Proceedings for TASI 2009 Summer School on "Physics of the Large and the Small": Introduction to the LHC experiments
aa r X i v : . [ h e p - e x ] A p r October 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc Proceedings for TASI 2009 Summer School on“Physics of the Large and the Small” :Introduction to the LHC experiments
E. HALKIADAKIS ∗ Department of Physics & Astronomy, Rutgers University,Piscataway, New Jersey 08854, U.S.A. ∗ These proceedings are a summary of four lectures given at the Theoretical Ad-vanced Study Institute in Elementary Particle Physics (TASI) in 2009. Theselectures provide a basic introduction to experimental particle physics and theLarge Hadron Collider experiments at CERN, with many general examplesfrom the (still running) Fermilab Tevatron.
Keywords : Elementary Particle Physics; Collider Physics
1. Introduction
The Standard Model (SM) of elementary particles, summarized in Fig. 1,has been quite successful in making predictions, confirmed to incredibleprecision in experimental data. Yet there are still many unanswered ques-tions about nature and the fundamental interactions. Some of today’s mostchallenging questions in physics are, but not restricted to: • Is there really a Higgs boson, as predicted by the Standard Modelof particle physics? If so, what is its mass? • If not, what is the origin of electroweak symmetry breaking? • Why is there a hierarchy of masses? • What are the origins of dark matter and dark energy? • Why is there no anti-matter in the universe? • How does gravity fit into all this?The dawn of a new energy frontier has arrived with the recent turn-onof CERN’s Large Hadron Collider (LHC). The LHC experiments at CERNuse state-of-the-art technology and will hunt for answers to many of the ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc open questions in high energy particle physics today. From the discoverypotential of the Higgs boson, to new particle and new phenomena searches,all particle physicists are focused on upcoming LHC results.What follows is a summary of four lectures given at TASI in 2009. Theselectures provide a basic introduction to experimental particle physics, withan emphasis on CERN’s LHC experiments. First, I begin with an overviewof particle accelerators (Sec. 2) with an emphasis on the currently run-ning hadron colliders, the Fermilab Tevatron and the LHC. Next, I reviewthe importance of luminosity (Sec. 3), the proton composition (Sec. 4) andhadron collisions (Sec. 5), followed by a summary of a few key definitionsevery high energy physics should know (Sec. 6). I then review how parti-cles interact with matter (Sec. 7) and how those interactions are used indesigning particle detectors (Sec. 8) and the identification of particles foranalysis (Sec. 9). Finally, I describe the importance of a trigger (Sec. 10),the current status of the LHC (Sec. 11), I highlight a few of the early LHCphysics measurements expected (Sec. 12) and conclude (Sec. 13). ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc
2. Particle Accelerators
Particle accelerators are shaped in one of two ways: • Linear colliders or LINAC:
An example of such an accelerator isthe Stanford Linear Accelerator Center (SLAC). • Circular or synchrotron accelerators:
These provide higher energiesthan a LINAC, such as the Fermilab Tevatron.Accelerators can also be arranged to provide collisions of two types: • Fixed target experiments:
When particles are shot at a fixed target.The center-of-mass energy, √ s , for this class of experiments is: √ s = p E beam m target (1) • Colliding beam experiments:
When two beams of particles are madeto cross each other. In this case, √ s = 2 E beam (2)Circular accelerators have been arranged to collide electrons and positrons(for example at LEP) and protons and (anti-)protons (hadron colliders).Scattering experiments have been also done by colliding leptons (electronsor positrons) and protons (for example at HERA). Examples of hadroncolliders are the Tevatron at Fermilab or the Large Hadron Collider atCERN. Hadron colliders provide much higher energies than e + e − collidersand do not suffer from synchrotron radiation. However, e + e − colliders canprovide us with clean environment for precision measurements.The two currently running hadron colliders, the Tevatron and the LHC,are further described in the following sections. Fermilab Tevatron
The Fermilab Tevatron, located roughly 30 miles west of Chicago, IL, ac-celerates protons and anti-protons to √ s = 1 .
96 TeV. The main ring isroughly 4 miles in circumference and when running collides 36 bunches ofprotons against 36 bunches of anti-protons, with roughly 100 billion parti-cles in each bunch. Once injected, the beam is stored and the same bunchesare collided typically for 20-30 hours.The Tevatron hosts two “general purpose” experiments, the ColliderDetector at Fermilab (CDF) and DO. Run I of the Tevatron lasted from1992-1996 and in 1995 the two experiments announced the discovery of the ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc top quark. At that point, the Tevatron entered a fixed target phase and then2001 marked the start of Run II and will continue until at least FY2011. Large Hadron Collider
The Large Hadron Collider at the CERN laboratory near Geneva, Switzer-land, is a proton-proton collider with a 27 km circumference. It is designedto provide collisions with a maximum √ s = 14 TeV. In November-December2009, the LHC turned on and collided protons at √ s = 900 GeV and forthe first time at √ s = 2 .
