A G, De Rujula A and Glashow S L, 1998, Astrophys. J. 495, 539.

BIBLIOGRAPHY 100

[35] P. Evenson, Astrophys. J., 176 (1972) 797.

[36] Sasaki, M., M. Nozaki, T. Saeki, et al., A search for antihelium with the BESS spec-

trometer, Proc. 27th Intl. Cosmic Ray Conf., 1711-1714, 2001.

[37] J. Alcaraz et al. [AMS-01 Collaboration], Search for anti-helium in cosmic rays. Phys.

Lett. B 461 (1999) 387-396

[38] A.M. Galper, A.G. Mayorov and V.V. Mikhailov, 22nd European Cosmic Ray Sym-

posium in Turku, Finland, 3 - 6 August 2010

[39] Jungman G, Kamionkowski M and Griest K, 1996, Phys. Rep. 267, 195.

[40] Bertone G, Hooper D and Silk J, 2005, Phys. Rep. 405, 279

[41] Bergstrom L, Edsjo J and Ullio P, 1999, Phys. Rev. D59, 43506.

[42] Cheng H C, Feng J L and Matchev K T, 2002, Phys. Rev. Lett. 89, 211301.

[43] M. Aguilar et al. The Alpha Magnetic Spectrometer (AMS) on the International Space

Station: part I - Results from the test flight on the space shuttle. Phys. Rep. 366, pages

331-405, 2002.

[44] C. Lechaoine-Leluc. AMS, a particle spectrometer in space. NIM B 214, pages 103-109,

2004.

[45] J. Burger and S. Gentile. The AMS-02 TRD for the International Space Station. 28th

Cosmic Ray Conference, Tsukuba, Japan, 2003.

[46] A. Oliva, High Charge Cosmic Rays Measurement with the AMS-02 Silicon Tracker.

PhD Thesis, 2007

[47] J. D. Sullivan, Geometrical factor and directional response of single and multi-element

particle telescopes, Nuclear Instruments and Methods 95 (1971) 5-11.

[48] AMS Internal Note, March 18th 2003, The Alternative Track Fitting Method for AMS,

J. Alcaraz (CIEMAT).

[49] M.Duranti and A.Oliva, Private Comunications, AMS-Tracker group.

[50] G. D’Agostini, A multidimensional unfolding method based on Bayes’theorem, Meth.

in Phys. Res. A362 (1995) 487.

[51] AMS Collaboration, Protons in near earth orbit, Phys.Lett. B472 (2000) 215-226.

Ringraziamenti

E dunque giunto il momento piu difficile, quello dei ringraziamenti. La maggior parte di

voi che leggera questi ringraziamenti lo sa bene che non sono certo la cosa che mi riesce

meglio ma in tale occasione sono doverosi visti gli sforzi profusi da alcuni di voi per farmi

arrivare al traguardo.

Innanzitutto mi sento di ringraziare il babbo e la mamma che in tutti questi anni di studi,

universitari e non, mi hanno sempre sostenuto senza mai pretendere nulla da me. Grazie

per avermi insegnato l’importanza dello studio senza mai fare alcuna pressione. Grazie per

aver sopportato (giustamente non in silenzio) i miei ritmi assurdi e gli orari indecenti in

cui spesso vi ho coinvolto.

Nondimeno per quanto riguarda questo ultimo anno mi sento di ringraziare tutto il gruppo

di AMS di Perugia a cominciare dalla Prof.ssa Bertucci, anch’essa costretta ad orari in-

decenti dal sottoscritto, per avermi seguito in questo lungo(?!?) lavoro di tesi. Grazie

per avermi dato la possibilita di partecipare ad un cosı grande e ambizioso progetto e

all’entusiasmante esperienza del lancio di AMS lo scorso maggio al JSC. Grazie per il suo

sostegno e i suoi insegnamenti e grazie anche per i doverosi rimproveri e per le preziose

correzioni, queste ultime anche nei casi che riguardavano solo me e non il gruppo (penso

infatti ai preziosi consigli che mi ha dato per Monaco e per la scelta che alla fine ho fatto).

Grazie a Pepe, che ormai mi segue fin dai tempi di DESY nel luglio 2010. Grazie per la

tua presenza costante, per il tuo supporto, per i tuoi consigli ed incoraggiamenti che non

sono mai mancati durante questo ultimo anno e mezzo.

