Studies of the response stability for long term photomultiplier operation in the ATLAS hadronic calorimeter and a new method for photomultiplier gain measurements

(1)

Università di Pisa

Dipartimento di Fisica

Corso di Laurea Magistrale in Fisica

Curriculum Fisica delle Interazioni Fondamentali

Studies of the response stability for long

term photomultiplier operation in the

ATLAS hadronic calorimeter and a new

method for photomultiplier gain

measurements

Candidate:

Advisor:

Giulia Di Gregorio

Dr. Fabrizio Scuri

(2)

(3)

To my parents and my boyfriend, for their support, encouragement and endless patience

(4)

(5)

Introduction

The Large Hadron Collider (LHC) is a proton accelerator operating at CERN laboratory. The proton-proton collider provides particle bunch collisions each 25 ns at a nominal centre of mass energy of 13 TeV and at a peak luminosity of 1034 cm−2_s−1_{. These conditions allow to investigate the Standard Model predictions} and to measure properties and couplings of the Higgs boson.

ATLAS is one of the general purpose experiments situated on the LHC ring. This thesis deals with its central hadronic calorimeter, TileCal, which is a sam-pling calorimeter made up of steel plates, as absorber material, interleaved by plastic scintillators as active material. The light produced by the passage of particles is transmitted to the readout photomultipliers (PMTs) by wavelength shifting fibres. The PMT signals are shaped, amplified and converted by ADCs. The digitized signals are processed in the ReadOut Drivers, the main component of TileCal back-end electronics, and propagated to the ATLAS ReadOut Systems. TileCal is designed for energy reconstruction of hadrons, jets, hadronic decays of τ leptons and missing transverse energy. To calibrate and to monitor the stability and performance of each part of the readout chain during the data taking, a set of calibration procedures is applied. The TileCal calibration system comprises:

• a hydraulic system with Cesium radioactive sources floating inside the de-tector and used to equalize the global response of the full readout chain; • a Laser system distributing pulses to all PMTs and used to monitor their

stability;

• a charge injection system used to calibrate the response of the readout electronics.

This thesis is focused on the Laser calibration system. The laser light pulses simulate the pulses produced by ionizing particles that pass in the calorimeter. The light is transmitted to all TileCal PMTs through a light distribution system made of beam expanders and optical fibres.

The aim of the Laser calibration is to monitor the drift of the PMTs response in the time between two subsequent global calibrations done with the Cesium sys-tem. The PMT response drifts can include several effects like cathode quantum efficiency loss, photomultiplier windows transparency degradation and photomul-tiplier gain drift. The Laser system uses a complex optical transmission system to send the laser light to the PMTs. Algorithms are used to estimate the stability

(8)

of the optical system which may affect the measurements of the PMT response to laser pulses. These algorithms are based on the assumption that the corrections for optical transmission instabilities can be computed using a subset of PMTs whose response is assumed to be stable in time. With the increasing peak lumi-nosity delivered by LHC, this assumption is no longer valid because all the PMT responses are now drifting due to the large amount of collected light.

In addition, in 2016 a water leakage was discovered in the hydraulic system. All the Cesium scans were suspended. After repair of the Cesium system, only one Cesium calibration was made before starting 2017 collisions. This calibration gives the reference for all the calibration constants which were updated after-wards using only laser calibrations and charge injection data calibrations.

For these reasons a new algorithm that estimates correctly the drift of the PMT responses in long periods (several months) was needed.

The aim of this thesis is to develop a new algorithm which is based on a statistical method to measure the PMT gain. The proposed method does not make any assumption on the PMT response stability. The only hypothesis is the presence of, at least, some PMTs in the calorimeter whose response drift is only caused by a gain drift. These PMTs read out the cells placed in the outer layer of the calorimeter. To derive any variation of the optical transmission system, the response of these PMTs is compared to their gain response. The drifts caused by the optical transmission system are then subtracted from the response of all the PMTs. After correction, the PMT response evolution and the gain evolution overlap within the errors. The only exception is for the PMTs that read scintillating counters placed in the gap between the electromagnetic and hadronic calorimeters. This difference could be caused by the cathode quantum efficiency loss and PMT windows transparency degradation.

My personal contribution was to define a new procedure to estimate the drift of the optical transmission system applying the statistical method to measure the PMT gain. Both the new procedure and the statistical method have been discussed in many ATLAS meetings and they have been included in the TileCal official software framework. This new algorithm has been proposed to become part of the official ATLAS Laser calibration procedure.

Due to the increasing peak luminosity, it is possible that the PMT response is no more stable even during a single run. This aspect has never been investigated before. In the final part of the thesis the PMT stability has been studied in physics runs where laser pulses are sent to the PMTs in empty sections of the LHC bunch train. In all 2015-2016 runs, it is found that the PMT response in physics runs is almost flat if the run has low peak luminosity (∼ 1031 _cm−2_s−1_{). Differently,} in case of runs with high peak luminosity (≥ 1033 _cm−2_s−1_{) the PMT response} evolution in time is characterized by a down-drift of the order of 10−6_/minute in the case of PMTs receiving larger amount of light. This means that, if a run lasts, for example, 40 hours, the photomultiplier response loss is about 0.3%. To verify the correctness of the algorithm, it is also applied to long cosmic runs and to short laser runs where low PMT activity is expected. In these special cases, the photomultiplier response is almost flat.

(9)

Introduction Using the proposed statistical method, it is also possible to study the PMT gain evolution in time during a physics run. These studies will help to understand the PMT behaviour in high-luminosity LHC environment, where the PMT response may vary inside a physics run.

Since the PMT stability in a single physics runs has never been analysed before, I have personally defined the analysis strategy and written all the pro-cedures. This study has been presented in many meetings of TileCal group to discuss its progress.

The structure of the thesis is as follows: Chapter 1 presents an overview of the theoretical picture of the Standard Model and a brief description of the ATLAS sub-detectors. Chapter 2 is dedicated to the general operation of the electromag-netic and hadronic calorimeters referring to the physics processes that happen when a particle passes through the calorimeter. Chapter 3 is focused on the de-scription of the TileCal characteristics and on the features of the TileCal PMTs. Chapter 4 is dedicated to the TileCal calibration procedure and in particular to the Laser system and in Chapter 5 the current laser calibration method is described. In Chapter 6 a statistical approach for PMT gain measurement is pre-sented and the studies done for validating the method are described. In Chapter 7 the new proposed Laser calibration procedure is described. Finally, in Chapter 8 an original study of the PMT response stability during a single physics run to understand the PMT short term stability as a function of the peak luminosity is presented.

(10)

(11)

Chapter 1 Overview

One of the principal aims of the construction of the Large Hadron Collider (LHC) was the observation of the Higgs boson. The discovery of the Higgs boson has been the most important breakthrough in experimental particle physics in the last decade and it has opened a whole new sector of studies at the LHC.

After a short summary of the Higgs boson properties, this chapter focuses on a general description of ATLAS experiment, a multi-purpose detector with full angular coverage around the interaction point of the accelerated particles.

1.1 Theoretical picture and the Higgs boson

The Standard Model of elementary particle (SM) [1, 2, 3] is a theory that describes three of the four fundamental interactions in Nature known as strong interaction, electromagnetic interaction and weak interaction. Only the gravita-tion force is not well described by the SM: that is why it cannot be considered a complete theory.