36 TeV, exceeding the center-of-mass energy of theTevatron. On March 30, 2010 the LHC achieved collisions at √ s = 7 TeV,launching a new era in particle physics. The LHC will also collide heavyions (Pb-Pb) for shorter running periods of roughly 1 month per year.The LHC tunnel rests 100 meters underground. The beams circle thering inside vacuum pipes guided by super-conducting magnets. There arethousands of magnets directing the beams around the accelerator, including1232 15 meter long, 35 ton dipole magnets shown in Fig. 2. These dipolemagnets have an ingenious configuration called a “2-in-1” design allowingthe two proton beams to point in opposite directions in each pipe. For a7 TeV energy beam, the dipoles are cooled to a temperature of 1.9 o Kproviding an 8.4 T magnetic field and a current flow of 11.7kA.The LHC is designed to collide a maximum of 2808 proton bunchesagainst another 2808 proton bunches. Each bunch is several cm long andcontains approximately 100 billion protons. In order to increase the prob- ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc ability of a hard collision, the beam is squeezed as much as possible at theinteraction point to a diameter of tens of microns. For these operating de-sign conditions, it is expected that on average 20 additional pp interactionswill occur.The LHC accelerator chain is shown in Fig. 3. Initially, 50 MeV protonsare produced in the LINAC and accelerated to 1.4 GeV in the Booster.They are then injected in the Proton Synchrotron (PS) where they reachan energy of 26 GeV and are further accelerated to 450 GeV in the SuperProton Synchrotron (SPS). Finally, they are injected in the main ring wherethey reach a maximum energy of 7 TeV (the maximum to-date has been3.5 TeV per beam).The collisions at the LHC take place at the location of the four experi-ments, which are: • Compact Muon Solenoid (CMS): One of the two large “generalpurpose” experiments. • A Toroidal LHC Apparatus (ATLAS): The other of the two large“general purpose” experiments. ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc − cm − s − . × (April 2010)Integrated Luminosity / year 10-100 fb − > fb − (2008) • LHCb: Designed to study the b -quark sector, CP violation and raredecays. • A Large Ion Collider Experiment (ALICE): A heavy ion experimentdesigned to study the nature of quark-gluon plasma.These lectures focus primarily on the CMS and ATLAS detectors. Fi-nally, Tab. 1 shows a summary of the LHC and Tevatron parameters forcomparison.
3. Luminosity
Important parameters in colliders are the energy of the beams and the rateof collisions ( R ), or the luminosity ( L ). R , is defined as: R = dNdt = L σ, (3)where dNdt is the number of hard collision events produced per second, and σ is the cross section of the process produced. Integrating over time, weget: N events produced = σ × Z L dt, (4)where N events produced are the number of produced hard collision eventsof the process with cross section σ and R L dt is the integrated luminositywhich is provided by the accelerator in a given time period. Unfortunately,a given high energy physics detector does not observe every collision eventthat is produced. For example, the trigger is inefficient, as is the identi-fication of the final state particles, and some fraction of the events maybe produced beyond the detector acceptance (see Sec. 12.1). These ineffi-ciencies need to be experimentally evaluated and once accounted for theexpression becomes: ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc N events observed = σ × Z L dt × ǫ (5)where N events observed is now the number of events observed in the detector,and ǫ is the total efficiency of identifying the collision event of interest (seeSec. 12.1).The units of a cross section are the same as the units of area and in highenergy physics are typically represented by a barn (1 barn = 10 − cm ),for example, mb, µb, nb , etc. The units of instantaneous luminosity are thesame as the units of [1 / (cross section × time)], for example cm − s − . Integrated luminosity has units of [1 / cross section], for example cm − or pb − , f b − , etc.An example of the difference between integrated and instantaneous lu-minosity is shown in Fig. 4. The top figure shows the initial luminosity de-livered by the Tevatron versus time and the increasing slope demonstratesthe challenges of increasing the luminosity at a hadron collider. It shouldbe noted that the instantaneous luminosity drops as the protons collide,until the next store or fill is dropped followed by (anti-)protons being re-injected and collisions resume. The bottom of Fig. 4 shows the integratedluminosity delivered by the Tevatron (black) and that acquired by the CDFexperiment (purple) as a function of time; it is impossible to record everycollision at a hadron collider and the difference between the two curvesshows how efficiently the experiment (in this case CDF) collects the datathat the accelerator delivers.Next, let us consider an alternate expression for luminosity: L = f n n πσ x σ y ≈ f n b N p πσ x σ y , (6)where n and n are the number of particles (protons) in each of the collid-ing bunches, f is the frequency with which they collide, σ x and σ y repre-sent the size of the transverse beam (e.g. the RMS if we assume a Gaussianshaped beam), n b is the number of bunches and N p is the number of par-ticles per bunch. So in order to increase the luminosity, it is important to squeeze as many protons in as small a transverse beam spot as possible. Exercises (1) Imagine a hadron collider such as the LHC or the Tevatron runs forone year with and instantaneous luminosity of 10 cm − s − , how much integrated luminosity will be delivered to an experiment? ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc × cm − s − ) delivered by the Tevatron vs. time.Bottom: Integrated luminosity in pb − delivered by the Tevatron (black) and acquiredby the CDF experiment (purple) vs. time. Answer:
A year is 3 × seconds, however, accelerators do not operateevery day. Assuming a good year of running is 10 seconds, we get arough estimate: Z L dt = 10 cm − s − × s = 10 cm − = 10 barns = 100 pb − (2) In 100 pb − p ¯ p → t ¯ t events will be produced at theLHC at √ s = 7 TeV? ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc Answer:
The p ¯ p → t ¯ t cross section at 7 TeV is ∼ pb . N events produced = σ × Z L dt = 165 pb × pb − = 16 , t ¯ t pairsPrecisely how many events are observed depends on the efficiency ofobserving them in the detector.(3) What size beam spot is needed for L = 1 × cm − s − at the LHC? Answer:
The LHC machine frequency is f = c/
27 km = 11kHz, and isdesigned to contain n b = 2808 bunches and N p = 1 × protons perbunch. Substituting this into Eq. 6 above and solving for σ (assuming σ x ≈ σ y ) gives: σ x,y = s kHz (2808)(10 ) π (10 cm − s − ) = 1 . × − cm = 15 µm So we will need approximately 15 µm beam size. For comparison, theTevatron beam size is ∼ µm .
