Performance and Radiation Hardness of the ATLAS/SCT Detector Module

Full text

(1)Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 839. Performance and Radiation Hardness of the ATLAS/SCT Detector Module BY. LARS EKLUND. ACTA UNIVERSITATIS UPSALIENSIS UPPSALA 2003.

(2) Dissertation for the Degree of Doctor of Philosophy in High Energy Physics presented at Uppsala University in 2003. )*564)+6. Eklund, L. 2003. Performance and Radiation Hardness of the ATLAS/SCT Detector Module. Acta Universitatis Upsaliensis. Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 839. 89 pp. Uppsala. ISBN 91-554-5630-8. The ATLAS experiment is a general purpose experiment being constructed at the Large Hadron Collider (LHC) at CERN, Geneva. ATLAS is designed to exploit the full physics potential of LHC, in particular to study topics concerning the Higgs mechanism, Super-symmetry and CP violation. The cross sections for the processes under study are extremely small, requiring very high luminosity colliding beams. The SemiConductor Tracker (SCT) is an essential part of the Inner Detector tracking system of ATLAS. The active elements of the SCT is 4088 detector modules, tiled on four barrel cylinders and eighteen end-cap disks. As a consequence of the high luminosity, the detector modules will operate in a harsh radiation environment. This thesis describes work concerning radiation hardness, beam test performance and methods for production testing of detector modules. The radiation hardness studies have been focused on the electrical performance of the front-end ASIC and the detector module. The results have identied features of the ASIC failing after irradiation and conrmed the good performance of the re-designed ASIC. The beam tests have been performed in the late prototyping and the pre-production phase, verifying the specied performance of the detector modules. Special eort have been made to evaluate the performance of irradiated detector modules. The assembly, quality assurance and characterisation of the detector modules will be done in the collaborating institutes. The thesis reports on methods developed for use during the production, to assess the electrical performance. Lars Eklund, Department of Radiation Sciences, Uppsala University, Box 535, SE-751 21 Uppsala, Sweden. c . Lars Eklund 2003. ISSN 1104-232X ISBN 91-554-5630-8 Printed in Sweden by Akademitryck AB, Edsbruk 2003.

(3) This thesis is based on the following papers, which will be referred to in the text by their Roman numerals: I L. Eklund et al., SEU Rate Estimates for the ATLAS/SCT Frontend ASIC, Submitted to Nucl. Instrum. Meth.. II Y. Ikegami ... L.Eklund ... et al., Results of early phase of series production of ATLAS SCT barrel hybrids and modules, Proceedings of the 8th Workshop on Electronics for LHC experiments, Colmar, France (2002). CERN-LHCC-2002-23 pp 116-120 III T. Kondo ... L. Eklund ... et al., Construction and performance of the ATLAS silicon microstrip barrel modules, Nucl. Instrum. Meth. A "&# (2002) 27. IV U. Unno ... L. Eklund ... et al., Beamtest of Non-irradiated and Irradiated ATLAS SCT Silicon Microstrip Modules at KEK, IEEE Trans. Nucl. Scie. "' (2002) 1868 - 1875 V W. Dabrowski ... L. Eklund ... et al., Progress in development of the readout chip for the ATLAS semiconductor tracker, Proceedings from 6th workshop on Electronics for LHC Experiments, Cracow, Poland (2000). CERN-LHCC-2000-041 pp 115-119..

(4)

(5) Contents Contents 1 Introduction 1.1 1.2 1.3. 1.4. High Energy Physics Motivation . . . . . . . CERN and the LHC Collider . . . . . . . . . The ATLAS Experiment . . . . . . . . . . . . 1.3.1 Overview . . . . . . . . . . . . . . . . 1.3.2 Inner Detector Physics Requirements . 1.3.3 Inner Detector Performance . . . . . . The SemiConductor Tracker . . . . . . . . . . 1.4.1 The Silicon Strip Detector Module . . 1.4.2 Detector Module Design Specications. 2 Technology and Methods 2.1 2.2. Device Physics . . . . . . . . . . . . . . . . 2.1.1 Radiation Damage in Silicon Sensors 2.1.2 Radiation Damage in Electronics . . Test Methods for Binary Read-out . . . . .. 3 Radiation Tolerance 3.1 3.2 3.3 3.4. Irradiation Set-up . . . Irradiation Procedures Irradiation Results . . Single Event Upsets .. . . . .. . . . .. 4 Beam Tests 4.1 4.2. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . . . . . .. . . . . . . . . .. . . . . . . . . .. . . . . . . . . .. . . . . . . . . .. . . . . . . . . .. . . . . . . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. The Beam Test Set-up . . . . . . . . . . . . . . . . . . . Calibration of Detector Modules . . . . . . . . . . . . . 1. 1 3. 3 8 11 11 14 18 22 23 27. 37. 37 38 40 42. 47. 47 49 54 58. 61 62 63.

(6) Contents. 2 4.3. Measurement Program . . . . . . . . . . . . . . . . . . .. 65. 5 Quality Assurance of Detector Modules. 71. 6 Concluding Remarks. 79. 5.1 5.2 5.3. Production Scheme . . . . . . . . . . . . . . . . . . . . . Quality Assurance Procedures . . . . . . . . . . . . . . . Electrical Tests in Production . . . . . . . . . . . . . . .. References. 71 73 75. 85.

(7) CHAPTER 1 Introduction. 1.1 High Energy Physics Motivation. Modern physics is said to start in 1900, with the demonstration by Max Planck that radiation is quantised, a conclusion derived from the blackbody radiation measurements of Rayleigh and Jeans. This eventually led to the formulation of quantum mechanics, which together with Einstein's theory of special relativity from 1905 is the starting point for the theories of particle interactions developed during the twentieth century. Experimental discoveries and measurements were made in parallel, guiding the development of theories. The electron discovered by Thomson in 1897 marks the beginning of the era of particle physics and the ongoing search for the fundamental constituents of matter. With the evidence of the proton found by Rutherford in 1919, a consistent set of fundamental particles had been observed; the electron, the proton and the photon. Quantum mechanics, was combined with the theories of special relativity by Dirac, forming relativistic quantum mechanics. The Dirac equation predicted the existence of a particle with the same mass as the electron but with all other properties mirrored. In 1933 Anderson conrmed this prediction by observing positrons in cosmic rays. Relativistic quantum mechanics was extended into quantum eld theory in which the number of particles does not have to be conserved. This enables for instance the creation of particle anti-particle pairs to as long as all conservation laws are fullled. Yukawa proposed a theory of interaction based on particle exchange, which he applied to the interactions 3.

(8) 4. Chapter 1 Introduction. between nucleons. The proposed particles were called π mesons and their mass was predicted by the size of the nucleus. The π meson was rst observed in cosmic rays 1947. In parallel, the experimental techniques developed rapidly. The early pioneers used photographic lm or scintillators to observe radioactive decays. Rutherford took the innovative step of to use the radiation from a radioactive source as a probe, studying the scattering of α particles from a gold foil to determine the structure of the atom. In the early 1930s, techniques for particle accelerators were developed making it possible to control the initial conditions of the particle interactions. Hence, taking particle physics from an observational science to methods of active experimentation. However, cosmic ray observations have remained an important source of experimental input in particle physics. The µ lepton rst observed in cosmic rays 1937 was the rst particle from what would be known as the second generation of fermions. During the following decades many new particles were observed. To structure the large number of dierent particles observed, mathematical group theory was applied and it proved to be successful in describing the observed particle spectrum. Theorists Gell-Mann and Zweig proposed the quark model where mesons and baryons are described as composites of three quarks: up, down or strange. The model was originally considered to be a only mathematical construction to explain the pattern in the observed particle spectrum. With the exhaustive experimental and theoretical evidence available today the quarks are considered to be physical particles, however without the possibility to appear as free particles. Experimental studies of particle interactions indicated the existence of three dierent kinds of forces with dierent characteristics. Firstly the strong force, as modelled by the meson exchange suggested by Yukawa. Secondly the week force responsible for e.g. β decays and nally the electromagnetic force. The idea of exchange particles as a force carrier, as proposed by Yukawa for the strong force, was developed for the weak force by Schwinger, Bludman and Glashow. Similar to the theory of the strong force, the weak force is mediated by massive bosons. The addition of a fourth quark to the model together with the unication of the theories for the weak and electromagnetic force completes the structure of what today is called the Standard Model (SM) of particle physics. As seen in Table 1.1, the particles of the SM are divided into three.

