The ATLAS and CMS Experiments at the LHC

Abstract

The ATLAS [1] and CMS [2 experiments are the two general purpose detectors at the LHC [3]. They will measure the decay products of proton–proton collisions at up to 14 TeV. Two more detectors are installed in the LHC ring: LHCb [4] is specialised on physics with b and c quarks and ALICE [5] is dedicated to the measurement of heavy ion collisions.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

The ATLAS [1] and CMS [2 experiments are the two general purpose detectors at the LHC [3]. They will measure the decay products of proton–proton collisions at up to 14 TeV. Two more detectors are installed in the LHC ring: LHCb [4] is specialised on physics with b and c quarks and ALICE [5] is dedicated to the measurement of heavy ion collisions.

The LHCb experiment will measure CP violation and rare decays of b and c hadrons in order to find indirect evidence for new physics beyond the Standard Model. Todays measurements in heavy flavour physics from B factories and the Tevatron [6,7] are fully consistent with the CKM mechanism. Nevertheless, LHCb is probing CP violation and decays of B _d, B _s and D mesons in greater detail to possibly find new sources of CP violation or effects of new, e.g. super-symmetric, particles. At luminosities of $2 - 5\times 10^{32} \textrm{cm}^{-2}\textrm{s}^{-1}$ which LHC will deliver to the LHCb experiment, several $10^{12}$ $\textrm{b} \overline{\textrm{b}}$ pairs will be produced per year. The LHCb detector is built asymmetrically around the interaction point, since b and $\bar{\textrm{b}}$ quarks can be measured equally well in both hemispheres. LHCb physics will cover a precise measurement of B _s oscillations especially of the mixing phase $\phi_s$, the determination of $\gamma = -\arg V_{ub}$ by studying hadronic B meson decays, measurements of rare decays like $B^0\to K^{*0}\gamma$, $B_s\to \mu^+\mu^-$, $B_s\to\phi\gamma$, and more. The LHCb detector is optimised for secondary vertex location, excellent mass resolution and particle identification to further improve the precision of CP physics with quarks.

The physics program of ALICE is dedicated to the study of QCD in extreme conditions. Collisions of heavy nuclei like Pb–Pb at $\sqrt{s}=5.5$ TeV allow the measurement of strongly interacting matter at very high energy densities in the regime $\varepsilon\approx 1 - 100 \textrm{GeV} \textrm{fm}^{-3}$. At such energy densities a new state of matter, the quark-gluon plasma (QGP) [9], consisting of deconfined quarks and gluons is expected to occur [10]. This was first discovered at the CERN SPS [11] and further studied at RHIC [12]. Also in ALICE the QGP state will be formed for only an extremely short time in each heavy ion collision, so that the measurement of the details of the QGP will be performed by determining charged particle multiplicities, particle momentum spectra and (elliptic) flow, production (respectively suppression) of heavy quarkonia like $J/\psi$ and $\varUpsilon$ mesons, c- and b-quark production, as well as production of high-p _T jets and photons [13,14]. All these serve as a probe to learn more about the formation and freeze-out mechanisms of the QGP. The LHC is planned to operate in heavy-ion mode during a few weeks per year delivering about $0.5 \textrm{nb}^{-1}$ of data to ALICE each year.

The following chapters will introduce the LHC collider and concentrate in more detail on the experimental techniques of ATLAS and CMS with focus on pp collision physics.

5.1 The Large Hadron Collider

The LHC [3] is installed in the LEP tunnel and accelerates bunches of protons in a ring of 26.6 km circumference from the injection energy of 450 GeV to the nominal beam energy of 7 TeV . Figure 5.1 shows the LHC underground installation. The protons are produced in a duoplasmatron device and accelerated in a linear accelerator (Linac2) to 50 MeV at a pulsed current of 180 mA. From there, the beam is injected into four Proton Synchrotron Booster (PSB) rings, ramped to 1.4 GeV, and transferred to the Proton Synchrotron (PS) where the beam energy reaches 25 GeV. In the PSB and PS the LHC bunch structure is prepared. The base structure is a sequence of 72 bunches with 25 ns spacing. Each proton bunch contains $8.28\times 10^{12}$ protons, when the LHC is operated at nominal luminosity. The rise-time of the beam ejection kicker magnet creates a gap of $(320\approx 13\times 25) \textrm{ns}$ after each 72 bunches.