4. Proton Composition
The proton is composed of three valence quarks (two up quarks and onedown quark) as well as gluons and sea quarks, but the exact composition isquite complicated. The mixture of partons inside the proton depends on theBjorken- x (the fraction of the proton’s momentum carried by the parton)and Q (the momentum scale that characterizes the hard scattering, suchas M , where M is the mass of the particle that is created by the scatteringprocess). These quantities, x and Q , are also what parameterize PartonDistribution Functions (PDF’s), as seen in Fig. 5, which help describethe content of the proton. For low values of Q ( Q < ) the protonbehaves predominantly as a single particle. For a medium energy range (1 < Q < GeV ), the proton interacts as a composite particle and thevalence quarks dominate in the interaction. At higher energies, the gluonsand sea quark PDF’s are dominant. PDF’s are obtained by global fits todata measurements from many experiments (deep inelastic scattering, fixedtarget, collider) and the constraints are summarized in Fig. 5 (green). Theyare essential inputs to perturbative calculations of production cross sectionsat hadron colliders. There are two main PDF fitting groups, CTEQ andMRST (now MSTW), which regularly provide updates to the PDF fitsand their uncertainties with new data. ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc Q vs. x (green). Also shown is the relationship between these parton variables and thekinematic variables for a final state produced with mass M and rapidity y assuming andLHC energy √ s = 14 TeV (blue). Figure 6 shows the PDF’s vs. x for the valence quarks (up and down),sea quarks (upbar) and gluons (divided by a factor of 10). Note that thePDF’s have a dramatic rise at low values of x and are dominated by gluonsin that region. The valence quarks are dominant for roughly x > .
1. Uncer-tainties in PDF’s quantify our understanding of parton content of protonsand the cross sections of processes. Therefore, making measurements whichare sensitive to constraining PDF’s are important since large uncertaintiesin PDF’s result in large uncertainties in predictions and processes whichare not well understood. PDF uncertainties can vary quite a lot (roughly2-30% or more) depending on the x range and parton of interest. For ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc x for up and down valence quarks, bottom sea quarks, and gluons(times 0.1) for a Q = 10000 GeV . The PDF’s shown are CTEQ6.1M and taken fromthis useful website. example, gluon PDF’s are poorly constrained in the range approximately x > .
5. Hadron Collisions
The collisions, or scattering, which occurs in hadron colliders is separatedinto hard and soft scattering. Calculations of the hard scattering process(when two of the constituent partons in the proton collide head-on) aredone using perturbative QCD. The soft processes (elastic, single diffrac-tive, double diffractive and non-diffractive inelastic scattering) are muchmore difficult to understand and suffer from non-perturbative QCD effects.The majority of the total pp collisions are soft. These soft processes (ev-erything except the hard scatter) is also generally referred to as the “un-derlying event”. The underlying event includes initial state radiation, finalstate radiation and interactions of other remnant partons in the proton.A schematic diagram describing the hard and soft processes in a hadroncollision can be seen in Fig. 7. Additionally, there is a lot about the colli-sion which we do not know, such as which partons collided with each other, ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc
12 Fig. 7. Schematic of a hard scattering proton-proton collision. what the momentum of the partons were when they collided, and what wasthe effect of the other partons in the proton.Figure 8 shows the cross sections for various SM processes as a functionof √ s . The two vertical lines at ∼ √ s . This also again emphasizes that themajority of the total inelastic cross section is coming from soft scatteringprocesses rather than hard collisions. For example, reading from the leftside of the y-axis, the total event rate produced for L = 10 cm − s − atthe LHC is ∼ events per second, whereas the event rate for W bosonproduction is ∼
200 events per second and for t ¯ t is ∼ .
6. Definitions
In this section I outline some definitions that all high energy physicists,both theorists and experimentalists, should know.