(9) 1.1 High Energy Physics Motivation. 5. fermions Generation. hadrons. leptons. I. up. down. e. νe. II. charm. strange. µ. νµ. III. bottom. top. τ. ντ. charge. 2 3. − 13. −1. 0. Table 1.1.. Table summarising the observed fundamental fermions in the Standard Model of particle physics Interaction. bosons. relative strength. Strong. gluons. 1. Electromagnetic. photons. 10−2. Weak. Z 0, W ±. 10−9. Gravitation. graviton. 10−38. Table 1.2. Table summarising fundamental interactions and the force mediating bosons in the Standard Model of particle physics generations of fermions. The existence of a fourth quark, the charm quark completing the second generation, was theoretically predicted in 1970 and experimentally observed in 1974 by two independent experiments. The agreement between theoretical predictions and experimental observations oered strong support for the SM. The story of the third generation is quite dierent. The observation of the τ lepton 1976, the rst observed particle from the third generation, was unexpected. More evidence for a third generation of fermions was given in 1976 by the discovery of another quark, the bottom quark. For the existing theories to hold a sixth quark had to be introduced, called the top quark. This was nally discovered in 1995 at Fermilab. The third generation was completed in 2000, by the rst direct observation of the τ neutrino..

(10) 6. Chapter 1 Introduction. Tables 1.1 and 1.2 describe the classication of particles according to their properties. Firstly, a division into fermions and bosons is done on the basis of the intrinsic spin. Fermions have half-integer spin and are therefore subject to the Pauli exclusion principle. Bosons, the force mediators, have integer spin and are not subject to the exclusion principle. Particles are also classied by their sensitivity to dierent interactions. Charged particles are subject to electromagnetic interaction. The dierence between hadrons and leptons is that hadrons have colour charge, hence they are subject to the strong interaction. Finally, all fermions are subject to the weak interaction. The remaining interaction, gravitation, is not included in the SM. It is well described by general relativity on the macroscopic scale, but on the microscopic scale the weakness of the interaction makes its contributions negligible. Furthermore, it is a continuous theory as opposed to the quantised theories for the other three interactions. However, it is a common assumption that all massive particles are subject to gravitation. In 1967, Weinberg and Salam proposed a unied theory for the electromagnetic and weak interaction. The theory of electroweak interaction describe the two forces in a unied theory, even though they appear as two dierent phenomena at low energies. The two interactions becomes indistinguishable at high energies, above the so called electroweak unication scale. The theory predicted the existence of three heavy force mediating bosons as presented in Table 1.2, the Z 0 and the W ± , which were discovered at CERN 1983. The theory of electroweak unication also predicts the existence of a Higgs boson, the only particle in the SM not yet observed. The Higgs eld was introduced by Higgs in 1964 as a mechanism explain the mass asymmetry observed for the force mediating bosons in the electroweak theory. Because of its role as the generator of particle masses in the theory of the SM, the Higgs eld is of decisive theoretical importance. The remaining important experimental conrmation of the theory would be the observation of the quantisation of the Higgs eld, the Higgs boson. During the decades that followed the formulation the SM of particle physics, many experimental results have conrmed the predictions of the SM. Examples of conrmations were given above, such as the discoveries of the Z 0 and W ± bosons and the particles needed to complete the third generation. Precision measurements of parameters in the theory have also conrmed its consistency, e.g. the measurement.

(11) 1.1 High Energy Physics Motivation. 7. of the width of the Z 0 resonance at the Large Electron Positron (LEP) collider at CERN. This width is a measure of the lifetime of the Z 0 boson. The prediction of the Z 0 lifetime is determined by the number of available decay channels. This in turn, is dependent of the number of generation of fermions, in particular the number of light neutrinos. The measured resonance width at LEP in consistent with a theory of three generations of fermions, but not with two or four. Hence, there is strong experimental evidence that there are exactly three generations. There are still open issues concerning the SM implying that our understanding of the fundamental constituents of matter is incomplete. The theory has been extensively tested up to energies of the electroweak unication scale. That is, for interactions with momentum transfers up to about 200 GeV. Attempting to extrapolate the theories up to much energies above gives computational problems. When calculating observables, e.g. the cross section for a reaction, the theory diverges. This implies that the theory must be incomplete and the missing constituents must appear at energies below 1000 GeV. The experimental search for the Higgs boson has been going on since the theory was rst put forward, and has intensied as more experimental evidence supporting SM has been found. Intensive search programs were conducted at the LEP collider during the operation of LEP from 1989 to 2000 without nding clear evidence for the Higgs boson within its energy range. Intensive searches have also been performed at the Tevatron accelerator, so far without success. The centre of mass energy of the Tevetron has been increased to 2000 GeV, and the search continues. Another scientic challenge in high energy physics is understanding the asymmetry between matter and anti-matter. Since the universe appears to consist of almost exclusively matter, there must be a preference for matter over anti-matter in nature. Decays of short-lived particles can be used to study this asymmetry, e.g. by looking with high precision at branching ratios into dierent decay channels. Mesons and baryons containing bottom quarks are a promising probe into the eld of matter anti-matter asymmetry. The avour of the quark involved has given the name to this eld: b-physics. By studying b-decays with great precision, it is possible to determine whether the observed asymmetry is within or exceeding the boundaries dened by the SM. There are many extensions to the SM that are appealing from a.

(12) 8. Chapter 1 Introduction. theoretical point of view. One example is super-symmetry (SUSY). In the theories of modern physics, symmetries play a crucial role. For example, translational symmetry implies that a theory is invariant under spatial translations, a property that every valid theory must have. Furthermore, theories must be symmetrical under rotations, time translations, relativistic Lorentz boosts and under gauge transformations. Super-symmetry is an extension of this concept, suggesting a symmetry between particles of integer and half-integer spins, or in other words symmetry between fermions and bosons. The theory predicts a whole spectrum of super-symmetric partners to the known particles, a fermion for each boson and vice versa. Non of these super-symmetric partners have yet been discovered, but searches have been performed and are ongoing. In the search for physics beyond the SM, it is of crucial importance to push the experimental frontier further. History has shown that the interplay between theoretical models and experimental observations has resulted in a detailed understanding of the fundamental laws of physics. To investigate the open issues presented above, and many more not mentioned here, there is the need for experimental exploration beyond the capacity of present day accelerators and detectors. As previously explained, the energy region up to a 1−2 TeV is likely to contain physics of the Higgs sector and possibly physics beyond the SM. A collider at the TeV energy scale like the LHC has a large potential for the discovery of many new phenomena, expected and unexpected.. 1.2 CERN and the LHC Collider CERN is the European Laboratory for particle physics located outside Geneva, Switzerland. The organisation originates from the Conseil Européen pour la Recherche Nucléaire (CERN) founded in 1951. In 1953 the Council decided to build the laboratory near Geneva. The organisation has 20 member states at present, and many more countries contributing to specic projects. Many successful scientic programs have been performed at CERN, as mentioned in Section 1.1. For instance the sp¯ ps collider with the discovery of the Z 0 and the W ± bosons and the LEP collider with the precision measurements of the Standard Model. The Large Hadron Collider (LHC) [1] is under construction at CERN to reach beyond present day accelerators in terms of collision energy and.