In the SM elementary particles are divided in two categories: fermions and bosons depending on their spin. The fermions have half integer spin and they obey to Fermi-Dirac statistic. They are the constituents of the matter and are classified in leptons and quarks. On the other hand the bosons have integer spin and they obey the Bose-Einstein statistic. The mediators of the forces are bosons. Fig. 1.1 shows a diagram of all the SM particles classified in quarks (violet), leptons (green), gauge bosons (red) and Higgs boson (yellow).

One special particle is the Higgs boson, a scalar particle associated to the Higgs field, which is responsible to give mass to all the particles through to the spontaneous symmetry breaking mechanism [4, 5, 6]. The Higgs mechanism was formulated by Higgs, Brout and Englert in 1964 but there was no direct evidence of this particle until 2012. The mechanism introduces a complex scalar field with four degrees of freedom: three of them are fixed when the Z boson and W±_bosons acquire mass; the fourth degree is the mass of the Higgs boson which is a free parameter of the SM and it must be determined experimentally.

The SM predicts different ways of producing the Higgs boson in p-p collisions, but the probability of generating the Higgs is small. This probability depends on the

(12)

Figure 1.1: Elementary particles of the Standard Model: quarks (violet), leptons (green), gauge bosons (red) and Higgs boson (yellow).

cross section of the process: the total cross section for proton-proton collisions at center of mass energy√s = 13 TeV is 108 _{nb [7] while the cross section for Higgs} production is 50.6 pb for a Higgs mass equal to 125 GeV [8]. This means that the Higgs cross section is 10 orders of magnitude smaller than the total cross section as shown in Fig 1.2.

The main processes for Higgs boson production at LHC are:

a) gluon fusion: at high centre of mass energy, the gluon-gluon fusion gg → H + X is the Higgs boson production with largest cross section. This pro-cess is mediated by the exchange of a virtual heavy top loop. Contributions from lighter quarks propagating in the loop are suppressed proportional to m2

q.

b) vector boson fusion: the vector boson fusion is the second process with the largest cross section. Higgs production via vector boson fusion qq → qq H proceeds by the scattering of two quarks, mediated by exchange of a W or Z boson. The Higgs boson is radiated off the weak-boson propagator. The scattered quarks give rise to two hard jets in the forward and backward region.

c) associated production: the associated production is the next most relevant Higgs boson production mechanisms. In this channel the Higgs boson is produced in association with a W or a Z gauge bosons.

d) top fusion: the top fusion production channel is important because it can provide information on the top-Higgs Yukawa coupling. In the top fusion the Higgs is radiated off by top quarks.

(13)

1.1. THEORETICAL PICTURE AND THE HIGGS BOSON

Figure 1.2: Cross sections for different processes measured at the Tevatron and LHC colliders as a function of center of mass energy.

Fig. 1.3 shows the Feynman diagrams for the Higgs production channels and Fig. 1.4 shows the Higgs production cross sections as a function of Higgs mass at center of mass energy√s = 14 TeV.

Figure 1.3: Feynmann diagrams for Higgs production: a) gluon fusion, b) vector boson fusion, c) associated production, d) top fusion [8].

[GeV] H M 100 200 300 400 500 1000 H+X) [pb] → (pp σ -1 10 1 10 2 10 = 14 TeV s LHC HIGGS XS WG 2010

H (NNLO+NNLL QCD + NLO EW) →

pp

qqH (NNLO QCD + NLO EW) →

pp

WH (NNLO QCD + NLO EW) →

pp ZH (NNLO QCD +NLO EW) →

pp

ttH (NLO QCD) → pp

Figure 1.4: Cross section for Higgs production as a function of Higgs mass at√s= 14 TeV [9].

As for all massive particles, the Higgs boson can decay in massive particles and in massless particles, like photons. The theoretical total decay width of the Higgs boson is 4 MeV for mH =125 GeV. Fig. 1.5 shows the predicted branching ratios of the different decay modes of the Higgs boson as a function of its mass.

(14)

Figure 1.5: Predicted branching ratios of Higgs boson as a function of its mass [10].

One way that the Higgs boson can decay is into a fermion-antifermion pair. Since the decay width is proportional to the fermion mass squared (Γ ∝ m2

f), the Higgs is more likely to decay into heavy fermions than light ones.

On the other hand the Higgs can also decay into a pair of massive gauge bosons: the most probable decay is into a pair of W bosons.

The third possibility for the Higgs is to decay into massless gauge bosons, as gluons or photons, through an intermediate loop of virtual heavy quarks (top or bottom) or massive gauge bosons. Although the decay into a photon pair is more rare, this channel was very important for the Higgs discovery since energy and momentum of the photons are measured with high precision and it is possible to separate the background from the signal with a high efficiency.

The LHC collider and the experiments ATLAS and CMS have been designed to give a definitive answer to the Higgs boson hypothesis and to measure its mass. In 2012 the ATLAS and CMS Collaborations announced the discovery of a new resonance with a mass of approximately 125 GeV [11, 12]. For a low-mass Higgs boson (110 GeV < mH < 150 GeV), where the natural width is only few MeV, the five decay channels that play an important role for the Higgs boson observation at LHC are listed in Table 1.1 with the corresponding mass resolution.

The Higgs boson decay channels are characterized by different features so different techniques and detector information are used for each channel in order to obtain the best signal to noise ratio. The fundamental requirements for good detector performances are:

1. good calorimetric resolution, both in the electromagnetic and in the hadron sections1_;

1_{The resolution in the electromagnetic calorimeter is important, for example, for the channel}

H → γγ; on the other hand the resolution in the hadron calorimeter is important to reconstruct the jets.

(15)

1.2. THE LARGE HADRON COLLIDER

Decay channel Mass resolution

H → γγ 1 − 2% H → ZZ → l+_l−_l0+_l0− _{1 − 2%} H → W+W− → l+_ν ll0−νl0 20% H → b¯b 10% H → τ+_τ− _15%

Table 1.1: The five principal decay channels for low mass Higgs boson searches at the LHC. The values of the mass resolution are referred to mH =125 GeV [8].

2. precise lepton identification and momentum measurements;

3. complete calorimetric coverage for missing transverse energy ET ,miss detec-tion from which the presence of neutrinos is indirectly inferred2_;

4. good secondary vertex reconstruction for b-tagging and τ-tagging; 5. fast and radiation-hard electronics and sensor elements;

6. high detector granularity to handle high particle fluxes and to reduce the influence of overlapping events;

7. large acceptance in pseudorapidity with almost full azimuthal angle cover-age;

8. good charged-particle momentum resolution efficiency;

The LHC and the detector characteristics will be illustrated in the next Section.

1.2 The Large Hadron Collider

The Large Hadron Collider (LHC) [13] is the world’s largest and most powerful particle accelerator. It is located close to Geneva in the CERN laboratory. It is a circular accelerator that allows to accelerate protons and heavy ions and to make collisions with an energy in the center of mass √s = 13 TeV and a peak luminosity of 1034 cm−2_s−1_.

The accelerator has a circumference of 27 km and it hosts more than one thousand superconducting magnets, cooled down to 1.9 K by liquid helium. These magnets provide a magnetic field of 8 T. A selection of the LHC design parameters is listed in Table 1.2.