Rapidity and Pseudorapidity
The natural coordinates of a typical collider experiment are cylindricalaround the beam-pipe. If we assume the z − axis to be in the direction of ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc the beam, we can define θ as the polar angle and φ as the azimuthal angle,and z = 0 is at the center of the detector or at the interaction point.The rapidity, y , of a particle is a function of the energy, E, and the z -component of the momentum, p z and is defined as: y = 12 log( E + p z E − p z ) = tanh − ( p z E ) . (7)In the coordinate system defined above, the polar angle θ is not Lorentz-invariant. However, what we can define is the pseudorapidity, η , as a func-tion of θ as: ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc η ≡ − log tan( θ/ . (8)We can then define the forward region as η ≥ θ ≈ η ≤ − θ ≈ π ) and the central region as η = 0 (or θ = π/ y , is Lorentz-invariant under boosts along thebeam direction, and for a massless particle (or a nearly massless particlewhere p >> m ) the rapidity and pseudorapidity are approximately equal. Itis also interesting to note, that we can calculate the η of a particle withoutknowing its mass (which is very handy for experimentalists). R Distance
In order to determine the separation in direction between particles, exper-imentalists use ∆ R as a measure of “distance” and is defined as:∆ R = p ((∆ η ) + (∆ φ ) ) , (9)where ∆ η and ∆ φ are the particles’ separation in pseudorapidity and az-imuthal angle, respectively. For example, this is very useful in the recon-struction of “jets”, where we use cones of ∆ R to group particles with eachother; more on this in Sec. 9. Transverse Quantities
Experimentalists also find it useful to focus on quantities measured in thetransverse plane, or the plane perpendicular to the beam z − axis.One quantity that is commonly used is the transverse momentum of aparticle, p T , defined as: p T = p sin θ. (10)Note that the p T is invariant under z − boosts. Particles that escape detec-tion (or end up in the forward region) have close to zero p T . In this sense,the transverse plane is opposite of forward.Additional transverse quantities that are often use are the transverseenergy, E T : E T = E sin θ, (11)and the transverse mass, m T : m T = q E T − p T . (12) ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc One of the most interesting and most difficult quantities for experimental-ists to understand is the missing transverse energy in an event, E T , definedas: E T ≡ − X i E iT ˆ n i = − X all visible ~E T , (13)where ˆ n i is the component in the transverse plane of a unit vector thatpoints from the interaction point to the i th calorimeter tower (see Sec. 8.2).It is an event-wide z -boost-invariant quantity and many new physics signa-tures are expected to show up with large E T . Experimentalists also find itinteresting to look at the measure of the scale of the visible p T in an event,or H T , loosely defined as: H T ≡ X i = objects | ~p T ,i | . (14)The definition of H T varies since it depends on which objects (leptons, jets, E T ) are included in the sum. This is also an event-wide z − boost-invariantquantity which could distinguish a SM final state from one produced bynew physics.So why are experimentalists so interested in the transverse plane? Whynot look for missing p z or missing E? Unfortunately, in hadron collisionsyou do not have the luxury of knowing the initial state exactly. Rememberwhat we said in Sec. 5, the proton itself is not what scatters. The particlesthat do scatter (underlying event) and escape detection have large p z sovisible p z is not conserved and is therefore not a useful variable. However,to a good approximation the visible p T is conserved, which is what makesit so useful.
7. Particle Interactions with Matter
To understand the various LHC detectors (and their differences) first re-quires a basic understanding of the interactions of high energy particleswith matter. Particles can interact with atoms and molecules, atomic elec-trons and the nucleus. These interactions result in several effects such asionization, elastic scattering, energy loss and pair-creation. There are sev-eral respectable sources on interactions of particles with matter andthe main one used here is the PDG. ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc Energy Loss of Charged Heavy Particles
The primary source of energy loss of moderately relativistic heavy chargedparticles, such as muons, pions and protons, in matter is via ionizationand atomic excitation. The average rate of energy loss is described by theBethe-Bloch equation: − dEdx = Kz ZA β [ 12 ln m e c β γ T max I − β − δ ( βγ )2 ] , (15)where z is the charge of the particle, Z is the atomic number of the materialthe charged particle is traversing, A is the atomic number of the material, K = 4 πN A r e m e c ( N A is Avogadro’s number, r e is the classical electronradius, and m e c is the mass of the electron), β and γ describe the rela-tivistic speed of the particle, I is the mean excitation energy and T max isthe maximum kinetic energy of a free electron in the collision. Equation 15is also referred to as the stopping power . The ionization, dE/dx , is typicallyexpressed in terms of M eV / ( g/cm ) and is dependent on the density of thematerial the charge particle is traversing. The minimum ionization is foundto be at a value of βγ ≈
3, and is independent of the charged particle’starget.Additionally multiple coulomb scattering off of nuclei is also an im-portant effect for high energy charged particles since as they ionize whiletraveling through materials, they end up changing their direction with eachinteraction. The distribution of this multiple scattering is described by aGaussian of width θ : θ = 13 . M eVβcp z p x/X [1 + 0 . ln ( x/X )] , (16)where βc is the velocity, p is the momentum, z is the charge of the scatteredparticle and x/X is the thickness of the material in units of radiationlengths X (defined in Sec. 7.2). Equation 16 holds for small scatteringangles, but for high scattering angles large non-Gaussian tails appear. Energy Loss of Electrons and Photons andElectromagnetic Showers
Electrons primarily loose energy via bremsstrahlung and ionization. Therate at which electrons loose their energy by bremsstrahlung is nearly pro-portional to its energy and the rate of ionization loss rises logarithmically.There is a critical energy, E c , at which the two loss rates are equal and it ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc X ( g/cm ) λ n ( g/cm ) H
63 52.4Al 24 106Fe 13.8 132Pb 6.3 193 depends strongly on the absorber. For example, this critical energy for leadis 9.5 MeV.The characteristic length that describes the energy decay of a beam ofelectrons is called the radiation length, X , defined as: X = 716 . gcm − AZ ( Z + 1) ln (287 / √ Z ) , (17)where A is the atomic mass of the material and Z is the atomic number. Itis the average distance the electron travels until its energy is reduced by afactor of 1 /e due to bremsstrahlung. By expressing the thickness in termsof X the radiation loss is approximately independent of the material. Theamount of energy loss of electrons by bremsstrahlung is: − dEdx = EX . (18)As is shown in Tab. 2, higher Z materials have shorter radiation lengths.For example, lead, which has a density of 11 . g/cm has X = 5 . Z materials make good electromagneticcalorimeters.The concept of a radiation length can also be applied to photons. Whenhigh energy photons lose energy in matter they do so via e + e − pair pro-duction. The mean free path, ℓ , for pair production by a photon is: ℓ = 97 X . (19)For electrons, as we just described, ℓ = X . So if we have a high energyphoton passing through an absorber, it will produce electrons, which thenradiate bremsstrahlung photons, and so on, the process repeats. This elec-tromagnetic cascade of pair production and bremsstrahlung generate moreelectrons and photons with lower energy and is referred to as an electro-magnetic shower. The transverse (lateral) development of electromagneticshowers scale with what is referred to as the Moli`ere radius, R M : ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc R M = 21 M eV X /E c , (20)where E c is the critical energy as described above. Hadronic Showers
Hadronic showers are produce by interactions of heavy particles with nuclei.These showers are described by the nuclear interaction length, λ n : λ n ≈ gcm − A / . (21)For heavy, or high Z, materials the nuclear interaction length is quite a bitlonger than the electromagnetic one and λ n > X (see Tab. 2). This resultsin hadronic showers starting later than electromagnetic showers and aremore diffuse. For example, from Tab. 2 lead, which has a density of =11.4 g/cm , has an interaction length of ∼
17 cm.