(13) 1.2 CERN and the LHC Collider. 9. luminosity. The collider will be situated approximately 100 metres under ground in the 27 km long tunnel previously used by the LEP collider. One fundamental dierence between LHC and its predecessor is that LHC is a hadron collider (proton-proton), whereas LEP was a lepton collider (electron-positron). The maximal collision energy reached by LEP, approximately 200 GeV, was limited by the synchrotron radiation emitted by the electrons and positrons due to their curved trajectories. By using protons, the losses due to synchrotron radiation is greatly reduced due to the much greater mass of the proton (mp = 1800 × me ). For the LHC the greatest design challenge, in terms of maximal collision energy, will be the maximum magnetic eld achievable in the bending magnets The design value of the eld is 8.6 T. Hadron colliders have a major complication over lepton colliders since hadrons are not fundamental particles. Protons have an internal structure, valence quarks and a quark-gluon sea. Electrons are at least to our present day knowledge point-like objects. A collision between fundamental particles like electrons and positrons is straightforward, the total energy of the two annealing particles is available for the production of collision fragments. The collision between two protons is more complex. The interaction takes place between the fundamental constituents of the protons. The momentum transfer occurs between for example two gluons or two quarks, each carrying only a fraction of the total momentum of the proton. Thus, even if the cross section for a proton-proton interaction is relatively large the transferred momentum in the collisions span over a large range. Only a small fraction of the collisions will have suciently large momentum transfer to open interesting physics channels. In addition, the cross sections for most processes to be studied are extremely small and to measure the parameters in the dierent theoretical models with precision requires large statistics. Hence the need for an accelerator that can deliver very high intensity beam. Some of the design parameters of the LHC can be found in Table 1.3. Two bunches of accelerated particles that are passing through each other is called a bunch crossing or an event. The high luminosity gives a large number of proton-proton interactions each bunch crossing, on average 23 at full design luminosity. Due to the very low fraction of high momentum transfer events and the small cross sections for the physics channels of interest, most events are discarded. This has two.

(14) Chapter 1 Introduction. 10. Collision energy (p-p). 14 TeV (CM). Collision energy (Pb-Pb). 3 TeV/nucleon. Maximal luminosity. 1034 cm−2 s−1. Collision rate. 40 MHz. Circumference. 27 km. Commissioning year. 2007. Table 1.3.. Table summarising some design parameters of the LHC.. important implications on the detector system that is built to record particles created at the collision. Firstly, there is a need for event selection or what is called a JHECCAH. The LHC experiments use several layers of trigger decisions. The rst trigger decision (Level 1) is crude but very fast and the selected events are subject to a rened analysis before the next level of trigger decision. This requires the detector systems to have the ability to temporarily store event data while awaiting trigger decisions. Secondly, high event rate and the large number of interactions per event generates a high uence of particles through the detectors. Thus, the detector systems has to cope with the large uence of particles without being saturated with data. Furthermore, the particle uence in itself corresponds to a considerable radiation dose, that potentially can damage the detector systems. The radiation doses are highest closest to the collision points where the particle densities is highest. This subject is further developed in Section 1.4.2. One experiment will be built at each one of the four collision points of LHC. The experiments are ALICE, ATLAS, CMS and LHCb. ALICE is a heavy-ion collision experiment, to be used when LHC is colliding Pb atoms, attempting to create quark-gluon plasma. This is a state of matter found in neutron stars and in the very early universe, fractions of a second after the Big Bang. LHCb is an experiment specialising in reconstruction and analysis of the decay of b-avoured hadrons. As described in Section 1.1, these decays are in particular interest for the study of matter anti-matter asymmetry, more precisely CP-violation. ATLAS and CMS are general-purpose experiments, designed to recon-.

(15) 1.3 The ATLAS Experiment. 11. struct all processes described in Section 1.1 and many more. The design goal of good performance in all the bench-mark physics channels make them well suited to study the unexpected and the unpredicted.. 1.3 The ATLAS Experiment 1.3.1 Overview The ATLAS (A Toroidal LHC ApparatuS) [2] experiment is a general purpose experiment designed to exploit the full physics potential of LHC. It has a coverage in pseudo-rapidity of |η| < 2.5, aiming to fully reconstruct all events within this acceptance. In high energy physics events, the processes under study take place very close to the interaction point, in general within a few mm. The particles discussed in in Section 1.1 are in general very short-lived, decaying in the vicinity of creation point within a time-frame to short to be resolved by any conceivable detector. Consequently, to study the processes modelled by theory one can only detect the decay fragments and reconstruct the original particles from those. The reconstruction is done by measuring the properties of each decay fragment as accurately as possible. For each particle the aim is to measure its linear momentum, energy and identity. The identity means classication into known groups of particles, e.g. electron, muon or pion. To achieve this, ATLAS consists of several sub-systems, each specialising in measuring certain properties of the decay products of the original particles. Figure 1.1 show an illustration of the ATLAS experiment. The innermost part is called the Inner Detector and has the purpose of measuring the trajectories of charged particles. The Inner Detector consists of three sub-systems, the Pixel Detector (Pixel), the SemiConductor Tracker (SCT) and the Transition Radiation Tracker (TRT). The sensing elements are mounted either on concentric cylinders around the collision point called barrels, or on disks placed as lids on the barrels called end-cap disks. The Pixel Detector consists of three barrel cylinders and eight end-cap disks covered by silicon pixel detectors. The barrels have radii between 4.8 and 16 cm and the distance between the outermost disks is 204 cm along the beam line. The SCT has four barrels with radii from 30 to 52 cm and eighteen disks stretching over a length of 5.5m. The barrels and disks are tiled with silicon micro-.

(16) 12. Chapter 1 Introduction. Figure 1.1. Schematic representation of the ATLAS detector. strip detectors. The SCT will be described in more detail Section 1.4. The third part of the Inner Detector is the TRT, a straw-tube detector giving a large number of space points with slightly lower precision. It stretches from 56 to 107 cm in radius and 6.8 m along the beam axis. In addition to providing space points for tracking the TRT is detecting transition radiation photons, emitted by electrons traversing the walls.

(17) 1.3 The ATLAS Experiment. 13. of the straws. The read-out of the TRT straw tubes has two thresholds. The rst threshold is tuned to trigger on the signals from charged particles traversing the straw. The second and higher threshold is tuned to detect signals from the transition radiation photons emitted by electrons traversing the walls of the straws. This is a powerful tool for electron identication, see Section 1.3.2. The Inner Detector [3] is surrounded by the solenoid magnet [4] (see Figure 1.1), generating a magnetic eld of 2 T parallel to the beam axis. The transverse momentum, pT , is extracted from the curvature of the charged particle trajectories in the Inner Detector. The coil of the magnet is superconducting and has a inner winding radius of 1.22 m for a length of 5.3m. The solenoid magnet is integrated in the inner vacuum vessel of the Electromagnetic (EM) Calorimeter [5]. The purpose of the EM calorimeter is to absorb and measure the energy of electrons, positrons and photons. Both the end-cap and barrel EM Calorimeters are of based on Liquid Argon (LAr) Accordion design. It is a sampling calorimeter with absorber plates of lead folded to a structure resembling the bellow of an accordion. LAr is used as an ionising medium, the charge drifting by the applied high voltage induce signals on segmented read-out electrodes. The Hadronic Calorimeter [6] will measure the energy of all strongly interacting particles not absorbed by the EM Calorimeter. Because of the dierence in particle density and radiation doses in the barrel and end-cap regions, two dierent technologies are employed. The barrel and the two extended barrel hadronic calorimeters have iron absorbers interleaved with plastic scintillating tiles read out with wavelength shifting bres. The end-cap hadronic calorimeter has Copper plates as absorbers alternating with LAr gaps. In the region close to the beam pipe the EM and hadronic calorimetry are combined in the Forward Calorimeters, also based on LAr. To cope with the extreme particle densities and radiation levels, a very thin gap of LAr is needed to avoid ion build-up that distorts the electrical eld. This is achieved by using the concept of a rod in a tube, with LAr lling the gap. The rods are equidistantly spaced, parallel to the beam axis and surrounded by absorbing material. For the EM part of the Forward calorimeter the absorbing material, rods and tubes are made of Copper, for the hadronic part the material is Tungsten. The muon spectrometer [7] surrounds the calorimeters (see Fig-.