The process of proton acceleration is obtained with a chain of machines that accelerate particles to higher energy step-by-step. Primary protons are extracted from hydrogen gas using an electric field and then they are injected into the Linac 2, a linear accelerator. Here protons are accelerated up to 50 MeV. Then

2_{The missing transverse energy E}

T ,missis defined as the momentum imbalance in the plane

(16)

Parameter Nominal value

Beam energy[TeV] 6.5

Center of mass energy[TeV] 13

Number of magnet dipoles 1232

Dipolar magnetic field[T] 8.386

Magnet temperature[K] 1.9

Peak luminosity[cm−2_s−1_] ₁₀34

Proton per bunch 1.05 · 1011

Bunch spacing[ns] 24.95

Table 1.2: LHC nominal parameters.

the beam is injected into the Booster, a circular accelerator which accelerates the particles to 1.4 GeV. Once protons reach this energy, they are injected into the Proton Synchroton (PS), which pushes them to 25 GeV; then they are sent into the Super Proton Synchrotron (SPS), where they are accelerated up to 450 GeV and finally they are injected into LHC. The LHC is the last element of the chain. The two beams are accelerated up to the energy of 6.5 TeV in opposite directions. Figure 1.6 shows the accelerator chain and the four positions where the two beams collide in the LHC. The collision points correspond to the positions of four detectors: ATLAS, CMS, LHCb and ALICE.

Figure 1.6: Schematic accelerator system and the four collision points. The two biggest experiments of LHC are ATLAS and CMS [14]. They are characterized by huge detectors and sophisticated technology in order to do gen-eral purpose measurements of the Standard Model parameters. The other two experiments, LHCb [15] and ALICE [16], are smaller: the first has the goal to measure the parameter of CP violation and rare phenomena connected with hadrons containing bottom quarks, the second, using collisions between heavy ions, tries to verify the existence of quark-gluon plasma which is a key issue in

(17)

1.2. THE LARGE HADRON COLLIDER Quantum Chromodynamics for understanding color confinement and chiral sym-metry restoration.

The maximum beam energy that protons can reach in the LHC is limited by the magnetic field in the storage ring. The nominal field is 8.33 T, that corresponds to an energy of 7 TeV per beam. The actual field achievable is limited by the heat load and by the temperature inside the magnets. The current maximum beam energy is 6.5 TeV.

The number of events, for a given process, generated in LHC collisions is given by the formula:

Nevent = σeventL (1.1)

where σeventis the cross section for the process under study and L is the integrated luminosity. The former factor depends on the physics process involved and it is proportional to the probability of a single final state to appear. More formally, the cross section for a process is defined as the number of interactions per unit time per target particle divided by the incident flux [17]. It is measured in barn or its submultipliers. This quantity is a function of energy: when energy increases, new phenomena contributing to the cross section may appear and become dominant and, at the same time, others can be suppressed.

The integrated luminosity is a parameter of the machine and it is defined as the integral of the instantaneous luminosity:

L = Z

L(t) dt (1.2)

Assuming that the beams have a Gaussian profile and collide head-on, the in-stantaneous luminosity is defined as:

L = f n1n2 4πσxσy

(1.3) where n1 and n2 is the number of particles in each bunch, f is the frequency of collision, σx and σy characterize the transverse beam sizes in the horizontal and vertical direction. The instantaneous luminosity is measured in pb−1 _s−1 _{or cm}−2 s−1 _units.

Equation 1.1 shows a linear dependency between the number of events and the integrated luminosity. Higher luminosity means larger amount of primary inter-actions: this is very important because it allows the study of rare events.

Figure 1.7 shows the integrated luminosity per day delivered by LHC (green) and recorded by ATLAS (yellow) at √s= 13 TeV from April 2016 to October 2016. Figure 1.8 shows the total integrated luminosity from April 2016 to October 2016: ATLAS has recorded data corresponding to 35.6 fb−1 _{in this period (yellow).}

The choice of protons as colliding objects inside the LHC has been driven both by physics motivations and by technical reasons. From the point of view of engineering, it would be impossible to build a synchrotron with such high energy and luminosities using electrons as interacting particles because of the cost and of the required technology. Another advantage of building a p-p collider is con-nected with the energy loss δE during a lap in a circular orbit: δE ∝ E4

(18)

Figure 1.7: Integrated luminosity by day recorded by ATLAS (yel-low) and delivered by LHC (green) in 2016 [18].

Figure 1.8: Total integrated lumi-nosity recorded by ATLAS (yellow) and delivered by LHC (green) in 2016 [18].

given energy, more massive particles lose less energy for emission of synchrotron light.

At the same time, a hadron collider has also disadvantages. A hadron is a com-posed particle so there is a high multiplicity of final particles from each primary interaction, this means that each event is characterized by large background. Another disadvantage is connected with the energy transferred at each collision: indeed it is always smaller than √s because each parton carries only a fraction of the total hadron momentum.

1.3 ATLAS experiment

ATLAS (A Toroidal LHC ApparatuS) [19] is one of the four experiments taking data at LHC and it is an example of a general purpose detector. This experiment is housed in a hall about 100 meters underground, in a beam inter-action point of the LHC ring. The dimensions of the detector are 25 m in height and 44 m in length.

The coordinate system and nomenclature used to describe the ATLAS de-tector and the particles emerging from the p-p collisions are briefly summarized here. The interaction point is defined as the origin of the coordinate system, the beam direction coincides with the z-axis and the x-y plane is the transverse plane to the beam direction. A cylindrical coordinate system is used to describe the detector: the azimuthal angle φ is measured around the beam axis and the polar angle θ is the angle from the beam axis. Usually, the θ coordinate is replaced by the pseudorapidity variable defined as η = − ln tan θ

2

. The ATLAS detector has cylindrical symmetry.

(19)

1.3. ATLAS EXPERIMENT

1.3.1 Detector description

The ATLAS detector is divided in three longitudinal regions: the central region called Long Barrel and the two lateral regions called Extended Barrel. The ATLAS detector consists of a series of sub-detectors: near the beam line there is the most internal sub-system, the inner tracker which is surrounded by a solenoidal magnetic field. Then there is the electromagnetic calorimeter surrounded by the hadronic calorimeter and finally there are the muon chambers. There is also a toroidal magnetic field, which gives the name to ATLAS, situated outside the calorimeters and within the muon system. Fig. 1.9 shows the ATLAS detector layout. A complete description of the ATLAS detector can be found in [19].

Figure 1.9: Cut-away view of the ATLAS detector [19].

A summary of the performance goals of the ATLAS sub-detectors is listed in Table 1.3 together with the η range covered.

Sub-detector Resolution η coverage

Tracking system σpT/pT = 0.05% pT ⊕ 1% ±2.5

EM calorimeter σE/E = 10% /

√

E ⊕ 0.7% ±3.2

Central Hadronic calorimeter σE/E = 50%/

√

E ⊕ 3% ±3.2

Forward Hadronic calorimeter σE/E = 100%/

√

E ⊕ 10% 3.1 < |η| < 4.9

Muon spectrometer σpT/pT = 10% at pT = 1TeV% ±2.7

Table 1.3: General performance goals of the ATLAS sub-detectors with their coverage [19].

(20)

Magnet system

The ATLAS magnet system [20] consists of a solenoid, a barrel toroid and two end-cap toroids, as shown in Figure 1.10.

Figure 1.10: Geometry of ATLAS magnet system: the solenoid (blue), the barrel toroid (pink) and the two end-cap toroids (green).

The solenoid is aligned on the beam axis and it is positioned between the inner tracker and the electromagnetic calorimeter. The inner and the outer diameters of the solenoid are 2.46 m and 2.56 m and its axial length is 5.8 m. It provides a 2 T axial magnetic field.