Grazie al Prof. Ambrosi per i recenti consigli, per gli insegnamenti e i suggerimenti che

mi ha dato per la presentazione della SIF e durante tutto questo periodo speso all’interno

del gruppo di AMS.

Un enorme ringraziamento va a Matteo, una delle piu grandi vittime dei miei orari. Grazie

per avermi dato la possibilita di romperti le scatole ripetutamente e assiduamente a tutte

le ore fin dall’inizio ma specialmente in questo ultimo periodo. Grazie per avermi insegnato

Root, per aver sopportato la mia scarsa dimestichezza con le librerie e i makefile. Grazie

per le risate che ce semo fatti a tarda sera nel tuo ufficio mentre mi prendevi per il c***

perche non ero buono o perche lasciavo qualche i+10 in giro. Grazie anche per i consigli

che mi hai dato nelle molteplici occasioni e in particolare per la presentazione di Monaco.

Grazie ad Alberto per il tuo supporto tecnico-metodologico e per i recenti tagli. Non capita

spesso che qualcuno venga ringraziato sul serio per i tagli, soprattutto ultimamente!

Grazie inoltre a Mimmo per le sue incursioni serali, compagno di lavoro after 8. Mi duole

ricordarti che in Italia c’e solo una grande squadra!

Grazie infine a Paolo, Sada, Nicola e tutti gli altri membri del gruppo di AMS. Un gra-

zie anche ai ragazzi della collaborazione con i quali ho speso delle belle serate in quel di

Houston.

Spero che la decisione che ho preso sia solo un arrivederci e non un addio!

Adesso passiamo invece ai ringraziamenti per gli amici, non che non ce ne siano tra

le persone gia citate ovviamente. Per cominciare permettetemi di usare un linguaggio un

po’ piu consono ai destinatari che altrimenti potrebbero annoiarsi. Dal momento che non

mi piace fare preferenze tra gli amici ho cercato un modo per apparire il piu imparziale

possibile e l’unico modo che mi viene in mente al momento e quello di nominarvi in ordine

alfabetico. L’altro modo che mi e venuto in mente era quello cronologico ma poi ho avuto

grossi problemi di pile up con i ragazzi del Pacchetto Fisici On The World per cui ho

lasciato stare.

Grazie Adro, il primo cittadino giust’appunto. Grazie per i piedi scalzi e per gli epiteti

femminili che spesso, cordialmente, ci scambiamo.

Grazie Alessio, con te le piu grosse risate sulle gaffe del Bomba.

Grazie Agnese per aver contribuito quel pomeriggio di inverno lungo il corridoio tra il

tuo ufficio e l’aula studio all’istituzione del gruppo che sicuramente passera alla storia..il

Pacchetto Fisici On The World. Grazie compagna acquisita ma anche compagna di fatto!

Grazie Andreone, new entry, per aver reso piu divertente il *MS center e anche per averlo

reso anche piu bianconero. Che altro dirti se non Forza JUVE?!?!

Grazie Aniello, caro amico particellare (uno dei tre moschettieri, quello senza baffi) con

cui ho condiviso tanto in questi lunghi anni. Dalle lotte alle innumerevoli manifestazioni,

dagli esami teorici ai laboratori...oh no scusa un laboratorio di particelle non l’hai voluto

fare con noi..dimenticavo! :-P

Grazie Antonio, compagno d’oltreoceano per l’amicizia che continua around the world!

Grazie anche a Belletti. Ecco te se l’unico oltre al Bomba che sei riuscito a far vacillare

l’ordinamento alfabetico. Hai cosı tanti soprannomi che non sapevo dove sbatterti, alla

fine ho optato per uno degli ultimi. Grazie quindi caro Bellico per la tua moltitudine.

Grazie Bomba, amico rustico col quale ormai condivido le giornate da piu di undici fottutis-

simi anni. Grazie per le risate indimenticabili delle superiori e per le altrettanto esilaranti

situazioni cui ho avuto il privilegio di partecipare in questi anni.

Grazie al Ciobo, amico da una vita. Grazie per condividere il senso di legalita e le innu-

merevoli cazzate del vecchio, ma soprattutto grazie per sopportarlo.