(18) 14. Chapter 1 Introduction. ure 1.1) and consists of Muon detectors and toroidal magnets [4]. Muons are the only charged particles penetrating the calorimeter system, a fact used for muon identication. The magnet system for the muon spectrometer consists of eight superconducting air-coils in the barrel region and eight in each end-cap, generating a toroidal magnetic eld. The inner diameter of the barrel toroids is 9.4 m, the outer diameter is 19.5 m and the length is 26 m. There are three measurement stations for muons both in the barrel and in the end-cap region, giving high precision space points for the measurement of the sagitta of the track. The overall length of ATLAS is 44 m and the diameter is 22 m, including the muon spectrometer. ATLAS has three levels of triggers in the event selection. Level 1 trigger [8] makes use of a limited set of information from the muon spectrometer and the calorimeters. The input rate is the full 40 MHz LHC bunch crossing frequency and the latency for the trigger decision is approximately 2µs. Awaiting the Level 1 decision, event data is stored in pipeline buers locally in the sub-detectors. The acceptance rate from the Level 1 trigger is expected to be 100 kHz. For the Level 2 [9] trigger decision, information from all sub-detectors is used, but only from a some regions of the detector acceptance. These regions of interest are dened by the Level 1 decision. The Level 2 trigger latency is expected to be 10 ms and the acceptance rate about 1 kHz. Finally, all data is read-out and the event fully reconstructed as a basis for the Level 3 trigger decision [9]. The accepted events are stored for future analysis with an expected rate of 100 events per second. The time budgeted for reconstruction and Level 3 trigger decision is approximately 1 s.. 1.3.2 Inner Detector Physics Requirements The Inner Detector is the detector system in ATLAS that performs what is called the tracking for analysis of high energy physics events. Tracking is the common name for extracting the topological structure and the reconstruction of the tracks emerge from the collision point. Hence the name tracking. The aim of this section is to illustrate the role of the Inner Detector for the data analysis and to motivate the key parameters to be extracted from physics events [3]. To put things in context, consider a simulated high energy physics event in ATLAS, for instance one used in the evaluation of the performance in searches for the Higgs boson [10]. Assuming that the Higgs boson exists, the.

(19) 1.3 The ATLAS Experiment. 15. ATLAS Barrel Inner Detector –. H→bb. b. –. b. Figure 1.2. A H → b¯b event viewed in the Inner Detector at low. luminosity, projected on a plane orthogonal to the beam axis.. production processes of the boson and its decay modes depend on the Higgs mass and other parameters in the theory. One plausible decay channel is H → b¯b, where the b and ¯b quarks subsequently decay. The Higgs boson has to to be reconstructed from the nal decay fragments. An example of how such an event could look in the Inner Detector is show in Figure 1.2. Each event in the Inner Detector gives raw data in the form of a large number of space points, indicating the presence of of a charged particle in that particular sensing element. In addition, there is transition radiation hits from the TRT indicating that particle passing through the straw probably was an electron. This set of space point hits, indicated as points in Figure 1.2, is processed to give the particle trajectory. Parameters of reconstructed track and ensembles of tracks are extracted. An ensemble of tracks, conned in a cone and originating from the same point is commonly known as a jet of particles. In this example, the following track analyses are important:.

(20) Chapter 1 Introduction. 16 • • • • •. Primary vertex reconstruction. Charge determination Secondary vertex. Transverse momentum. Lepton b-tagging.. The interaction point of the colliding particles is constrained by the beam envelope. Each particle bunch has a Gaussian distribution of protons in space, with σx = σy = 15 µm and σz = 5.6 cm. Hence the collision point, the primary vertex, is not well dened by the beam envelope and has to be extracted from the reconstructed tracks. Schematically described, all tracks are tted to a common vertex, iteratively excluding tracks that appears to originate from a secondary vertex. The remaining tracks are considered to originate from the same vertex and the interaction point is extracted from the t. The b and ¯b quarks hadronise into b-mesons which have a comparably long lifetime (cτ ≈ 460 µm). They will decay a measurable distance from the interaction point. This is a useful property for identifying jets of particles originating from the decay of a b-avoured hadron. In the decay channel in our example, this identication is crucial to reconstruct the decaying Higgs boson. The secondary vertex can be explicitly reconstructed, determining the decay point of the b-avoured hadron. Alternatively the transverse impact parameter d0 , the minimal orthogonal distance to the z -axis, is determined and used as a discriminating parameter for the identication of b-avoured jets. The identication of b-avoured jets is a trade-o between the eciency of the identication b and the purity. The criteria on the discriminating parameter can be optimised for the requirements of a particular physics analysis. Identication of jets originating from b-avoured hadrons is called b-tagging in the jargon of particle physics. The transverse momentum, pT , is maybe the most crucial parameter to be determined. The momentum of a particle is a key ingredient to determine its four-vector, necessary to reconstruct the invariant mass of the reconstructed Higgs boson in the example of Figure 1.2. Furthermore, the momentum can be matched with the energy measured by the calorimeters as a tool for particle identication as described below. The transverse momentum can be extracted from the curvature of the trajectory of a charged particle in the magnetic eld of the Inner Detector. In practise the momentum is determined as a parameter from the t of.

(21) 1.3 The ATLAS Experiment Tracks. Tracks. 17. 0.08. 0.2. 0.06. 0.04 0.1 0.02. 0. 0. 1. 2. 3. Ecore/p. 0. 0. 0.2. 0.4. 0.6. Fraction of TR hits. Figure 1.3. Left: The ratio of calorimeter energy over track momentum. Right: The fraction of transition radiation hits per track. Distributions of electron tracks from b-decay (shaded) and charged particles in gluon jets (open).. helix tracks to the space points. The sign of the charge is determined from the direction of the trajectory curvature. Lepton b-tagging illustrates the use of the performance of the Inner Detector combined with the Electromagnetic Calorimeter to increase the b-tagging eciency. The method makes use of the presence of associated electrons in b-avoured jets, originating from the decay b → eX which has an inclusive probability of 17% for electrons with pT > 1 GeV. The electron can be identied by matching the measured momentum from the Inner Detector with the energy measured by the Electromagnetic Calorimeter, under the assumption that the track originates from a particle with the mass of an electron. Alternatively, by calculating the fraction of hits above the transition radiation threshold in the TRT over the total number of hits associated with a track. This fraction is a measure of the probability that the track originates from an electron. The discriminating power of this method can be seen in Figure 1.3. This method increases the eciency and purity of the identication of b-avoured jets..

(22) Chapter 1 Introduction. 18. System. Position (cm). Pixel b-layer. Resolution σ (µm). 140 ×. 106. Coverage |η|. read-out channels. r = 4.8. Rφ = 12, z = 66. 2.5. cylinders. r = 11 − 16. Rφ = 12, z = 66. 1.7. disks. z = 47 − 111. Rφ = 12, R = 77. 1.7 − 2.5. SCT. 106. 6.2 ×. read-out channels. cylinders. r = 30 − 52. Rφ = 16, z = 580. 1.4. disks. z = 84 − 278. Rφ = 16, R = 580. 1.4 − 2.5. TRT. 420 ×. 103. read-out channels. barrel. r = 56 − 107. 170 per axial straw. 0.7. end-cap. z = 83 − 340. 170 per radial straw. 0.7 − 2.5. Table 1.4. Table summarising the principal design parameters of the Inner Detector. 1.3.3 Inner Detector Performance The analysis described in Section 1.3.2 and other measurements performed by the Inner Detector can be partitioned into two parts. The rst step is combining the set of space points into particle trajectories. This process is called pattern recognition or track reconstruction. The second step is extracting the relevant parameters from the reconstructed tracks. Both steps are taken into account in the design and the construction of the detector. Some basic design parameters of the Inner Detector are given in Table 1.4. The coordinate system commonly used for ATLAS is a cylindrical coordinate system with the z -axis positioned along the nominal beam axis. The azimuthal angle is denoted φ and the polar angle θ. R is the radial distance from the beam axis. In addition to the polar angle, the pseudorapidity is used to quantify the direction in the R − z plane. The pseudorapidity is dened as η = −log(tan(θ/2)), frequently used in the analysis of physics events. Another often used quantity is the transverse momentum pT . The track reconstruction makes use of advanced pattern recognition.