Differently the barrel toroid is formed by eight coils situated outside the calorime-ters and within the muon system. The magnetic field goes through 25.3 m in length, with inner and outer diameter of 9.4 m and 20.1 m, respectively. Also the two end caps toroids are formed by eight coils and they are situated at the end of the barrel toroid; they have inner and outer diameter of 1.65 m and 10.7 m and a length of 5 m. The barrel toroid provides a magnetic bending over the range |η| < 1.4, instead the magnetic bending of the end-cap toroids is over the range 1.6 < |η| < 2.7. The region over 1.4 < |η| < 1.6 is usually called transi-tion region. Here the magnetic deflection is given by a combination of barrel and end-cap fields.

The three superconducting toroids generate a magnetic field for the muon detec-tors. This configuration creates a magnetic field mostly orthogonal to the muon trajectory in order to minimize the degradation of resolution. The toroidal mag-netic field is not uniform: the barrel toroid provides 2-6 Tm while the end cap toroids contributes with 4-8 Tm. The transition region is characterized by lower bending power.

Inner detector

The ATLAS Inner Detector (ID) [21] is designed to identify charged particles, to measure their momentum and to provide both primary and secondary vertex

(21)

1.3. ATLAS EXPERIMENT measurements for tracks with the transverse momentum pT above 0.5 GeV. Since ID is the nearest detector from the collision point, it is designed to resolve a large track density. To achieve the required momentum and vertex resolutions, the ID is made by high granularity sub-detectors. The sub-detectors of the ID are: the Pixel detector, the Semi-Conductor Tracker (SCT) and the Transition Radiation Tracker (TRT). The layout of the Inner Detector is illustrated in Fig. 1.11.

Figure 1.11: Cut-away view of ATLAS inner detector [19].

The ID is immersed in a 2T magnetic field generated by the central solenoid which allows to curve the trajectory of charged particles. The region covered by the ID is |η| <2.5.

The ID detector is divided in three separated regions: the barrel section and the two end-cap sections. In the barrel region the three sub-detectors are arranged on concentric cylinders around the beam axis, while in the end-cap regions they are located on disks perpendicular to the beam axis.

At inner radii, the high density track information is given by silicon pixel and microstrip layers which offer high granularity and pattern recognition capabilities. At larger radii, the Transition Radiation Tracker provides continuous tracking and improves the momentum resolution.

Calorimetric system

The ATLAS calorimetric system has the task to measure energy of particles that interact with the material of which it is made. It covers a huge η region (|η| < 4.9) using different techniques so with different resolutions.

The calorimetric system is divided in two sections: the electromagnetic calorime-ter and the hadronic calorimecalorime-ter. The first one provides a measurement of elec-tromagnetic showers, while the second one provides a measurement of hadronic showers. Fig. 1.12 shows the layout of the ATLAS calorimetric system.

(22)

Figure 1.12: Cut-away view of ATLAS calorimetric system [19].

The ATLAS calorimeters consist of a number of sampling detectors with full φ-symmetry and coverage around the beam axis in order to have a good Emiss

T measurement.

The electromagnetic calorimeter (EM) [22] is a sampling calorimeter made of steel cladded lead absorbers and liquid argon, arranged in an accordion structure and plunged in the liquid argon. It is divided in three regions: the barrel part covers the range |η| < 1.47 and the two end-cap parts cover the range 1.375 < |η| < 3.2. The total thickness of the electromagnetic calorimeter is more than 22 radiation lengths (X0) in the barrel part and more than 24 X0 in the end-caps.

In the region of |η| < 1.8, a pre-sampler calorimeter is used to correct for the energy lost by electrons and photons upstream of the calorimeter. This pre-sampler consists of an active LAr layer 1.1 cm thick in the barrel region and 0.5 cm thick in the end-cap region.

The hadronic calorimeter [23] is also a sampling calorimeter. It is composed by three parts: the Tile Calorimeter (TileCal), the Hadronic End-cap Calorime-ter (HEC) and the Forward CalorimeCalorime-ter (FCal). The TileCal is placed around the EM calorimeter envelopes and it is made by steel plates as absorbers, and plastic scintillating tiles as active materials. It is composed by three barrels: the central barrel covers the region |η| < 1.0 and the two extended barrels cover the range 0.8 < |η| < 1.7.

The HEC consists of two independent wheels per end-cap, located behind the end-cap electromagnetic calorimeter. They cover the region 1.5 < |η| < 3.2 and they use the liquid argon technology with copper plates.

The FCal is another sampling calorimeter made by copper and tungsten as ab-sorbers and liquid argon as active material. It is integrated into the end-cap cryostats and it covers the region 3.1 < |η| < 5.

Even if the calorimetric structure consists of many detectors, it offers high her-meticity and granularity. Since this thesis is focused on a calibration procedure of TileCal, this subdetector will be described in details in Chapter 3.

(23)

1.3. ATLAS EXPERIMENT Muon system

A specific track system is built to detect muons not interacting in the calori-metric system since they do not originate electromagnetic showers3_{. The muon} system [24] is the outer part of the ATLAS detector and it relies on magnetic deflection of muon tracks. The layout of the muon system is shown in Fig. 1.13.

Figure 1.13: Cut-away view of ATLAS muon system [19].

This system measures muon momentum for pseudorapidity range |η| < 2.7 thanks to the magnetic bending. The magnetic configuration creates a field which is orthogonal to the muon trajectories while minimizing the degradation of res-olution due to multiple scattering. The muon system is also designed to trigger on muons in the region |η| < 1.4.

In the barrel region (|η| < 1.4) tracks are measured in chambers arranged in three cylindrical layers around the beam axis at radii of approximately 5 m, 7.5 m and 10 m; in the transition and in the end-cap regions (|η| > 1.6) the chambers are installed in planes perpendicular to the beam and located at distances of |z| ≈ 7.4 m, 10.8 m, 14 m and 21.5 m from the interaction point.

The muon system is composed by different detectors: at small η values (|η| < 2.0) there are Monitored Drift Tubes (MDT) which provide a precision measurement of the tracks; differently at large pseudorapidity values (2 < |η| < 2.7) there are Cathode Strip Chambers (CSC) which are multiwire proportional chambers with cathodes segmented into strips. These chambers are characterized by high granularity and they are used in the innermost tracking layer due to their higher rate capability and time resolution.

The muon trigger system consists of Resistive Plate Chambers (RPC) in the barrel region and of Thin Gap Chambers (TGC) in the end-cap regions. The

3_{In the EM calorimeter, particles lose their energy for Bremsstrahlung. The energy loss is}

(24)

trigger chambers for muon spectrometer have different purposes: provide bunch-crossing identification, provide pT thresholds and measure the muon coordinate in the direction orthogonal to the beam axis.

1.3.2 Trigger and data acquisition

The ATLAS Trigger and Data AcQuisition (TDAQ) [25] is designed to select and store all the interesting events. It is composed by three separate levels: level 1 trigger (LVL1), level 2 trigger (LVL2) and Event Filter (EF). Each trigger level refines the decisions made at the previous level and it applies additional selection criteria.

The trigger system is very important since the proton-proton interaction rate at LHC is about 1 GHz, but the rate of selected events must be reduced to ∼ 100 Hz. The overall rejection factor is of the order of 107_{. The trigger system provides} such a reduction factor. This is acceptable since the rate of interesting events is only a small fraction of the total rate. Therefore, the trigger must be able to reject the majority of events while accepting interesting events with a high efficiency in a very short time. Fig. 1.14 shows a scheme of the TDAQ system.

