LHC Detectors ATLAS and CMS Howard Gordon Brookhaven

LHC Detectors: ATLAS and CMS Howard Gordon, Brookhaven National Laboratory, Jiří Dolejší, Charles University Prague Physicists passed a long way from the table-top accelerators like the first cyclotron invented and built for about 25$ by Ernest Lawrence in 1930 towards huge accelerators for about 1 G$ hidden under the landscape like LHC at CERN. . . Replica of Lawrence’s cyclotron at CERN Microcosm 1

CERN LHC, to be finished in 2007 2

Why are physicists building such huge and expensive machines? ? ? Because there are still many unanswered questions, like: What gives particles their mass? Where is the awaited Higgs boson? Are there any extra dimensions predicted by some theorists? Do the predicted supersymmetry particles exist? A rather simple question might also be: Is the Nature fully described by the today's Standard Model, nothing beyond? The answer could be hardly yes! The new machines are huge and therefore expensive to explore the new energy regions and to enable studies of extremely rare processes. . . if something was not observable in the past, we should create the chance to observe it tomorrow. LHC will accelerate particles, but we should be able to “see” them – to have appropriate detectors. Have a look at them: 3

Here is one of them: A Toroidal LHC Apparatu. S ATLAS 22 m 44 m 4

And here the second: Compact Muon Spectrometer CMS 15 m 22 m 5

Why are the detectors at LHC so big? ? ? They should deal with all particles flying from the collision of accelerated protons. The protons are not two like on the animation, but plenty of them grouped into bunches: 2808 bunches in each beam, 1, 15× 1011 protons in each bunch, bunch spacing 25 ns what corresponds to 7. 5 m distance (some bunch positions are empty). 6

The collision point is “watched” by surrounding detector. Some particles just escaped from the collision zone, the next collision threatens. Each meeting of two bunches results in about 23 proton-proton collisions. The mean number of particles born in all these collisions is about 1500. The detector should record as many of them as possible. The detector should: • have large coverage (catch most particles) • be precise • be fast (and cheap and. . . ) Each proton carries energy 7 Te. V. So each bunch with 1011 protons carries energy 1011× 7× 1012 e. V = 7× 1023 e. V = 44 k. J. This is a macroscopic energy!!! In order to reach such kinetic energy on a bike, you go with a speed of more than 30 km/h! So boring to paint 1011 protons in each bunch. . . 7

The real detector should have no “holes” and expose to particles sufficiently thick layer of material to detect them (see the chapter Particle physics experiment for processes which happen when particles fly into matter). The collision point is “watched” by surrounding detector. Here many particles escape “detection“. The Collisison point surrounded by layers of different detectors 8

Let us have a look at interaction of different particles with the same high energy (here 300 Ge. V) in a big block of iron: 1 m electron The energetic electron radiates photons which convert to electron-positron pairs which again radiate photons which. . . This is the electromagnetic shower. The energetic muon causes mostly just the ionization. . . muon pion (or another hadron) Electrons and pions with their “children” are almost completely absorbed in the sufficiently large iron block. The strongly interacting pion collides with an iron nucleus, creates several new particles which interact again with iron nuclei, create some new particles. . . This is the hadronic shower. You can also see some muons from hadronic decays. 9

Expert pages! You don´t need to understand them, but it is a challenge! Try to answer the following questions: What about interactions of high energy photons? What about neutral pions which decay very quickly (the mean lifetime is just 8× 10 -17 s, ct = 25 nm) to two photons? To answer these questions think about the evolution of the electromagnetic cascade. . . For a little bit deeper insight to the electromagnetic and hadronic showers we may remember the exponential probability of a projectile to survive without interaction or without absorption (see the chapter “Particle physics experiment”) in the depth t of the target: where we introduced the mean interaction length t. This quantity determines the mean distance between collisions of hadrons with nuclei of the material and therefore it tells us where the hadronic shower will probably start and how fast it will evolve. The radiation length X has almost the same meaning in evolution of the electromagnetic cascade – it determines the mean path of an electron to radiate the photon and also the mean path of a photon to convert to the electron-positron pair. Look at values of these quantities for several materials: Material Radiation length X Nuclear interaction length t water 36, 1 cm 83, 6 cm iron 1, 76 cm 16, 9 cm lead 0, 56 cm 17, 1 cm 10

Here is the general strategy of a current detector to catch almost all particles: Magnetic field bends the tracks and helps to measure the momenta of particles. electron muon Hadronic calorimeter: offers a material for hadronic shower and measures the deposited energy. Neutrinos escape without detection hadrons Tracker: Not much material, finely segmented detectors measure precise positions of points on tracks. Electromagnetic calorimeter: offers a material for electromagnetic shower and measures the deposited energy. Muon detector: does not care about muon absorption and records muon tracks. 11