(23) 1.3 The ATLAS Experiment. 19. algorithms, applying models of expected particle trajectories to join the available space points into continuous tracks. To quantify the success of the reconstruction two quantities are used, the eciency and the fake rate. The eciency is the fraction of accepted reconstructed tracks and the fake rate is the fraction of the reconstructed tracks that are considered to be fake tracks. One possible denition of accepted tracks could be: • Associated hit in the innermost layer. • Associated hit in in minimum one of the remaining two pixel layers • Associated hits in minimum 9 precision planes (of 3 pixel and 8 SCT planes. • Constraints on the impact parameter.. These constraints are invoked to avoid that sets of random hits in the detector generate fake tracks. It is dicult to exactly dene what is meant by fake tracks, since they normally consist of parts of real tracks wrongly reconstructed. The relevant denition depends on the exact physics analysis in question, but criteria complementary to those presented in the list above could be considered. This issue is of particular importance when the LHC has reached its maximal design luminosity, when there will be several proton-proton interactions in each bunch crossing (see Section 1.2). The large number of concurrent interactions will generate a vast number of particle tracks rendering the task of pattern recognition more dicult. The large number of space points provided by the TRT is important as a starting point in the search for tracks, however the quality of the reconstruction is normally determined via the hits in the two precision detectors. Clearly, a high detection eciency for charge particles is an important design criteria to facilitate the track reconstruction, as well as a low rate of noise hits. The momentum for a charged particle is measured by the curvature of its trajectory in the magnetic eld. Hence the directly measured quantity is the sagitta of the trajectory, based on the space points from the tracking detectors. Since the errors on the space points are Gaussian, the errors on 1/pT are expected to be Gaussian. Thus, the resolution is given as σ(1/pT ). The simulated momentum resolution of the Inner Detector versus pseudorapidity η is shown in Figure 1.4. As expected, the resolution in 1/pT is at in the central region of the de-.

(24) Chapter 1 Introduction Relative σ(1/pT). -1. σ(1/pT) (TeV ). 20. 0.8. 0.6. 1.4. 1.2. 0.4 1 TRT SCT pixels. 0.2 0.8 0. 0. 0.5. 1. 1.5. 2. 2.5. |η|. 0. 0.5. 1. 1.5. 2. Relative Detector σ. Figure 1.4. Left: Simulated pT resolution for a 200 GeV muon versus pseudorapidity. Right: Relative pT resolution as a function of detector resolution of the three sub-detectors. tector (|η| < 2). The steep rise in the forward region is a consequence of the particles exiting the tracking volume before reaching its outer radius, hence shortening the lever arm. For low pT particles, another source of error becomes dominant: multiple scattering in the detector material distorting the trajectory. The correlation between the momentum resolution and the spatial resolution of the individual sub-detectors is shown in Figure 1.4. Similarly to the momentum resolution, angular resolution for the Inner Detector is degraded for low pT particles due to multiple scattering. However, in the high momentum range, the angular resolution is dened purely by the geometrical layout and the spatial resolution of the detector. As seen in Table 1.4, the resolution is φ is much higher than the resolution in θ. This design choice is made to maximise the resolution in Rφ, important for the momentum measurement. Simulations show that the angular resolution in φ is better than 1 mrad for high pT , degrading for low momentum due to the multiple scattering, particularly in the forward region, where σ(φ) increase to 4 mrad. For the polar angle θ, the resolution is better than the specication value of 2 mrad in the barrel region for a high momentum particle. It rises as cot(θ) in the forward region, since the space points in the end-cap.

(25) 21 800. σ(z0) (µm). σ(d0) (µm). 1.3 The ATLAS Experiment. pT = 200 GeV With B-layer Without B-layer. 30. pT = 200 GeV With B-layer Without B-layer. 600. 20 400. 10. 200. 0. 0 0. 0.5. 1. 1.5. 2. 2.5. |η|. 0. 0.5. 1. 1.5. 2. 2.5. |η|. Figure 1.5. Simulated impact parameter resolution versus pseudorapidity for a pT = 200 GeV track. Left: Transverse impact parameter resolution. Right: Longitudinal impact parameter resolution. region are measured on planes orthogonal to the beam axis. The impact parameter, the closest distance between a track and the primary vertex, is often split into two projections. The transverse impact parameter (d0 ) is the closest distance projected on an axis parallel to the beam axis, and the longitudinal impact parameter (z0 ) is the projected distance along that axis. The impact parameter resolution is dominated by the Pixel Detector because of its proximity to the collision point and its high spatial resolution. Especially important is the innermost layer, called the b-layer because of its contribution to the secondary vertex resolution in b-hadron decays. For the analysis of such decay, every improvement in resolution has a direct impact on the physics performance, essentially without lower limit. The impact parameter resolution is shown in Figure 1.5 for a high pT track. The importance of the b-layer is clearly visible. Multiple scattering will also in this case degrade the resolution at low momentum. For instance, a pT = 1 GeV track will have σ(d0 ) = 60µm and σ(z0 ) = 300µm at η = 0 with the b-layer, rising sharply in the forward region. The Inner Detector is a detector system combining the performance and features of three dierent detector concepts. The TRT provides a large number of space points at large radius. Due to the long lever arm.

(26) 22. Chapter 1 Introduction. it becomes the most inuential detector for the momentum resolution. The large number of space points are an asset for the track reconstruction. The Pixel Detector provides high resolution space points close to the interaction point, making it the key player in the vertex resolution. The SCT is the intermediate detector system, both in terms of position and resolution. Hence it gives essential contributions to all areas described above.. 1.4 The SemiConductor Tracker The active sensing elements of the SemiConductor Tracker (SCT) [3] are silicon micro-strip detector modules, described in detail in Section 1.4.1. The 4088 detector modules are tiled on the four barrel cylinders and the eighteen end-cap disks, as shown schematically in Figure 1.6. Due to the problems induced by multiple scattering in detector material as described in Section 1.3.3, the whole detector system is designed to minimise the amount of material in the tracking volume. This constraint inuences both the design and the choice of materials used. The support structures, barrel cylinders and end-cap disks, are made by carbon bres. Cooling, cabling and optical bres are routed to each detector module on the carbon bre structures. The barrel detector modules are attached to the cylinders with precision brackets. The end-cap modules are attached via the cooling blocks onto the disks. The cooling system is a bi-phase system using uorocarbons as cooling medium. Each detector module dissipates 5 − 7W thermal energy from the front-end electronics which is absorbed by the cooling system. The high and low voltage necessary for the operation of the SCT detector modules is supplied via low mass power tapes along the support structures. The power tapes are replaced by cables with lower resistance at the boundary of the tracking volume since the material constraints become less severe further away from the collision point. This scheme is repeated again further out from the collision point, to balance the two conicting demands of minimal material and low voltage drops in the supply cables. The clock and commands are distributed via optical links with separate bres to each detector module. There is redundancy option built in since two neighbouring modules can share clock and command in the case of failure on a optical link..

(27) 1.4 The SemiConductor Tracker. 23. Figure 1.6. The 4088 SCT detector modules tiled on the barrel cylinders and the end-cap disks.. 1.4.1 The Silicon Strip Detector Module An SCT detector module [11][12] has four single sided silicon strip sensors mounted pairwise back to back. The strips are read out via frontend Application Specic Integrated Circuits (ASIC) mounted on the detector module. Each detector plane gives a the position of a traversing particle in one dimension, orthogonal to the strips. The second dimension is achieved by combining the information from the two detector planes. The sensors are mounted at an angle of 40 mrad between the sides. The small angle gives rise to a large dierence in resolution in the two directions. A local coordinate system for the module is dened with the y -axis orthogonal to the strips on the top side corresponding to the Rφ direction in the ATLAS coordinate system. The rectangular distributed uncertainty from the 80 µm strip pitch gives the standard deviation σstrip = √8012 = 23 µm. Combining the information of the two σstrip σ √ = 580 µm and σy = strip = 16 µm as planes gives σx = sin(40mrad) 2 stated in Table 1.4. The design of the detector modules for mounting on the barrel cylinders and on the end-cap disks are dierent in many point. The dierences are due to the dierent geometries required and to some deliberate design choices. Figure 1.7 shows a drawing of a SCT barrel detector module. The four silicon sensors are mounted on a baseboard giving mechanical stability and providing a heat path from the sensors to the cooling block. The baseboard is made of a substrate of Very High Thermal Conductivity Pyrolytic Graphite (VHCPG) encapsulated in epoxy. The interface to the cooling block consists of Beryllium Oxide (BeO).