Grazie a quel r******ne maledetto di Danilo, amico vero sempre presente. Senza di te non

ci sarebbero cori all’altezza. Per quanto so che vari lettori non apprezzeranno io so che tu

apprezzerai sempre un bel viva la **** e forza Perugia ale!!

Grazie Debby, amica ritrovata senza la quale i doppi sensi non ne avrebbero manco uno!

Grazie ad Eleonora ed Enrico, la bella coppia dai cognomi meglio assortiti che la storia

ricordi. Grazie per l’affetto e per il piccolo Michele al quale riservo un ringraziamento

particolare che seguira.

Grazie Erro, compagno di banco delle elementari. Ultimamente stai riflando una sola dopo

l’altra ma sei sempre ’l mejo. Le risate con te non bastano mai.

Grazie Fabri, per l’amicizia e il sostegno, per i bei momenti e per quelli brutti che abbiamo

affrontato insime che fanno della nostra amicizia un’amicizia sincera e adulta.

Grazie Fiocco, compagno di sventure. Se non c’avessi ’n amico vero (ma rintronato) come

te non saprei come fare, non saprei a chi telefonare alle 1.30 di notte in mezzo alla setti-

mana per andare a ballare.

Grazie a quel buho del Fra, Compagno con la C maiuscola di lotta, di resistenza e di

divertimento, a volte un po’ azzardato :-P. Grazie che mi ricordi che c’e sempre almeno

una squadra piu scarsa della Juve degli ultimi anni. Ssse tu sssse buho dillo!!!

Grazie Giorgia, altra recent entry, senza di te il nome di first lady perderebbe di signifi-

cato.

Grazie Giulia per difendere ingiustamente il Fabri dalle accuse di f*******o. Sappiamo

tutti che ti paga! :-P

Grazie alla Giusy, che nonstante il tradimento non ha mai smesso di volermi bene.

Grazie a quella bucciotta castelana chiacchierona della Gloria. Penso di parlare a nome

di tutti i maschi, Ser compreso, se ti dico grazie per allietare (ma dove??!?!?!) le tristi

giornate nel dipartimento. ahahhaha!!

Grazie Helder Jan Ernesto Martin Adrea Dante l’unico con 5 nomi e un cognome che pare

un nome. Potrei dilungarmi molto su di te ma per pieta del lettore mi limito a dirti che

sti 5 nomi si riflettono anche sul peso e lo sai bene. PS: mo che parto ti saranno vietate le

tue scappatelle a fisica, quindi organizzati!

Caro Ilario, gia citato nelle vesti ben piu adatte di vecchio, che altro devo aggiungere in

piu a quello che dico sempre??? A te dico solo grazie, tu sei un buon intenditor!

Grazie a Il Porco, senza di te mi sarei sentito a disagio a scroccare ai banchetti. Di fronte

a te nessuno puo competere.

Grazie Jampier, uno degli ultimi ad essere aggiunto al pacchetto ma ne sei subito diventato

uno dei maggiori protagonisti. Torna presto che ti devo aiutare con le tue mogli.

Grazie a quella VECCHIONA della Lety, se non me la pio con te con chi me la pio???

Grazie all’altro Compagno con la C maiuscola Luchino, prima di ringraziarti ti devo

chiedere scusa per la sola clamorosa che ti ho passato per le mostra del PCI...Hasta Siem-

pre!

Grazie a Lucia, altrimenti detta Bombe, chissa perche...grazie per questi lunghi anni pas-

sati insieme, grazie per D’Artagnan, grazie per esserti emozionata per qualsiasi cosa..ma

soprattutto ’mo preparete per la prossima avventura che saranno cazzi!

Grazie Lucone, uno degli irriducibili. Ti hanno dato del ripieno di saggezza popolare...frase

certamente appropriata, ma mai quanto la tua! Quando ’l c**o arbatte all’anca, se ’nen

quattro diti de f****a poco ce manca. Rimarra alla storia. Preparete ’mo che venimo su!

Grazie Marina, grazie perche dei due sensi non ne cogli nemmeno uno.

Grazie Marta, nome in codice Martino. Grazie per la tua maestria nell’usare certi utensili.

Sappiamo che ne fai buon uso...Danilo ringrazia!