8. Particle Detectors
The goal of every collider experiment is to completely surround the collisionby arranging layers of different types of subdetectors. In Sec. 7 we justlearned how different particles interact with matter so in order to identifythem we exploit these differences. The key information of the particles thatwe want to extract is their momentum and charge, their energy, and theirspecies.Figure 9 shows schematic drawings of the CMS (top) and ATLAS (bot-tom) detectors, which have the traditional layered detector structure. Thesedetectors have the following general features, starting from center movingoutwards: • Tracking detectors within a magnetic field: measures the charge, trajec-tory and momentum of charged particles • Electromagnetic calorimeter: measures the energy and position of elec-tromagnetic particles • Hadronic calorimeter: measures the energy and position of hadronicparticles • Muon chambers: measures the trajectory and momentum (along withthe tracking detector) of muons ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc
In the sections below I provide a brief description of these detectors, givingexamples from both the LHC experiments and Tevatron experiments. Addi-tional details on particle physics detectors can be found in these Refs. .I summarize the detector technologies used in the CMS and ATLAS detec-tors in Sec. 8.4.
Tracking Detectors
The main goal of tracking detectors is to measure the momentum, chargeand trajectory of charged particles. Ideally, we we want tracking detectorsto contain as little material as possible in order to minimize multiple scat-tering. There are two main technologies of tracking detectors in particle ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc physics: • Gas/wire drift chambers:
These devices are made of wires in a volumefilled with a gas, such as Argon-Ethane. They measure where a chargedparticle has crossed when it ionizes the gas. There is an electrical po-tential applied to the wires so atomic electrons knocked off the atoms inthe gas drift to a positively charged sense wire. The chamber are con-nected to electronics which measure the charge of the signal and whenit appears. To reconstruct the tracks of the charged particles severalchamber planes are necessary. Advantages to drift chambers is theirlow thickness (in terms of X ) and are the traditionally preferred tech-nology for large volume detectors. Typical single hit resolutions rangefrom ∼ − µm . An example of such a device is the CDF ex-periment’s Central Outer Tracker (COT) which has approximately30,000 wires. • Silicon detectors:
Silicon detectors are semi-conductor detectors whichare modified by doping. For example, doping with Antimony gives ann-type semiconductor or with Boron which gives a p-type semiconduc-tor. This doped silicon is then used to create a p-n junction, to whicha very large reverse-bias voltage is applied. This creates a “depletionzone” and once the silicon device is fully depleted we are left withan electric field. When charged particles cross the detector they ionizethe depletion zone and create an electrical signal. Figure 10 shows aschematic drawing of a charged particle interacting in a silicon device,which has a typical thickness of ∼ µm . Silicon detectors come intwo varieties, either metal strips (as seen in Fig. 10) or pixels (shownin Fig. 11) which provide much higher granularity and a higher preci-sion set of measurements. For example, the CMS silicon strip resolutionranges from 8 − µm and for its pixel detector is ∼ − µm . Addi-tionally, the number of pixel sensor channels at CMS is ∼
65 million andat ATLAS is ∼
80 million. These detectors are radiation hard and areimportant for detection secondary vertices (for example, from b-hadrondecays as described in Sec. 9) close to the primary interaction. Siliconis now the dominant sensor material in use for tracking detectors at theLHC and especially for CMS.Since a magnetic field is applied within the detector the momentumand charge of the particle is measured using a few points of the particle’strack (trajectory) which we can use reconstruct the curvature of the track.The transverse momentum ( p T ) of charged particles is proportional to the ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc Fig. 11. Schematic of the silicon pixel detector at CMS. radius of curvature and to the B field. In particular, p T = 0 . q B r, (22)where the reconstructed track p T is measured in GeV/c, B is in Tesla, thetotal particle charge is qe (e is the magnitude of the electron charge) and r is measured in meters and is the radius of curvature of the track. ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc Electromagnetic and Hadronic Calorimeters
Electromagnetic calorimeters are designed to measure the energy of elec-tromagnetic particles (both charged and neutral) and their position. Thisis done by constructing them using a heavy, high Z material to initiate anelectromagnetic shower, as described in Sec. 7.2, to totally absorb the en-ergy and stop the particles. The important parameter for the material usedin electromagnetic calorimeters is the radiation length X , and have typicalvalues of 15-30 X . Additionally, it is key to have as little material beforethe calorimeter as possible (this means the tracker) so that the particles donot radiate before they reach it.The relative energy uncertainty (or resolution), σ E , of calorimeters de-creases with the energy E of the particle and can be parameterized asfollows: σ E E = a √ E ⊕ b ⊕ cE , (23)where a is referred to as the stochastic term and quantifies statistics re-lated fluctuations, b is the constant term and c is primarily due to noise(for example, in the electronics). The three terms in Eq. 23 are added inquadrature (denoted by the symbol ⊕ ).There are two types of calorimeter detectors: • Homogeneous calorimeter:
These detectors are generally made of aninorganic heavy, high Z material which is also scintillating. The ideais to create an entire volume to generate the electromagnetic signal, asseen in Fig. 12 (top). Examples of these calorimeters include a varietyof crystals such as CsI, NaI, and PbWO, and ionizing noble liquids suchas liquid Ar. Energy resolutions of these types of detectors are typically σ E E ∼ • Sampling calorimeter:
These calorimeters are made of an active mediumwhich generates signal and a passive medium which functions as an ab-sorber as seen in Fig. 12 (bottom). Examples of active medium mate-rials are scintillators, ionizing noble liquids, and a Cherenkov radiator.The passive material is one of high density, such as lead, iron, cop-per, or depleted uranium. Energy resolutions of sampling calorimeterdetectors are typically σ E E ∼ ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc The purpose of hadronic calorimeters is to measure the energy of heavyhadronic particles. Hadronic calorimeters are similar to electromagneticcalorimeters but in this case the important parameter of the absorber is theinteraction length λ n . In general, a hadronic calorimeter has λ n ≈ − Muon Chambers
Recall that the muon signature is that of a minimum ionizing particle andextraordinarily penetrating and therefore the detectors for identifying themare the outer-most layer of a collider detector. These detectors are made upof several layers of tracking chambers as described in Sec. 8.1. Their primarypurpose is to measure the momentum and charge of muons. The measure-ments from the muon chambers are combined with the tracks reconstructedwith the inner tracker to fully reconstruct the muon trajectory.Muon chambers in LHC experiments are made from a series of differenttypes of tracking chambers for precise measurements and some examplesinclude: • Drift Tubes (DT’s): Wire chamber devices, so when muons travelingthrough kick off atomic electrons in the gas and drift to the positively ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc From left to right:Drift Tubes, Resistive Plate Chambers and Cathode Strip Chambers. charged wire. • Cathode Strip Chambers (CSC’s): Wires crossed with metallic stripsin a gas volume, so when muons traverse the detectors electrons driftto the positively charged wire as described above. Additionally, thepositive ions in the gas drift to the metallic strips and induce a chargedpulse perpendicular to the wire, giving a two dimensional coordinate ofthe traveling muon. • Resistive Plate Chambers (RPC’s): Oppositely charged parallel platescontaining a gas volume, creating drift electrons when muons cross thedetector.Schematic drawings of DT’s, RPC’s and CSC’s are shown in Fig. 13.
The ATLAS and CMS Detectors
In this section I give a brief summary of the details of the CMS and ATLASdetectors shown in Fig. 9. Additional detailed information can be found inthe technical design reports (TDR) for the two experiments.Both the CMS and ATLAS detectors are large scale experiments inevery sense. CMS is 21 m long, 15 high m and 15 m wide and weighs 12,500tons. The dimensions of ATLAS are even larger, 46 m long, 25 m high and25 m wide, and weighs 7000 tons. CMS is located at Point 5 around theLHC ring in Cessy, France, whereas ATLAS is located at Point 1 and is inMeyrin, Switzerland (see Fig. 3). Due to the high intensity of the collisionsthe detectors will experience, they both have been designed to be veryradiation hard, in particular the tracking detectors closest to the beam-pipe. The ATLAS and CMS experiments have designed their subdetectorsusing different approaches and a summary of the detector technologies usedis shown in Tab. 3. ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc ∼ ∼
9. Particle Identification
In this section, I describe how these detectors described in Sec. 8 are usedfor particle identification. Figure 14 shows a schematic of a transverse sliceof the CMS detector outlining the identification of various particles. Onemay find it useful to refer to this diagram while reading the descriptionbelow. • Electrons and Photons:
Electrons are identified as an energy depositin the electromagnetic calorimeter, and is required to have a showershape (energy loss) consistent with an electromagnetic shower. It isalso required to have little or no energy in the hadronic calorimeter.Since electrons are charged particles it needs to be associated with atrack reconstructed in the tracker, and is therefore required to havea matched position measurement in the calorimeter with the one from ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc the track. If the electromagnetic cluster of energy does not have a trackpointing to it then it becomes a candidate for being a photon. • Muons:
Muon identification begins by reconstructing a track in themuon system which is then matched with a track in the inner tracker.Additionally, since muons are minimum ionizing particles, they are ex-pected to deposit little or no energy in the calorimeters. • Jets:
Jets are created when a quark or gluon is knocked out of theproton and due to parton confinement subsequently a hadron is cre-ated. This hadron forms a jet once it decays and fragments into manyparticles (hadronization), which are essentially collimated object. Thereconstruction of a jet is the experimentalists representation of a par-ton. There are several algorithms for reconstructing jets but overallwhat these reconstruction algorithms do is attempt to group the parti-cles from the hadronization process together and measure the energy ofthe parton. There are two main categories of jet algorithms that exper-imentalists and theorist use to reconstruct jets: (1) Cone algorithms when one draws circles of ∆ R around clusters of energy according tosome rule, and (2) Recursive cluster reconstruction such as the anti- k T algorithm which is now the default jet algorithm of choice for theLHC experiments.Measuring the jet energy has several challenges since it is impossibleto determine which particles came from which hadronization process.There are several effects which contribute to the complication of thejet energy measurement, such as multiple pp interactions, spectatorpartons interacting and noise in the calorimeters. However, experimen-talists have ways of correcting for such effects and this calibration thejet energy is generally called the Jet Energy Scale (JES) and oftendepends on the p T and the η of the jet. • b-hadrons: There is a special category of jets coming from b hadronswhich are long-lived (with cτ ∼ µm ) and massive. There are twostandard techniques for identifying a b hadron decay, referred to as b-tagging . One can look for displaced tracks forming a secondary vertexaway from the primary vertex of the interaction. Alternatively one canidentify soft leptons (electrons or muons) inside the jet, which wouldbe a signature specific to semi-leptonic b decays. • Tau Leptons:
The identification of tau leptons is for hadronically de-caying taus, which decay ∼
49% of the time to a single charged hadronand neutrinos and ∼
15% of the time to three charged hadrons and neu-trinos. Leptonically decaying taus are indistinguishable from “normal” ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc electrons and muons. The reconstruction algorithms for taus assumethat taus form narrow jets in the calorimeter. First one forms a ∆ R cone around clusters of energy and tracks (a signal cone) and a secondlarger ∆ R cone around the signal cone (an isolation cone) where thereis little or no calorimeter and track activity. In the signal cone, oneor three tracks are required as well as electromagnetic energy in thecalorimeters from neutral particles (such as π s). • Neutrinos or E T : Neutrinos are weakly interacting particles and passthrough all the material in the LHC detectors. They are identified indi-rectly by the imbalance of energy in calorimeters. This missing energywas previously defined above in Eq. 13 and recall that it is one of themost interesting and most difficult quantities for experimentalists. Vari-ous effects could contribute to the complications of the E T measurementsuch as dead calorimeter cells or a jet whose hardest hadron enters acrack (between cells) in the calorimeter or an improperly calibratedcalorimeter. Therefore, we need to carefully understand this quantityas it is very important for searches of new physics processes which couldproduce additional weakly interacting particles.
10. Selecting and Storing the Interesting Events: Triggerand Computing
At design a center-of-mass energy of 14 TeV and a luminosity of10 cm − s − , Fig. 8 shows that the total cross section at the LHC willbe ∼ nb. The rate for all collisions will be around 40 MHz. Since it isnot possible to record every collision event, quick decisions need to be made a priori selecting the interesting events worthy of analysis. This filter, ortrigger, needs to single out rare processes and reduce the common processes.We also want to keep less interesting events for “standard-candle” measure-ments (such as jet and W boson and Z boson production cross sections),calibrations, and so on. It is critical to consider carefully the make-up ofthe trigger and make wise choices, otherwise the events will be thrown awayforever.A typical trigger table will contain triggers on: electroweak particles(photons, electrons, muons, taus) at as low an energy as possible, veryhigh-energy partons (jets), and apparent invisible particles ( E T ). Theoryvery often plays a role in guiding these choices, therefore it is important tohave good communication between theorists and experimentalists.The LHC experiments have two levels of triggers, one which bases itsdecision on hardware electronics (L1), and a second level which based on ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc software programming (the high level trigger or HLT). Recall, the startingtrigger rate is 40 MHz, which gets reduces after the L1 trigger to a rateof around 100kHz. The HLT trigger further prunes this down to roughly150-200 Hz, which is the event rate that the experiments record. Therefore,the final decision of the trigger is to keep ∼ The LHC will produce roughly 15 petabytes (15 million gigabytes) ofdata annually. Finally, there is the challenging task of distributing therecorded data around the world for analysis. The LHC has a tiered com-puting model to distribute this data around the world, referred to as theGrid.
11. Status of the LHC
These TASI lectures were given in June 2009 and the LHC has since turnedon and the experiments have been collecting data. It was on November 20,2009 when the LHC first came back online, circulating proton beams of450 GeV and three days later the beams collided for the first time at at √ s = 900 GeV. Then in December, protons collided for the first time everat √ s = 2 .
36 TeV, exceeding the center-of-mass energy of the Tevatron.And on March 30, 2010 the LHC achieved again the highest ever energycollisions at √ s = 7 TeV, and a new era in particle physics commenced.Since then the machine and the experiments have been running smoothlyand has so far achieved a luminosity around 10 cm − s − , with a goal of ∼ cm − s − . The plan is to continue running the accelerator at √ s = 7TeV through 2011 (with a short technical stop at the end of 2010) untilit has delivered > f b − of good collision data to the experiments. Thisdataset will be enough to make potentially very exciting new discoveries inthe near future. ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc
12. Early LHC Physics Measurements
With the initial data from the LHC, in order to have confidence of anypotential claims of discovery, the very first job of the experimentalists is tounderstand the detector. The early LHC measurements will be focused oncalibrating, and aligning the detector as well as rediscovering the SM sinceit is the SM particles which are the only ones we are guaranteed to see.A complete list of expectations for physics measurements from the CMSand ATLAS experiments can be found in their TDR’s and updatedresults located at the experiment websites.