(28) 24. Chapter 1 Introduction. Figure 1.7. Drawing of a SCT Barrel Detector Module. facings laminated to the baseboard. The front-end ASICs are mounted on a ex circuit glued on two Carbon-Carbon (C-C) bridges. CarbonCarbon is a unidirectional material with very good heat conduction and high youngs modulus in the direction of the carbon bres. The direction of the bres is chosen to give a good thermal path from the chips to the cooling interface. The C-C bridges also gives mechanical stability to the ex circuit and the conducting bres give an additional ground plane improving the electrical performance of the detector module. The ex circuit is wrapped around the sensor/baseboard assembly, with the feet of the C-C bridges glued onto the BeO facings providing a gap between the electronics hybrid and the silicon sensors. The electrical connection from the strips to the front-end ASIC is done via ultrasonic wire-bonding and a glass pitch adaptor. A second ex circuit is connected to the detector module housing the optical links for the clock, command and data links. An expanded view of a end-cap detector module [12] is shown in Figure 1.8. The the end-cap detector modules are mounted on disks in the plane orthogonal to the z -axis with the strips on the top-side sensor in the radial direction. The strip pitch varies along the length of the sensors so the strip implants are pointing in the the R direction across the whole detector module. Furthermore, there are three avours of end-cap detector modules depending on their position on the disks; there are inner, middle and outer detector modules. The.

(29) 1.4 The SemiConductor Tracker. 25. Figure 1.8. Expanded picture of a SCT End-cap Detector Module. layout of the silicon sensors of the three avours are adapted to the their respective radial position. Hence the calculation of the detector resolution presented above is not strictly valid for the end-cap, the sensors have strip pitches ranging from 50 to 90 µm. The four sensors are mounted on a spine providing mechanical stability and a thermal path between the detectors and the cooling block. The spine is made of Thermal Pyrolytic Graphite (TPG) coated with a 10 µm layer of Parylene-C for mechanical protection and electrical insulation. To improve the mechanical stability, the spine is reinforced with Aluminium Nitride ceramics. Similarly to the barrel modules, the front-end ASICs are mounted on a ex circuit attached to a C-C substrate. As seen from Figures 1.7 and 1.8 the positioning of the read-out electronics for the barrel and end-cap design diers. In the case of the barrel, the strips are read out at the mid-point of the two daisy-chained sensors whereas for.

(30) 26. Chapter 1 Introduction. Figure 1.9. Block diagram of the ABCD3T front-end ASIC. the end-cap the sensors are read-out at the end of the pair.In the case of the end-cap detector modules, the decoding circuits for the optical links are located on the electronics hybrid. Each silicon sensor has 768 p+ in n strips with a strip length of approximately 6 cm, the two sensors on each side are daisy-chained with wire bonds. The strip in-plants are capacitively coupled to the read-out pads connected to the pre-amplier. The sensor is 285 µm thick, making the most probable deposited charge 3.6 fC for a minimum ionising particle. The reverse bias for the sensor is supplied via the back-plane of the silicon sensors. There are twelve front-end ASICs per detector module, named ABCD3T [13][14][15], each one supporting 128 read-out channels. The ASICs are produced by ATMEL (Nantes) using the radiation hard DMILL [16] BiCMOS process. A block diagram of the ABCD3T chip is shown in Figure 1.9. The ASIC has an analogue front-end, an input register for masking of failing channels and a pipeline memory. The operation of the front-end is controlled by a number of Digital to Analogue Converters (DAC). The event data is buered and compressed before transmission to the o-detector system. The analogue part is made in bipolar technology featuring preamplier, shaper and discriminator. The discriminator gives only hit/no hit information for each channel. This binary readout was chosen to reduce the data quantity already on board the ASIC and consequently.

(31) 1.4 The SemiConductor Tracker. 27. reduce the complexity of the o detector data acquisition system. The front-end has the provision of tuning the bias currents in the preamplier and shaper circuitry, providing a facility for performance optimisation. The discriminator threshold is set globally for all channels on the chip. However, due to the small but inevitable mismatch in device parameters between channels, there is a certain threshold spread. To reduce the channel-to-channel variation a re-programmable oset correction circuit for each channel has been implemented. This procedure is called trimming and the values for the oset corrections are set by DACs. There are four dierent step sizes for the DACs, the smallest step size give a very good matching when the oset spread is low, before the eects of radiation damage appears. The larger step sizes will be used toward the end of the SCT operation, when the channel matching will be poor due to radiation damage. The digital part of the chip is implemented in CMOS technology and is operated at 40M Hz , synchronous with the LHC bunch crossing clock. It has a 132 cell long pipeline for storage of event data awaiting the Level 1 trigger decision followed by a read-out buer for selected events. The ABCD3T chip has a mask register to suppress the output of failing channels. The ASIC has on chip zero suppression and encoding of event data for transmission to the o detector systems. The rst ASIC on each side of the module is denes as master, with its dataoutput connected to the optical links. The event data from all chips is read out via the master, transmitted from chip to chip via a token passing scheme. In case of a failure, an ASIC can be bypassed in the read-out chain or a whole module can be read from one of the masters. The ABCD3T chip has a on-chip charge injection circuit, that injects signals in the pre-amplier for verication and characterisation of the electrical performance. The ASIC is powered via separate low voltage supply lines for the analogue and digital part.. 1.4.2 Detector Module Design Specications Section 1.3.2 describes the role of the Inner Detector in the analysis of the high energy physics events that the ATLAS detector is designed to investigate. Section 1.3.3 is dening the tools and the concepts used in this analysis and how the Inner Detector is performing in measuring the relevant parameters. This section will show how this translates into specications and requirements on the SCT detector module. There.

(32) 28. Chapter 1 Introduction. are many competing requirements, sometimes one in direct opposition to another, so the nal specication is by necessity a compromise. The choice of silicon strip layout as opposed to for example pixel layout is made to reduce the number of read-out channels. For a two dimensional read-out with strip layout, the number of channels grows like 2n with increasing granularity. For a pixel layout, that is a 2D matrix of sensing elements, the channel number increases as n2 . However, the strip layout introduces the problem of ghost-hits in the detector. If a detector module has two tracks traversing the detector planes in the same event, it will generate two hits on each detector plane. These hits can be combined into four space points, two corresponding to the correct combination of hit strips and two indistinguishable ghost hits as a result of the wrong combination. This ambiguity can only be resolved via pattern recognition of tracks. In the extremely high track density of the Pixel detector this problem becomes unmanageable, hence the choice of a pixel layout. Since the SCT is situated further away from the collision point, the track density has decreased suciently to allow a strip layout. This explain the large dierence in read-out channels seen in Table 1.4. The strip pitch, 80 µm in the case of barrel detector modules, is a compromise between the requirement of spatial resolution and the number of read-out channels. The dependence of the pT resolution on the spatial resolution can be seen in Figure 1.4. The theoretical spatial resolution given in Table 1.4 is the uncertainty of the track position in the local coordinate system of the detector module, assuming a mechanically perfect module. There will be two further contributions to the uncertainty of the track position. Small mis-alignments of the silicon sensors will introduce errors in the track position with respect to the local coordinate system. Secondly, the position is transferred to the ATLAS coordinate system based on the location of the detector modules. Consequently, uncertainties in the exact position of each detector module will be translated into uncertainties in the track position. This leads to the specication of the alignment of SCT, both within a detector module and the alignment of detector modules in the ATLAS coordinate system. The philosophy behind the requirements is that the resolution should not degrade more than 20% for any of the parameters mentioned in Section 1.3.3 due to alignment errors. Using this upper limit, the tolerances for mis-alignment can be distributed in the.