Figure 1.14: Block diagram of the ATLAS Trigger and Data AcQuisition system. Since the bunch spacing is 25 ns, the bunch crossing rate is 40 MHz and this is the input frequency for LVL1 trigger. This trigger makes an initial selection based on reduced-granularity information from a subset of detectors. The first selection is based on measurement of high pT muons, electrons, photons, jets, τ-leptons decaying into hadrons, missing transverse energy and total transverse energy. In case the trigger is activated by electron/photon or hadron/τ-lepton,

(25)

1.3. ATLAS EXPERIMENT track isolation can be required. The decision is made in less than 2.5 µs and the target latency for LVL1 trigger is 2 µs. With the LVL1 the data rate is reduced to approximately 75 kHz.

In each event, the LVL1 trigger also defines one or more Region-of-Interest (RoI), the coordinates η and φ of the trigger sectors over threshold. The RoI data in-clude information on the type of feature identified and the criteria passed. If the event passes this first selection it is processed by the LVL2 trigger. The LVL2 trigger refines the previous choices making use of a more complete informa-tion from the calorimeter and from the inner detector. This level takes decision in about ∼ 40 ms and in this case the latency, variable from event to event, ranges between 1 ms and 10 ms. The output rate of LVL2 trigger is about 3.5 kHz. After LVL2 and before event storage, a final selection is performed in the EF. All algorithms are executed offline. The EF, a farm of processors for parallel data reduction, takes information from all ATLAS detectors to create raw-data to be stored at a frequency of ∼ 100 Hz. The average event processing time of the EF is of the order of 4 s.

Other important elements of the TDAQ are the Read-Out Drivers (ROD) which are detector-specific elements of the front-end systems. They achieve a high level of data compression: once an event is accepted by LVL1 trigger, data flow from the front-end pipeline to RODs for reconstruction.

(26)

(27)

Chapter 2 Basic concepts of calorimetry

A calorimeter is a device used to measure the energy of particles interacting in it. Particles entering the calorimeter initiate a particle shower and a large fraction of primary particle energy is deposited in the device by means of sub-sequent interactions of the shower in the calorimeter. Calorimeters constructive characteristics depend on the nature of the particles to be measured.

2.1 Calorimeters in High Energy Physics

A calorimeter [26, 27, 28] is a device which measures the total or part of the energy of the incident particle. The energy deposition of the particles travelling across a calorimeter is produced through the development of a shower or a cascade of particles with decreasing energy. Showers can be either electromagnetic or hadronic. The interaction processes depend on the energy and on the nature of the incident particles. A small fraction of the deposited energy is used to produce a detectable signal which can be measured. The energy deposited by particles of the shower, which can be detected in the form of charge or light, serves as a measurement of the energy of the incident particle. The quality of the calorimeter is given by its intrinsic energy resolution and response linearity to the energy of the incident particle.

The calorimeters can be divided in two main categories: homogeneous calorime-ters and sampling calorimecalorime-ters. The difference consists in the absence or in the presence of passive materials. The passive materials are blocks of matter in which the particles interact and produce showers. In the active materials the particle interactions produce electrons, ions or photons in a quantity proportional to the deposited energy. Charges and photons are collected by the readout system of the calorimeter. The active materials are usually scintillators, ionizing noble liquids or Cherenkov radiators. Absorber materials are lead, iron, copper and uranium. Homogeneous calorimeter uses only active material so the energy deposit can be measured in the entire volume. In a sampling calorimeter the active material is interleaved with passive absorber material. Typically, the sampling fraction is a few percent of the total1.

(28)

The advantage of the sampling calorimeters is that they allow to choose the den-sity of the passive medium according to the constraints of the experiment. For example, it is possible to select very dense materials to produce showers that evolve quickly in a limited space. On the other hand, a disadvantage of this type of calorimeter is that a fraction of the shower energy is deposited in the absorber material and it is not measured. Due to the large statistical fluctuation of the amount of energy deposited in the passive material, the energy resolution is degraded.

Another classification is made considering the type of particles to be detected. Electrons and photons, which lose their energy through the electromagnetic in-teractions (e.g. Bremsstrahlung, pair production), are measured with electromag-netic calorimeters. Differently, devices dedicated to the measurements of hadrons (pions, protons and neutrons) are called hadronic calorimeters.

Calorimeters play an important role in high energy physics experiments for the following reasons:

• they are sensitive to the energy deposit of all incident particles. This means that they are able to detect charged and neutral particles, and, often, calorimetry is the only method to detect the presence of the latter;

• the energy release is associated to the total absorption of the incident par-ticles. Several inelastic collisions degrade their energy and the shower de-velops producing secondary particles with lower energy;

• calorimeters are the only devices which can directly measure the jet energy; • generally, calorimeter response is fast. This feature allows to operate at

high particles rates and for event selection in fast trigger systems. Other features common to all calorimeters are:

• the energy measurement in the calorimeters is associated to stochastic pro-cesses. The average number ¯n of secondary particles produced in the shower development is proportional to the energy E of the incoming particle:

¯

n = E

W (2.1)

where W is the average energy needed to create a secondary particle. The uncertainty in the energy measurement due to the stochastic fluctuations of ¯n is expressed by a term ∝ _√1

E in the intrinsic energy resolution σ(E)

E ; • the shower development is different depending on the nature of the

pri-mary particle. The shower longitudinal and transverse profiles are used for particle identification in the calorimeters.

(29)

2.2. ELECTROMAGNETIC SHOWER

2.2 Electromagnetic shower

The electromagnetic showers are produced by a particle that only interacts via the electromagnetic force with matter such as a photon, electron or positron. Different electromagnetic processes contribute to the shower formation depending on the particle type and energy. Since electrons and photons interact with mat-ter via a few well-understood QED processes, the main shower features can be expressed with simple empirical functions with material dependent parameters.

The average energy lost by electrons in lead and the photon total cross-section in lead as a function of energy are shown in Fig. 2.1 and in Fig. 2.2. Two main regimes can be identified: for energies greater than 10 MeV, the main source of electron energy loss is Bremsstrahlung, while photon interactions produce mainly electron-positron pairs. For larger energies, above 1 GeV, both these processes be-come roughly energy independent. Differently, at low energies, electrons lose their energy mainly through collisions with atoms of the material originating ioniza-tion. Photons lose energy through photoelectric effect and Compton scattering.

Figure 2.1: Fractional energy loss in lead as a function of electron or positron energy.

Figure 2.2: Photon total cross sec-tions as a function of energy in lead. The contributions to the total cross section are the photoelectric effect σp.e., Rayleigh scattering σRayleigh, Compton scattering σCompton, pair production in nuclear field knuc and pair production in electron field ke.

Electrons of sufficiently high energy that hit on a block of material produce secondary photons by Bremsstrahlung, high energy photons produce secondary electrons and positrons by pairs production. These secondary particles produce other particles by the same mechanism generating a shower of particles with

(30)

progressively degraded energies.

To describe the structure and the dimension of the electromagnetic showers, two quantities are used:

• the radiation length X0 which is defined as the mean distance over which a high-energy electron loses 1

e of its energy by Bremsstrahlung. X0 is usually normalized to the material density and it is approximated by X0[g · cm−2] ' 180 · A

Z2 where A is the mass number and Z is the atomic number of the material. X0 sets also the scale of the interaction length for photons: at high energy the mean distance of the photons is 9

7X0.