Here I will only list a few examples of early LHC physics measurementsof SM processes. Without measurements such as the ones listed below, wecan not be assured of any claims of discovery of new physics. • Charged track track multiplicity: This measurement has already beenmade by the ALICE, CMS and ATLAS experiments with therecently collected 900 GeV and 2.36 TeV data. • Inclusive jet cross section • Z and W boson production cross sections • t ¯ t pair production cross sectionIn the next section, I will highlight a few of the key elements which goin making an example measurement such as a production cross section. Example Analysis: Measuring a Cross Section
Measurements of the production cross sections of known processes producedin high energy pp collisions provide important tests of the SM. Althoughthe measurement of a cross section is primarily a counting experiment, oneshould not be fooled into thinking it is a simple analysis; it is actuallyrather complex with many ingredients which need detailed understanding.Experimentally, the cross section for a process of interest is measured as: σ = N obs − N bkg R L dt × ǫ , (24)where N obs is the number of observed candidate events selected in the datasample, N bkg is the estimated number of background events mimicking thesignal, R L dt is the integrated luminosity of the data sample analyzed, and ǫ is the overall efficiency of observing the produced events of interest.The evaluation of the background processes which fake your signal canoften be a difficult task. In general, backgrounds are evaluated using a com- ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc bination of Monte Carlo simulations of select processes (such as electroweakproduction) and evaluating them directly from the data (for example, jets faking leptons).The total efficiency, ǫ , typically has several components. Overall, oneneeds to evaluate: ǫ = Number of events used in the analysisNumber of events produced . (25)Some of the key ingredients to evaluating the total efficiency are the productof: • Trigger efficiency:
Modeling of the trigger in collider experiments hasbeen found to be quite complex. Generally, trigger efficiencies are ob-tained from data, measuring the efficiencies of the different componentsthat make up a particular trigger from an other trigger with looser re-quirements. These efficiencies can have a dependence on p T or η forexample and a trigger turn-on curve as a function of these variablesneed to be evaluated. • Particle identification efficiency:
The identification of the final stateparticles as described in Sec. 9 are often not highly efficient. One needsto determine how often an object that should have been identified failedthe selection criteria. The identification efficiencies are generally ob-tained from data. For example, for leptons, Z → ℓℓ decays are ideal formeasuring their efficiencies. Z boson decays provide a clean environ-ment and a precisely known mass resonance. The efficiencies are thenmeasured by selecting Z candidate events where only one of the leptonsis rigorously identified, while the other lepton has its selection criteriasignificantly loosened, and then counting how often the loose leptonfails the full selection. • Reconstruction efficiencies:
Again, experimentalists rely on the data tohelp them evaluate the efficiency of the reconstruction of tracks, thereconstruction of clusters in the calorimeters, etc. • Kinematic acceptance:
An additional ingredient to knowing the totalefficiency ǫ is also knowing the fraction of the decays which satisfy thegeometric constraints of the detector (for example η coverage) and thekinematic constraints (for example E T or p T of the final state objects)of the event selection criteria. The acceptance is primarily determinedfrom a Monte Carlo simulation of the signal process.There are a lot of examples of cross section measurements in the avail-able literature which describe in detail the complexities of the analysis, and ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc I provide a reference to rather complete one of the inclusive W and Z crosssections at the Tevatron for further reading.
13. Concluding Remarks
With the startup of the LHC, we are at the dawn of a new era of parti-cle physics. In these lectures, I was only able to touch the surface of thechallenges experimentalists face when trying to understanding the data tothe point of confidently making a discovery. With these lectures I provide astarting point for understanding the physics of how particles interact withmatter and how we exploit those interactions in the state-of-the-artCMS and ATLAS detectors to be used in analyses. It is a great time tobe working on the energy frontier as we are all looking forward to upcomingdiscoveries at the LHC.
14. Acknowledgments
I would like to thank the TASI organizers for their hospitality and for theirkind invitation to give these lectures. It was a wonderful opportunity andI hope I interact again in the near future the highly enthusiastic groupof students who attended the lectures. Their energy was refreshing and Iencourage them to continue to have lively discussions with experimentalcolleagues.
References .2. https://physicslearning2.colorado.edu/tasi/tasi 2009/tasi 2009.htm.3. http://cdsweb.cern.ch/ .4. .5. N. Kidonakis, “Higher-order corrections to top-antitop pair and single topquark production”, arXiv:hep-ph/0909.0037.6. J.M. Campbell, J.W. Huston, W.J. Stirling, Rept. Prog. Phys. , 89 (2007).Also, see arXiv:hep-ph/0611148.7. J. Pumplin, D.R. Stump, J. Huston, H.L. Lai, S. Kuhlmann, J.F. Owens andW.K. Tung, J. High Energy Phys. , 046 (2003).8. A.D. Martin, W.J. Stirling, R.S. Thorne and G. Watt, Phys. Lett. B http://durpdg.dur.ac.uk/hepdata/pdf3.html .11. C. Amsler et al. (Particle Data Group), “The Review of Particle Physics”,Physics Letters B667, 1 (2008) and 2009 partial update for the 2010 edition. ctober 23, 2018 1:5 WSPC - Proceedings Trim Size: 9in x 6in tasi09˙proc
12. W. R. Leo,
Techniques for Nuclear and Particle Physics Experiments , 2ndedn. (Springer, 1994).13. T. Affolder et al. , Nucl. Instrum. Meth.
A526 et al. , “Run II jet physics”, arXiv:hep-ex/0005012.23. M. Cacciari, G. P. Salam, G. Soyez, J. High Energy Phys. , 063 (2008).24. http://lcg.web.cern.ch/lcg/.25. “CMS Physics TDR: Volume II, Physics Performance”, CERN-LHCC-2006-021, J. Phys. G, . 995 (2007).26. http://atlas.ch/.27. The ALICE Collaboration, Eur.Phys.J. C65 , 111 (2010).28. The CMS Collaboration, J. High Energy Phys. , 041 (2010).29. The ATLAS Collaboration, arXiv:hep-ex/1003.3124.30. A. Abulencia et al. , J. Phys. G: Nucl. Part. Phys.34