(33) 1.4 The SemiConductor Tracker System. 29. ∆R (µm). ∆z (µm). ∆Rφ (µm). SCT Barrel. 100. 50. 12. SCT End-cap. 50. 200. 12. Table 1.5.. Maximal allowed total alignment error for the SCT detector modules, causing a degradation of less than 20% in track parameter resolution. three spatial dimensions on the three sub-systems in the Inner Detector. Simulations of the impact of alignment errors on track parameter resolution [17] resulted in tolerances presented in Table 1.5. There is of course some freedom in the exact distribution of the errors, but this choice is based on what is judged feasible for the dierent systems. The alignment of the SCT detector modules in the ATLAS coordinate system is done in several steps. After the assembly of the detector modules onto cylinders and disks, there will be an x-ray survey of the assembled SCT. The detector modules will be used as sensing elements in the tomography where collimated beams of x-rays will be sent through the detector [18]. Furthermore, there is an in-situ laser interferometry system making precision measurements of the detector module positions [19]. This data will be used as a starting point for the ne tuning of the alignment via tracks. The method makes use of tracks traversing the overlap region between two detector modules, giving their relative position with high precision [20]. The result of this procedure will be a database of the exact location of each detector module, from which o-line corrections on the track parameters can be done. To facilitate this complicated alignment process, the specication on the assembly of SCT detector modules is such that it can be treated as one rigid and mechanically perfect object. This requirement puts harsh constraints on the silicon sensor alignment in the detector module assembly. For instance, the error due to mis-alignment within the module in the most critical direction (Rφ) should be negligible compared to the 12 µm stated in Table 1.5, hence the tolerance is set to 4 µm RMS. The survey of the mechanical assembly precision is done by measuring the position of ducial marks in the lithography on the silicon wafers. The position and direction of the four wafers with respect to the.

(34) Chapter 1 Introduction. 30. Figure 1.10. Denition of the coordinate system and the parameters for the survey of the alignment precision on a SCT detector module.. two mounting holes give three degrees of freedom per wafer and one degree of freedom in the direction of the line between the mounting holes. The convention for module survey [21] for SCT detector modules denes thirteen parameters to be measured and their accepted tolerances. Parameter. mnemonic. Tolerance. Mounting hole. mhx, mhy. 30 µm. Mounting slot. msx. 100 µm. Mounting slot. msy. 30 µm. Mid-point, front pair. midxf. 10 µm. Mid-point, front pair. midyf. 5 µm. sepf, sepb. 10 µm. Sensor angle. a1, a2, a3, a4. 0.13 mrad. Stereo angle. stereo. 0.13 mrad. Wafer separation. Table 1.6. Allowed maximal deviations from the design value for the parametrisation of the mechanical precision of a SCT detector module..

(35) 1.4 The SemiConductor Tracker. 31. These parameters will also be stored in the SCT production database for future detector module selection and possibly for o-line corrections. The denition of the alignment parameters is illustrated in Figure 1.10. The mid-point and the orientation of each of the four silicon wafers are determined from the ducial marks. From the line joining the centres of the pair of wafers on the same side, the mid-point of each side is determined. The survey coordinate system has its origin half way between the mid-points of the two sides, with the Xm-axis bisecting the two centre lines of the front and back side. In this coordinate system, the positions of the two mounting holes (msx, msy, mhx, mhy) and the mid-point of the front side (midfx, midfy) are expressed. Furthermore, the half stereo angle between the mid line of the front side and the Xm-axis and the angles between the wafers and the centre lines on the two sides (a1, a2, a3, a4) are determined. The last two parameters are the separation between the two silicon wafers on the front and back side (sepf, sepb). The requirements expressed in Table 1.5 give the accepted tolerances of the thirteen parameters presented in Table 1.6. The atness of the detector module is important for two reasons. Deviations from the ideal shape introduce an uncertainty of the space points in z and R direction for barrel and end-cap detector modules, respectively. Hence if the deviations from the ideal shape are not negligible compared to the tolerances given in Table 1.5, o-line corrections have to be made. Furthermore, the clearance between the detector modules mounted on the barrel cylinders are very small. The detector modules have to t within an envelope of a few hundred µm to be installed and operated safely. A consequence of the high luminosity requirements, as explained in Section 1.2, is a high event rate. The LHC 40 MHz bunch crossing rates allows for 25 ns to process each event in the front-end of the ABCD3T. Hence the requirement of a shaping time of approximately 20 ns and for double pulse resolution of 80% detection eciency for two hits in the same channels two 25 ns time slots apart. The requirement of local buering of event data awaiting the Level 1 trigger decision is 2.5 µs plus latency introduced by the SCT read-out system. The pipeline of 132 cells gives 3.3 µs storage time. The pipeline is followed by an eight events deep read-out buer to store selected events waiting to be transmitted to the o-detector data processing. The depth of the read-out buer is optimised for an average occupancy of 1% and.

(36) 32. Chapter 1 Introduction. a Level 1 trigger acceptance (L1A) rate of 100 kHz, as expected at maximal luminosity. The 100 kHz L1A frequency allows for 10 µs on average to transmit one event, with the transmission speed of 40 Mbit/s, corresponding to 400 bits of data. For each hit channel, the address of the channel (11 bits) plus a three bit hit-pattern and three bits of header is transmitted, giving 17 bits per hit. Hence at 1% occupancy there are 261 bits plus event header (19 bits) and trailer (16 bits) to transmit if a full module of 1536 channels is read out via one optical link. In terms of noise performance, the critical parameter for a system with binary read-out is the noise occupancy. Noise occupancy is dened as the fraction of channels that are registering a hit due to front-end noise. High noise occupancy cause problems in several areas, principally the track reconstruction (Section 1.3.3) and the data transmission. Simulations of the inuence on track reconstruction [22] [23] show that the track reconstruction eciency is not much aected by the noise hits, but the fake rate increases steeply. The absolute number of fake rates depends on their exact denition. Quantitatively, if the noise occupancy increases from 0 to 0.5% the fake rate will quadruple, reaching levels of 10−3 −10−2 depending of the denition of fake tracks [23]. As seen from the bit transmission estimates above, there is some margin above the expected 1% occupancy from tracks. Dead-time simulations [24] using expected track distributions and the eight event deep read-out buer of the ABCD3T show that the system will reach a non-zero dead-time at 1.5% occupancy and 1% dead-time at 2% occupancy. To allow for margin against simulation errors and noise introduced by the system, the requirement for a single detector module is a noise occupancy of 5 × 10−4 at the expected operating threshold of 1 fC. The signal in the silicon sensor is generated by a deposited charge drifting in the electrical eld, and this signal is amplied by the charge sensitive amplier of ABCD3T. The most probable value of the deposited charge is 3.6 fC, hence for a estimate of the required Signal to Noise ratio (S/N) it is desirable to express the noise specication in Equivalent Noise Charge (ENC). The ENC is often expressed as the standard deviation of the noise amplitude spectrum in units of the elementary charge (1.6 × 10−4 fC), denoted electrons or e− . Assuming perfect channel-to-channel matching of the discriminator threshold and a Gaussian noise amplitude spectrum, the requirement of a noise occupancy of 5 × 10−4 implies that the threshold should be at 3.3σ . Since.