• the critical energy Ec which is defined as the energy at which the electron loss rates by ionization and Bremsstrahlung are the same. For solids and liquids the critical energy can be parametrized as Ec = 610 MeV_Z+1.24 where Z is the atomic number of the absorber.

The radiation length X0 is used to describe the longitudinal development of the shower. The mean longitudinal profile can be described as:

dE

dt = E0b

(bt)a−1e−bt

Γ(a) (2.2)

where t = x

X0 is the depth inside the material in radiation lengths and a and b are the parameters related to the nature of the incident particle. Approximately the maximum shower depth is located at:

tmax ' ln E0

Ec

+ t0 (2.3)

where E0 is the incident particle energy and t0 = -0.5 (+0.5) for electrons (pho-tons). Usually the calorimeter thickness contains 95% of the shower energy, for example the ATLAS electromagnetic calorimeter which is designed for particle energy of the order of TeV, has a thickness of 45 cm corresponding to 24 X0.

The size of an electromagnetic shower in the transverse plane of the shower axis is due to the multiple scattering of electrons and positrons away from the shower axis. A measurement of the transverse size, integrated over the full shower depth, is given by the Molière radius RM given by:

RM[g/cm2] ' 21MeV X0 Ec[MeV]

(2.4) On average, about 90% of the shower energy is contained in a cylinder of radius ∼1 RM.

2.3 Hadronic shower

Unlike electromagnetic cascades, in which the evolution of the shower is deter-mined by well-understood processes, hadronic showers are more complicated. The

(31)

2.3. HADRONIC SHOWER shower development in a hadronic cascade is governed by interactions between hadrons and nuclei of the absorbers.

Two main effects should be taken into account to understand the hadronic shower:

• in each hadronic collision about 1

3 of the produced pions are π

0 _{that decay} almost immediately in two photons. This means that no more nuclear in-teractions happen and this part of the shower develops electromagnetically; • a large part of the energy lost by charged hadrons is converted in nuclear

excitations and only a small fraction of this energy is detectable.

Consequences of this two effects are the richness and the complexity of the hadronic cascades; an other consequence is the interplay between the hadronic and the electromagnetic fractions of the shower. In particular, the average hadronic shower fraction Fh can be parametrized as:

Fh = E

E0 k

(2.5) where E0 ∼ 1 GeV is a cutoff for the hadronic processes, E is the initial energy and k ∼ −0.2 is the power which describes the suppression of the hadronic components. Looking at the expression of Fh, the higher the energy, the lower is the average hadronic fraction of the shower.

The calorimeter response to the full shower can be expressed as follows:

R = hEh+ eEe (2.6)

where h and e represent the detection efficiency of the calorimeter for the hadronic and electromagnetic components of the cascade, whose energies is re-spectively Eh and Ee. The ratio E_Ee

h depends on the event and its fluctuations are determined by the variation, event by event, of the number of produced π0.

The ratio between the response of the calorimeter to electrons and to pions of the same energy is referred to as e

π and it is defined according to the following expression: e π −1 = 1 − 1 − h e Fh (2.7) A calorimeter with e

h = 1 is said to be a compensating calorimeter; differently if e

h 6= 1 it is a non-compensating calorimeter. The main implications on the calorimeter response for non-compensating calorimeters are:

• the resolution of the calorimeter is influenced by the fluctuations of the π0 production.

• the calorimeter signal is not proportional to the energy of the incident particle.

• the calorimeter resolution does not improve with the energy according to the _√1

(32)

The hadronic shower shape is different from the electromagnetic case due to the fluctuations in the electromagnetic fraction and to the nature of the strong interactions. The average longitudinal development of the shower is measured in units of nuclear interaction length λ, defined as:

λ [cm] = A

NAρσ

(2.8) where A is the mass number of the absorber material, NA is the Avogadro’s number, ρ is the density of the absorber and σ is the total cross section of the process. λ represents the mean distance travelled by a high energy hadron until its energy has been reduced to a fraction 1

e of the initial energy. A numerical approximation of λ, times the material density, is given by:

λ[g · cm−2] ' 35A13 (2.9)

Hadronic calorimeters are designed to contain the whole shower so their length must be a sufficient number of interaction lengths. In the LHC experiments the hadronic calorimeters have a typical thickness of about ten interaction lengths.

The shower depth also depends on the shower energy and on the type of primary particles. For example protons and pions behave differently in hadronic calorimeters: in particular, λπ± > λ_p. It is important to consider the case in which a particle escapes detection because it exits the calorimeter without any sort of nuclear interaction. This type of event is called punch-through and it is a source of degradation in the signal resolution.

The transverse profile of a hadronic shower is much wider than the radial elec-tromagnetic one. The hadronic shower shows a collimated elecelec-tromagnetic core, due to the π0 _{production, surrounded by a large hadronic halo and it develops} more deeply in the medium.

2.4 Calorimeter segmentation and hermeticity

To measure the shower shape, the calorimeters are segmented. Segmenta-tion is useful in particle identificaSegmenta-tion and improves the linearity response of the hadronic calorimeters.

In collider experiments, the calorimeters are often segmented in projective towers for η−Φ measurements of the energy deposit. The granularity is expressed in terms of ∆η × ∆Φ and it is determined by the calorimeter segmentation. A longitudinal segmentation can be achieved by dividing each tower in segments. This segmentation allows to reconstruct the longitudinal shower shape and it helps in jet calibration.

Each segmentation cell requires a dedicated channel of the readout electronic. This implies an insertion in the detector volume of cables and structures through empty gaps that may degrade the calorimeter hermeticity.

Calorimeter hermeticity is required for ET ,miss measurements. Missing energy is commonly used to infer the presence of non-detectable particles such as the

(33)

2.5. LINEARITY AND ENERGY RESOLUTION neutrino and it is expected to be a signature of many predicted physics events that contain particles that do not interact with the detector. The more hermetic the calorimeter, the more precise is the estimate of the missing transverse energy ET ,miss, the more powerful is the neutrino detection.

2.5 Linearity and energy resolution

The main goal of the calorimeter is the energy measurement of the incoming particles. The type and the energy of the particles passing through a calorimeter are unknown a priori and they depend on the initial energy and on the type of primary interaction. Good accuracy over a large energy range is a calorimeter requirement. The quality of the device is mainly measured by two quantities: the response linearity and resolution.

The linearity defines the scaling with the energy: considering a particle of energy E that generates a signal S, the calorimeter is linear if a particle of the same nature with energy kE generates a signal kS. This important property must be ensure in the energy range in which the calorimeter operates.

The second important parameter is the resolution. The energy resolution is important to reconstruct the energy of jets. A jet is a narrow cone of hadrons produced via the fragmentation of quarks and gluons. The jets are the key in-gredient for many physics measurements and for searches for new phenomena. The resolution defines the precision of the energy measurement and it can be parametrized as: σ(E) E = a √ E ⊕ b E ⊕ c (2.10)

where ⊕ represents the sum in quadrature and E is in GeV.

The first term of Eq. 2.10 represents the Poisson statistics-related fluctuations such as the intrinsic shower fluctuations, photoelectrons statistics, dead material and sampling fluctuations. The coefficient a is at a few percent level for a homo-geneous calorimeter and it is typically 10% or more for a sampling calorimeter. The second term of Eq. 2.10 is contributed by the electronic noise of all the read-out channels and pile-up 2 _{effect. For high energy this term can be neglected.} The third term of Eq. 2.10 is related to the energy independent fluctuations given by the detector non-uniformity, the calibration uncertainty, the calorimeter non-compensation and the non-full containment of the shower. Additional con-tributions to the constant term of the calorimeter resolution could be given by the effects of the radiation damage of the active medium.