Grazie al Matthew, grande assente della specialistica. Tanti esami e tante puttanate dette

negli anni. Senza di te il grado di sobrieta e aumentato di parecchio.

Grazie a quell’ (ir)responsabile di Mattia che arriva sempre tardi anche quando se paga.

Grazie a Maura, the strangest! Con te qui in dipartimento tutti noi ci siamo sempre potuti

vestire al buio senza pensieri.

Grazie Meri per la tua simpatica ingenuita.

Grazie Nino, l’amico terrone che gira ’l mondo. Con te e Alessio le peggio risate. Penso

che tutti e tre dobbiamo ringraziare nuovamente il Bomba. Pero me mancano il casatielll

e a pizza i spachett.

Grazie Paolino. Amico sincero sempre e comunque. L’altro moschettiere, quello religioso.

Senza le tue mani sudate e le tua miriade di puttanate questi anni non sarebbero stati gli

stessi. Mi sei mancato questo ultimo mese.

Grazie Pietro per la compagnia, per il database invidiabile, le mogli e per i capelli brizzo-

lati che ci fanno sentire giovani.

Grazie alla Sara, amica da sempre e per sempre. Qualunque momento della mia vita io

guardi, ci sei tu. Ogni tanto se te scanzassi ’n sarebbe male pero :-P!!

Grazie a Sara Leccarelli, ma solo dalle dieci di sera in poi.

Grazie Simone B. per le risate di quest’ultimo periodo. Preparete anche tu per Monaco

che ce ne faremo delle altre insieme a quel ripieno de lucone.

Grazie Simone C., il cugino collega...vede’mpo’ !!

Grazie Simone M., l’unico ciellino quasi tollerabile in un consiglio studentesco. Pero non

te monta la testa mo, che anche se non ce saro piu io a fisica comanderanno Alessia e

Valeria!

Grazie Ser per gli esami che abbiamo fatto insieme, praticamente tutti e per tutte le botte

che ce semo dati, o meglio per quelle che ho piato. Me dispiace solo per il pentatlon...vinci

solo nel calcolo delle sezioni d’urto.

Grazie Toro, l’unico che quanto a stranezze riesce a battere pure Maura.

Grazie Torazzo 100%, ogni colpo lo mette a segno! Grande amico e grande collega...non ti

dimenticare di quel progettino ambizioso. Io non me dimentico!

Grazie infine a tutti gli altri del Pacchetto Fisici On The World e ai miei cugini siculi

Ivana, Giuseppe e Fabio.

Grazie alla nonna che oggi vorrebbe essere qui ma non e potuta venire.

Adesso voglio ringraziare i piccoli Michele e Michelangelo per aver finalmente svecchi-

ato un po’ la combriccola. Era ora, benvenuti. Devo pero chiedervi scusa per la scarsa

presenza del sottoscritto zio, non ho scuse, nemmeno ste 100 pagine de lavoro bastano

come scusa. Spero di poter rimediare in qualche modo. L’augurio che voglio farvi non e

certo l’ormai inflazionato e troppo di moda Stay Hungry Stay Foolish, a voi auguro invece

di ricordare sempre, e come ne avro occasione lo faro io, che Chi lotta puo perdere, ma chi

non lotta ha gia perso (E. Che Guevara).

And now we arrive to the crucial point....immancabili i ringraziamenti per la persona,

non se ne abbiano a male gli altri, piu importante di tutti non solo in questo ultimo anno

ma in generale. Un grazie speciale va a Te Annalisa che da piu di otto mesi a sta parte

mi sei stata piu vicina di tutto e tutti. Anche oggi che non ci sei ti sento vicina. A Te che

mi hai sopportato in questo periodo particolarmente stressante, a Te che hai sopportato

le mie lametele e le mie incazzature rendendo questo periodo unico e irripetibile come

tutto cio che facciamo insieme. A Te che sei piombata inaspettatamente, non solo per me,

nella mia vita. A Te che rendi tutto speciale, vivo e affascinante. Grazie per l’amore e la

compagnia che giorno dopo giorno mi dimostri. Un grosso abbraccio, TH!

That’s all folks!

Matteo

Documents

The AMS-02 experiment: rst data and performances.ams.pg.infn.it/sites/default/files/TESI/Tesi_Palermo_AMS.pdf · AMS-02 has been designed to perform high precision measurements of