Fig 5.1

The LHC underground installation. The ATLAS and CMS experiments are installed at opposite sites of the main ring at access points 1 and 5. ALICE and LHCb are close to the ATLAS site at point 2 and 8, respectively. The protons are injected from the SPS into the LHC via beam transfer lines. Eventually, the protons are stopped at the end of the beam lifetime into a beam dump system at point 6

The last element is the Super Proton Synchrotron (SPS) where the beam energy is increased from 25 to 450 GeV, the LHC injection energy. Additional time gaps are introduced into the final bunch structure by the rise-time of the SPS and LHC injection kickers, which is shown in Fig. 5.2. In total, 2808 bunches each containing $1.15\times 10^{11}$ protons are circulating in both directions of the LHC ring, where they are accelerated to their final energy of 7 TeV. The beam current is 0.58 A and the expected filling time is about 3 min. At the highest intensities, an energy of 362 MJ is stored in each beam.

Fig 5.2

Time structure of the LHC proton bunches. In total, 2,808 bunches are injected per proton beam [3]

The LHC machine is divided into eight 3 km long arc sections and eight straight sections, each 523 m long. One arc is composed of 23 identical FODO cells, with the typical focusing and defocusing magnetic multipole structure. Each cell is 107 m long with 3 dipoles and 1 quadrupole per half-cell. The superconducting dipoles provide the magnetic bending field which varies from 0.54 T at beam injection to 8.35 T when the protons reach 7 TeV. Inside the dipoles, the two proton beams circulate in two beam pipes separated by 197 mm, as displayed in Fig. 5.3. The straight sections are equipped with dispersion suppressors to match and adapt the beam optics in the straight sections, also called insertion regions (IR), to the arc. The beams are injected in IR8 and the RF structures are installed in IR4. They accelerate the protons using a 400 MHz superconducting cavity system. The frequency matches the bunch length, which varies between 1.7 and 1.1 ns at 450 GeV and 7 TeV, respectively, to capture the beam with minimal losses. Eight cavities with 5.5 MV/m field strength provide in total 16 MV accelerating voltage per beam. A photograph of one cavity is shown in Fig. 5.3.

Fig 5.3

(a) Cross-section through a LHC dipole magnet inside its vacuum vessel with the two beam pipes, the cold screen and the superconducting coils [3]. (b) One of the four superconducting cavities that are combined in each of the two RF modules which supply the 400 MHz RF power to the proton beam

In two other insertions the beams are cleaned: particles with large momentum offset are absorbed by collimators in IR3, while particles with large horizontal, vertical, or combined betatron amplitudes are filtered out in IR7. The beam abort system with the beam dump is installed in IR6.

To focus the beam at the interaction points (IP), a triplet of 31 m long quadrupole magnets is installed on each side of the IP. Beam separation and recombination is performed by two pairs of dipoles separated by 88 m. A set of four more quadrupoles provides the matching of the beam optics in the IP region to the remaining ring.

There are 1,232 main dipole magnets installed, together with several hundred quadrupoles for focusing and defocusing of the beam, completed by many thousand sextupoles, octupoles and decapoles for orbit correction. Each multipole creates betatron oscillations in the vertical and horizontal plane. In the LHC, the number of oscillations per turn, also called horizontal and vertical tune, are carefully chosen to be $Q_h=64.31$ and $Q_v=59.32$, respectively. This is to avoid the resonance condition

$$m Q_h + n Q_v=p\;\;(m,n,p=\textrm{integer numbers})$$

((5.1))

which leads to beam instabilities and eventually beam loss.