All the detectors are wrapped around the beam pipe and around the collision point: here a schematic and less schematic cut through ATLAS The Tracker or Inner detector The Hadronic calorimeter The Electromagnetic calorimeter The Muon detector 12

ATLAS and CMS follow the same principles but differ in realization: ATLAS Tracker or Inner Detector Silicon pixels, Silicon strips, Transition Radiation Tracker. 2 T magnetic field CMS Silicon pixels, Silicon strips. 4 T magnetic field Electromagnetic Lead plates as absorbers with Lead tungstate (Pb. WO 4) calorimeter liquid argon as the active crystals both absorb and medium respond by scintillation Hadronic calorimeter Iron absorber with plastic scintillating tiles as detectors in central region, copper and tungsten absorber with liquid argon in forward regions. Stainless steel and copper absorber with plastic scintillating tiles as detectors Muon detector Large air-core toroid magnets with muon chamber form outer part of the whole ATLAS Muons measured already in the central field, further muon chambers inserted in the magnet return yoke 13

So, why are the detectors at LHC so big? ? ? Many tempting questions Challenging theoretical predictions Towards higher energy LHC, 7+7 Te. V Curiosity to explore the unexplored ATLAS and CMS in their complexity Many very energetic particles to be recorded analysed 14

How to get the data from the detector? The detectors will sense the collisions of proton bunches every 25 ns, i. e. with the frequency of 40 MHz. With 23 pp collisions in every bunch crossing it means pp collision rate almost 1 GHz. Few GHz is the frequency of current computer processors, so how it could be possible to collect and elaborate data from such a huge detector? ? ? Destiny of ATLAS after first data taking? One should have in mind, that new beam particles come to the interaction region with a speed of light, but signals from the detector move in the cables always slower. One could therefore expect, that information from the detector will cumulate inside and sooner or later explode. Almost every student knows the feeling of the potentially exploding head from some lectures or seminars. The solution is quite “human” - to concentrate on the most interesting events and to forget about all others. This task is performed by the trigger system. The trigger planned for ATLAS has three levels and in these three steps reduces the event rate to about 100 – 200 events per second which are written to storage media. The size of data from one event is about 1 MB. 15

What to do with that amount of data? The data heap will grow fast – more than 100 MB per second, about 10 TB per day, 1 PB (1015 B) per year. You can translate this amount of data to usual media – ATLAS will need to burn a CD every 7 seconds, more than ten thousands CDs per day, more than million CDs per year. . . You may notice that our estimates are quite rough. We calculate with a year having 107 seconds instead of having p× 107 seconds. We expect that not the whole year could be used for running the experiment and recording the data. The computing power needed to analyze this huge amount of data is larger than what is available now. LHC experiments are actively participating in the development of a new computing tool to facilitate the analysis. The solution is a distributed computing and the corresponding key word is the “grid”. The word “grid” as used here is analogous to the power grid: the distributed requests for computing resources, data or computational power will be satisfied by the tiered structure of computing centers (see figure on the following page). 16

17

How these collaborations work? Where they get money? The ATLAS Collaboration includes about 1850 physicists and engineers from 175 institutes in 34 countries. CMS has a similar list of participants often from the same countries, but not completely overlapping. Each institute has specific responsibilities as formalized in a Memorandum of Understanding. Financial support comes from the funding agencies of individual participating states. 18

19

Both these experiments have a well defined democratic structure for steering all affairs. There has been a heavily documented process for each subdetector: setting the objectives Review developing the detailed technical specifications full prototyping of each component procurement and placing contracts Review installation testing fabrication Review commissioning operation 20

These collaborations have organized meetings to resolve specific design issues and to divide the work. The meetings can occur all over the world, often using telephone or video conferencing, but are mostly held at CERN. 21

Decisions and technical specifications are documented in Technical Design Reports, drawings and other documents that are available on the World Wide Web that was invented at CERN by particle physicists. The NEXT cube, the first WWW server at CERN Microcosm and Tim Berners-Lee which together with Robert Cailliau invented the World Wide Web. 22

What is happening now? Leading industrial companies from all over the world fabricate components of the detector. Many of the components are assembled in the various collaborating institutes. Final installation and commissioning of each component is done at CERN with the participation of the collaborating teams. The cryostat for liquid argon electromagnetic calorimeter. Hadronic calorimeter being assembled in the ATLAS experimental cavern. Toroid magnets of the muon system 23

To be continued 24