(34)

2.5.1 Energy resolution of the ATLAS calorimeters

The energy resolution of the ATLAS calorimeters for pions can be parametrized as [29]: σ(E) E = 52% pE[GeV] ⊕ 1.6 GeV E[GeV] ⊕ 3% (2.11)

At low energy the pile-up has an important role so the term 1.6 GeV

E[GeV] is dominant. Differently at high energy ( of the order of TeV ) the constant term is the dominant one.

The contributions to the constant term are:

• calorimeter non-compensation: partial measurement of the energy deposited by hadrons due to the fluctuations, event by event, of the electromagnetic component of the shower;

• leakage: particle energy deposited outside the calorimeter;

• calorimeter non-uniformity: the effective sampling fraction may depend on particle trajectory;

• detector mis-calibration.

The constant term in the resolution of ATLAS hadronic calorimeter is 3%. All contributions to the constant term listed above contribute about 1%. For this reason it is important to have a calibration system which provides sub-per-cent precision.

(35)

Chapter 3 The Tile Calorimeter design

TileCal is the central section of the hadronic calorimeter of the ATLAS ex-periment and it provides important information for reconstruction of hadrons, jets, hadronic decays of τ leptons and missing transverse energy. It is a sampling calorimeter made up of scintillating plastic plates called "tiles" interleaved by steel plates. The light produced by the passage of particles is transmitted to the readout photomultipliers (PMTs) by wavelength shifting fibres. The PMT signals are am-plified with two gains. Different readout channels are added to form calorimeter trigger towers. The digitized signals are processed in the ReadOut Drivers and propagated to the ATLAS ReadOut Systems.

Constructive details of TileCal are described in this chapter, whose last part is devoted to a brief description of the characteristics of PMTs used in the ATLAS hadronic calorimeter.

3.1 Tile Calorimeter layout

The Tile Calorimeter (TileCal) [23] is the central section (|η| < 1.7) of the hadronic calorimeter of the ATLAS experiment. It is a sampling calorimeter made by steel plates and plastic scintillation tiles: the steel plates have the function of absorbers while the scintillating tiles are the active material. It surrounds the LAr electromagnetic calorimeter and it has a cylindrical structure with an inner radius of 2.28 m, an outer radius of 4.25 m and a length of 5.8 m. It is subdivided into three regions: a central section called Long Barrel (LB) and two external sections called Extended Barrel (EB). The LB covers the region of |η| < 1 and the EB covers two regions of 0.8 < |η| < 1.7. The two lateral barrels are referred to as EBA and EBC. Also the Long Barrel region is considered split in two different parts called LBA and LBC even if it is a mechanically a single component, as illustrated in Fig. 3.1. Between the LB and the EB there is a gap of 600 mm, which is needed for routing the Inner Detector and the Liquid Argon cables, electronics and services. Part of the gap contains an extension of the EB called Intermediate Tile Calorimeter (ITC) designed to maximize the volume of active material in this region, while still leaving space for cables.

(36)

scin-Figure 3.1: Section of TileCal barrels.

tillating tiles at particle crossing is collected by wave-length shifting fibres. The fibres deliver the light to the photomultipliers (PMTs) located in the outer radius iron structure that also houses the front-end electronics. The PMT outputs are amplified, shaped and finally digitized by ADCs and then stored in the front-end pipeline memory.

3.1.1 Tilecal basic requirements

The design of TileCal was driven by its capacity to reconstruct hadrons, jets produced in the p-p interactions, and to provide a good ET,miss measurement. The particle and jet energy in large centre of mass energy collisions require good calorimeter performance over a large energy range extending from few GeV up to several TeV.

The guidelines for the detector design were:

• good linearity in the response from few GeV to 1 TeV of deposited energy; • full uniformity in both η and φ coordinates;

• good hermeticity;

• good energy resolution over the whole η range covered; • strong radiation hardness.

These guidelines derive from the requirement of the intrinsic resolution for jets of ∆E

E '

50%_√

E ⊕ 3%.

The main limitation of TileCal is the non-linearity in the energy response to hadron showers. The non-linearity is of the order of 1-2 % and it is due to the non-compensating nature of the calorimeter. The lack of linearity is restored

(37)

3.2. TILECAL MECHANICS AND OPTICS using off-line algorithms which apply different weight to the measured signal. It is very important to restore linearity. For instance, it is estimated that a 5% uncorrected non-linearity may simulate an increase in the jet cross-section at large pT such as in the case of a compositeness scale of 20 TeV at 4 TeV [30].

Another important goal of TileCal is to provide good ET ,miss measurement be-cause it is the only way to detect neutrinos since they do not interact with detec-tors. The measurement of ET ,miss depends on the acceptance of the calorimeter, that’s why the calorimeter must be hermetic.

3.2 TileCal mechanics and optics

TileCal is segmented in depth in three layers, called respectively A, BC (just layer B in the EB) and D. The lengths of the layers are approximately 1.5, 4.1, 1.8 interaction lengths (λ) for the LB and 1.5, 2.6 and 3.3 λ for the EB, respectively. The total detector thickness at the outer edge of the tile-instrumented region is 9.7 λ at η=0 including the EM calorimeter. The projective layout of cells and naming convention are shown in Fig. 3.2.

Figure 3.2: The layout of the TileCal cells and their numbering.

Each barrel is divided azimuthally in 64 modules of trapezoidal shape and each module covers an azimuthal angle φ 0.1 rad wide. Fig. 3.3 shows a typical TileCal module structure: tiles, wavelength-shifting fibres and photomultipliers compose the optical readout system. The tiles are placed in the planes orthogonal to the colliding beams and radially staggered in depth in order to ensure a good sampling homogeneity. The tiles are separated by steel and the ratio in volume of steel and tiles is 4.67:1 in order to contain all the hadronic shower. All the plates have a hole where pipes are inserted for the Cesium hydraulic system.

As shown in Fig. 3.3, the light produced in the scintillating material is col-lected at the edge of each tile using wavelength-shifting fibres. The wavelength shifting fibres convert the scintillation light to a longer wavelength to cope with the PMT cathode spectral sensitivity. The fibres are grouped together and cou-pled to the PMTs which are housed at the outer edge of each module.

(38)

Figure 3.3: Schematic showing how the mechanical assembly and the optical readout of TileCal are integrated together. [23]

Cells are defined by the fibre groups collecting light sent to the photomulti-pliers. The granularity of each cell is ∆η x ∆φ = 0.1 x 0.1 for the first two layers A and BC, while the cells of the D layer have a granularity of ∆η x ∆φ = 0.2 x 0.1.

Fibres grouped in a bundle to form a cell are glued into the fibre-insertion tube, as shown in Fig. 3.4 and 3.5. These tubes are then fixed into the girder positioned at the outer edge of each module. The role of the tubes is to guarantee a precise match to the position of the photomultipliers.

All the cells are read out on both sides by fibres that deliver the light to PMTs. Conventionally, the two PMTs which read the same cell are called or even PMT and odd PMT either left PMT and right PMT. Each cell is readout by two PMTs to provide redundancy.

As already mentioned, the module structure ends on the external side with the girder component. This has a double role in supporting the module and providing the space for photomultiplier tubes and the front-end electronics. To assure fast and easy access to the TileCal read-out system, a movable drawer is inserted inside the girder. The drawer contains both front-end electronics and photomultipliers.