(37) 1.4 The SemiConductor Tracker. 33. the expected operating threshold is 1 fC, this implies an ENC of 0.3 fC or 1900 e− , including contributions from common mode noise. The assumption of perfect discriminator matching is not realistic, in practise the matching will be determined by the step size of the threshold correction DACs. The channel to channel variation in threshold will aect the noise occupancy in exactly the same way as a noise charge. In the early operation of SCT, before any eects of radiation damage degrades the performance, the ABCD3T chips are operated using the smallest step size of the threshold correction DACs. The small step size (4 mV) and the high gain (60 mV/fC) gives a threshold spread of 0.02 fC, negligible when added in quadrature with the front-end noise. However, toward the end of the operation of SCT, when radiation damage is considerable (Chapter 3), the oset spread between the channels will demand operation with a larger step size for the threshold correction DACs. Furthermore the gain will decrease, increasing the step size expressed in charge even further. Assuming operation at the highest step size (16 mV) and a gain of 35 mV/fC, the standard deviation of the spread will be 0.13 fC. This will add in quadrature with the noise charge, eectively reducing the allowed ENC in the front-end to 0.27 fC or 1700 e− for a noise occupancy of 5 × 10−4 at 1 fC. The operating threshold is of course a trade-o between the noise occupancy and the detection eciency at that threshold. The eciency is dened as the probability that particle traversing a sensor plane will induce a hit in the corresponding channel. The consequences of reduced eciency will be a decreased track reconstruction eciency (see Section 1.3.3) and a smaller average number of space points per track. Simulations [23] show that the track reconstruction eciency will degrade a couple of percent at 95% detection eciency for the SCT, depending of the denition of accepted tracks. This simulation assumes randomly distributed ineciencies, which will probably not be the case since failures most likely will occur in front-end ASICs or detector modules. Hence in reality the degradation will probably be greater. The resolution of track parameters, e.g. the transverse momentum, degrades monotonously with a decreased number of space points per track. To leave some room for future failures and simulation failures, the specication is 98% eciency for detector modules to be installed in SCT. This specication is normally divided into 99% of perfect channels on produced modules (Chapter 5) and a 99% eciency on perfect chan-.

(38) Chapter 1 Introduction. 34. Position. TID [kGy/yr]. NIEL 1 MeV equiv. n/cm2. SCT Barrel 3. 8.3. 1.7 × 1013. SCT Barrel 6. 3.8. 1.1 × 1013. SCT Disk 1. 6.0. 1.4 × 1013. SCT Disk 9. 6.8. 1.8 × 1013. Table 1.7. Estimated yearly radiation levels in some locations in the SCT. The values for the end-cap disks are for the position of the front-end electronics.Simulations by the ATLAS Radiation Task Force [25] force using the layout of June 2002. nels (Chapter 4). Thus, the combined requirement is to nd a window of operation for the threshold, where both the specication on noise occupancy and eciency are fullled. As discussed in the Section 1.3.3, material in the tracking volume leads to multiple scattering of the tracked particles. This degrades the resolution on all track parameters. Obviously, the material in the tracking volume must be kept to a minimum. There is no hard specication on the amount of material allowed, but every single component has to be optimised to minimise multiple scattering. This constraint is normally competing with other performance criteria, and a compromise has to be found in every particular case. The extremely small cross sections of the reactions under study at LHC require very high luminosity. However, the total reaction cross section is much larger, giving a large number of proton-proton interactions. Consequently, as explained in Section 1.2, the components of ATLAS will be subject to very high radiation doses. The radiation levels are expressed either as the Total Ionising Dose (TID) or as the Non-Ionising Energy Loss. The TID is the deposited ionisation energy per unit weight in units of Gray 1 Gy = 1 J/kg or 1 Rad = 10−2 Gy. The NIEL is scaled to the energy loss of an equivalent uence of 1 MeV neutrons. In these units, the specication for radiation hardness for the SCT components is to be fully functional after 10 MRad = 100kGy TID and 2 × 1014 1 MeV neutrons/cm2 equivalent. This specications are based on simulations of the radiation environment for SCT [25] [26],.

(39) 1.4 The SemiConductor Tracker. 35. assuming ten years of operation and a 50% safety margin. Table 1.7 show estimated radiation levels for dierent regions of SCT for one year of high luminosity running. The expected energy spectra for dierent particle types can be found in [26]..

(40) !$. Chapter 1 Introduction.

(41) CHAPTER 2. Technology and Methods 2.1 Device Physics Charged particles traversing matter interact with the electrically charged particles present in the material, primarily the electrons. This leads to ionisation in the material. The amount of energy deposited by this process is described by a theory presented by Bethe, Bloch and others [27]. This ionisation energy is the source of the signal in many radiation detectors. The deposited energy can be detected by the free charges created by the ionisation: electrons, holes or ions. In many types of ionising radiation detectors the charge is collected by drift an applied electrical eld. Semiconductors are used as radiation detectors, using a pn-junction as the active detection element [28]. A reversed bias is applied across the pn-junction enlarging the volume depleted from space charges. To maximise the active volume, the applied voltage is set to fully deplete the semiconductor. An ionising particle traversing the depleted region will excite electron-hole pairs in the semiconductor. This is often referred to as the deposited charge since it arises from the deposited energy. The deposited charge will drift in the applied eld, and the drift of the charges will induce a signal on the bias contacts. This signal generation process can be used for energy measurements in which the deposited energy is determined. Using the geometry of the devices, this technique can also be used to make position sensitive detectors. The processing technology of semiconductor devices [29] puts very little constraints on the layout of the active pn-junction. Hence position 37.

(42) 38. Chapter 2 Technology and Methods. sensitive semiconductors can be designed to any desired geometry. The treatment of charged particles' interaction with matter described by the Bete-Bloch theory gives the average energy deposited per unit track length. The energy loss in a given volume for a particular track is governed by random processes, hence the amplitude is described by a probability distribution. For thin absorbers, the distribution is described by Landau theory [30] illustrated in Figure 2.3. The ionisation charge can be expressed as the number created electron-hole pairs or the the equivalent charge in Coulomb. For the 285 µm thick silicon sensors used for SCT detector modules, the most probable value of the deposited charge is 22300 electron-hole pairs or 3.6 fC. The particles generating the signal also damages the silicon by its deposited energy. The received radiation dose of a device is expressed both as the ionising and non-ionising energy loss. The Total Ionising Dose (TID) is the amount of ionising energy deposited per unit weight. The Non-Ionising Energy Loss (NIEL) arise from elastic or inelastic interactions with the silicon nucleons in the lattice.. 2.1.1 Radiation Damage in Silicon Sensors Both ionising and non ionising energy deposition damage the silicon sensor, however the governing damage is caused by interactions with the silicon lattice [28]. The processes involved are for instance displacement of silicon atoms or nuclear reactions. The primary defects formed by the interaction of the traversing particle are normally not stable; vacancies and interstitials are mobile at room temperature. These primary defect can either anneal by e.g. an interstitial lling a vacancy or they can interact with other defect to form stable complexes. The defect can interact either with dopant atoms, with imperfections originating from the fabrication or with other radiation induced defects. The electrical properties of the defects are of course aecting the properties of the silicon on the macroscopic scale. Donors and acceptors can be removed by the formation of inactive defect complexes containing the dopant atoms. Moreover, defect complexes can act as eective donors or acceptors. Hence the apparent doping of the silicon change with irradiation dose. The eective doping of n-doped silicon will gradually decrease with increasing dose, reaching the point of intrinsic silicon. If the dose is further increased the creation of acceptor levels will continue and the doping polarity.

(43) 2.1 Device Physics. 39. Figure 2.1. The eective doping measured as a function of NIEL equivalent 1 MeV neutron uence [31] .. of the silicon changes. The silicon will become p-doped. The eective doping as a function of normalised uence is shown in Figure 2.1 [31]. This phenomenon is called type inversion and has to be taken into account when designing silicon detectors for harsh radiation environments. Defect complexes creating energy levels in the band gap will act as generation-recombination centres. The currents generated by these centres are contributing to the increase of the reversed bias leakage current [32]. The leakage current introduce shot noise in the pn-junction and hence degrades the performance of the detector. This also has implications on the cooling of irradiated silicon sensors. The leakage current in a reversed biased diode depends exponentially on the temperature so the heat generated by the leakage current induces a further increase in current, which can lead to thermal run-away. Defect complexes also act as trapping centres, where electrons or holes are trapped and re-emitted with some time delay. This is problematic due to the fast shaping time used by the LHC experiments. The signal is generated by the charge collected withing a 25 ns time window, hence signal is lost due to trapping [33]. The requirements on the SCT detector module is that 91% of the deposited charge should be collected.

No results found