(39)

3.3. TILECAL FRONT-END ELECTRONICS

Figure 3.4: Glued fibre bundle which brings the light to the photomulti-plier in the girder.

Figure 3.5: Fibre routing for TileCal module.

3.3 TileCal front-end electronics

The TileCal front-end electronics is responsible for the amplification, shap-ing and digital conversion of the PMT analog signals. The TileCal on-detector electronics is housed in the drawers (Fig 3.6). Any module of each barrel has its drawer. There are 256 drawers in total. Each drawer houses the PMT block, the amplifier/shaper cards, the Motherboard (MB) that receives the amplified PMT signals and routes them to the Digitizer card. The Mother Board hosts also three interfaces with the back-end electronics.

Figure 3.6: Tile Calorimeter and on-detector electronics drawers.

The PMT block contains four components: a light mixer coupling the fibre insertion tube to the PMT window, the PMT, the PMT voltage divider and the 3-in-1 board which receives the PMT anode signal. The PMT is responsible for

(40)

providing the electronic signals from the light pulses received from the fibres; the 3-in-1 card provides the pulse amplification and shaping.

The MotherBoard has several functionalities. It controls the 3-in-1 card, hosts the digitizers and the interfaces with the data acquisition of trigger system (TDAQ). Fig. 3.7 shows the main components of the front-end electronics and their connections: each MotherBoard controls two Digitizer modules; each Digitizer is composed by two Data Management Unit (DMU) units; each DMU unit receives the data sample from three PMTs blocks.

Figure 3.7: Schematic view of the hierarchical structure of TileCal read-out chain in a module.

3.3.1 PMT blocks

As already mentioned, each TileCal cell is read by two PMTs, for a total of 45 and 38 PMT blocks in Long Barrel and Extended Barrel respectively1_{. For} technical reasons the module structure is composed of 48 PMT blocks and the missing photomultiplier boxes are left empty. Usually the PMT and its relative electronic chain are called channel.

The PMT block is installed in a dedicated hole inside the drawer.

Since this thesis is focused on the stability of PMT response, a brief description of the TileCal PMTs will be reported in Section 3.4.

The function of a PMT block is to convert light signals from the calorimeter cells into electronics signals. Each PMT block contains a PMT, a light mixer which serves as the interface between the PMT and the fibre bundle, a High Voltage divider, and a 3-in-1 board which acts as interface with the read-out electronics. Fig.3.8 shows an example of PMT block. Referring to the Fig. 3.8, the fibre bundle arrives from left and it is coupled with the mixer to the PMT; the electron signal is then collected into the PMT and finally the signal is shaped in the 3-in-1 card. The whole structure is shielded with iron and µ-metal cylinders. The function of the light mixer is to couple the fibres with the photomultiplier uniforming the signal on its photocathode surface. Thanks to the light mixer,

1_{In case of the LB, the total number of PMT blocks is odd because, exceptionally, the D0}

(41)

3.3. TILECAL FRONT-END ELECTRONICS

Figure 3.8: Details of a PMT block.

the output signal of the photomultiplier is independent from the fibre pattern in the bundle.

Fig. 3.9 shows the 3-in-1 card: the PMT signal is shaped and split. The resulting two shaped signals are sent to two independent amplification chains. The outputs have a gain ratio of 64. These two chains are referred to as Low Gain and High Gain branches and they are connected with the Digitizer board for further information. The signal from the Low Gain branch is also sent to the Level-1 trigger system. The importance of the multi-gain structure is that it allows to measure signals from few MeV up to TeV maintaining a high resolution in the whole energy range.

Figure 3.9: The 3-in-1 board diagram.

The second function of the 3-in-1 card concerns the possibility to control and inject at the input stage of a shaper a known signal of programmable amplitude. This system is called Charge Injection System and it is part of the calibration chain of TileCal. It is described in Section 4.2.3.

(42)

The last function of the 3-in-1 card is implemented with the slow integrator. The integrator signal is used to process the information from the signals produced during a calibration with the Cesium system, another component of the calibra-tion system (see Seccalibra-tion 4.2.1). The integrated signal is used also to measure the average anode current produced by Minimum Bias events, a diagnostic tool developed to provide an independent cross-check of the global calibration (see Section 4.2.4).

3.3.2 Motherboard, Digitized board and Interface board

All the signals from the drawers to the back-end electronics are transmitted with a three-layers structure. The first layer below the PMT block is the Moth-erboard. The Motherboard carries the low voltage power and digital control signals for the PMT block. In particular it drives the Charge Injection System, distributes the timing and controls signals from the ATLAS TDAQ system.

The middle layer is the Digitizer board which receives the analog output signals from the 3-in-1 card in the PMT block and it is responsible for digitizing the signal. The shaped and amplified signals generated by the two amplification chains (High Gain and Low Gain) in the 3-in-1 cards are sampled every 25 ns and digitized by 10-bit ADCs in the DMU inside the Digitizer board. For each channel two ADCs are required due to the two gain branches. The default data samples are the ones from the High Gain branch, unless saturation in the HG-ADC has occurred. In case of saturation, the Low Gain samples are passed. The saturation in the High Gain branch occurs for charge above 12.5 pC, while the Low Gain region ends for input above 800 pC. In terms of energy, the maximum energy read is about 13 GeV and 850 GeV in HG and LG region respectively. The data samples are stored temporarily inside the TileDMUs in pipeline while waiting for the Level 1 Accepted Signal. If the event does not satisfy the LVL1 trigger conditions, the sample is rejected, otherwise it is transferred to the buffer memories.

Finally the outer layer is the Interface Card which has the role to transmit the sampled signals to the back-end electronics.

3.4 TileCal Photomultipliers

The photomultiplier is a device which converts light pulses into electron cur-rent pulses through an amplification. The essential elements of the PMT are:

• photocathode which converts light flux into electron flux;

• electron-optical input system which focuses and accelerates the electron flux; • electron multiplier which consists of a series of electrodes, called dynodes,

responsible of electron secondary emission; • anode which collects the amplified electron flux.

Studies of the response stability for long term photomultiplier operation in the ATLAS hadronic calorimeter and a new method for photomultiplier gain measurements

Università di Pisa

Dipartimento di Fisica

Corso di Laurea Magistrale in Fisica

Curriculum Fisica delle Interazioni Fondamentali

Studies of the response stability for long

term photomultiplier operation in the

ATLAS hadronic calorimeter and a new

method for photomultiplier gain

measurements

Candidate:

Advisor:

Giulia Di Gregorio

Dr. Fabrizio Scuri

Contents

Introduction

Chapter 1

Overview

1.1

Theoretical picture and the Higgs boson

1.2

The Large Hadron Collider

1.3

ATLAS experiment

1.3.1

Detector description

1.3.2

Trigger and data acquisition

Chapter 2

Basic concepts of calorimetry

2.1

Calorimeters in High Energy Physics

2.2

Electromagnetic shower

2.3

Hadronic shower

2.4

Calorimeter segmentation and hermeticity

2.5

Linearity and energy resolution

2.5.1

Energy resolution of the ATLAS calorimeters

Chapter 3

The Tile Calorimeter design

3.1

Tile Calorimeter layout

3.1.1

Tilecal basic requirements

3.2

TileCal mechanics and optics

3.3

TileCal front-end electronics

3.3.1

PMT blocks

3.3.2

Motherboard, Digitized board and Interface board

3.4

TileCal Photomultipliers