The machine luminosity depends on the beam parameters and can be written for a Gaussian beam distribution as

$$L=\frac{N_b^2 n_b f_{rev} \gamma_r}{4\pi \varepsilon_n \beta^*} F\;,$$

((5.2))

where N _b is the number of protons per bunch, n _b the number of bunches per beam, $f_{rev}$ the revolution frequency, $\gamma_r$ the relativistic gamma factor, $\varepsilon_n$ the normalised transverse beam emittance, $\beta^*$ the beta function at the IP. The factor F is the geometrical luminosity factor due to the beam crossing angle $\theta_c$:

$$F=1/\sqrt{1+\left(\frac{\theta_c \sigma_z}{2\sigma^*}\right)^2}\;,$$

((5.3))

where $\sigma_z$ is the RMS bunch length and $\sigma^*$ the transverse RMS beam size at the IP. The normalised transverse emittance is related to $\sigma^*$ by

$$\varepsilon_n=\gamma_r\varepsilon=\gamma_r\frac{(\sigma^*)^2}{\beta^*}\;,$$

((5.4))

so that the luminosity may also be written in a more classical way

$$L=\frac{N_b^2 n_b f_{rev}}{4\pi (\sigma^*)^2} F,$$

((5.5))

inversely proportional to the transverse beam area.

Table 5.1 summarises the parameters which need to be reached to achieve the peak luminosity of $1.0\times 10^{34} \textrm{cm}^{-2}\textrm{s}^{-1}$ at 14 TeV centre-of-mass energy. With a luminosity lifetime of 15.5 h, an integrated luminosity of about $60-80 \textrm{fb}^{-1}$ per year can be expected for the final LHC performance.

Table 5.1. LHC beam parameters for 7 TeV beams at peak luminosity at the interaction points of the ATLAS and CMS experiments [3]

There are several effects that limit the LHC luminosity. The mechanical aperture of the beam pipe is about $34.6 \textrm{mm}\times 44 \textrm{mm}$, and the transverse beam size is 1.2 mm, applying a 10$\sigma$ safety distance of the beam profile to the wall. With a maximum value of the $\beta$ function of 180 m, the normalised transverse emittance is $3.75 \upmu\textrm{m}$. The minimal $\beta^*$ at the IP and maximum crossing angle is limited by the aperture of the quadrupole triplets.

In the interaction region, beam–beam interaction induces a so-called tune shift of:

$$\xi = \frac{N_{b} r_p}{4\pi \varepsilon_n}$$

((5.6))

where r _p is the classical proton radius $r_p={\alpha_\textrm{QED}}/m_p$. From experience with proton-proton machines, the total linear tune shift must stay below 0.015 for stable running. For 3 IP’s, this corresponds to $\xi <0.005$, or, with $\varepsilon_n=3.75 \upmu\textrm{m}$, to a maximum number of protons per bunch of $N_b=1.5\times 10^{11}$. This is close to the nominal value of $N_b=1.15\times 10^{11}$.

One important factor is the electron cloud effect. It is induced by synchrotron radiation which creates a 6.7 keV energy loss of the 7 TeV proton beam per turn. The UV photons hit electrons off the beam pipe wall. These electrons are accelerated in the field of the proton beam and can initiate secondary electron emission, which eventually builds up an electron cloud. This cloud leads to beam instabilities, growth of the emittance and increases the heat load in the cold beam screen. This screen is kept a temperature 5–20 K and shields the 1.9 K cold core of the superconducting magnets from quenches. The cooling capacity of the screen is 1.15 W/m, where the average arc heat load is already 0.66 W/m, e.g., due to synchrotron radiation and resistance of the wall. The remaining 0.5 W/m are attributed to the electron cloud heat load. The heat load is increasing also with shorter bunch spacing. For a future LHC upgrade, higher luminosity can therefore not be reached by simply increasing the number of bunches. In nominal LHC running, the electron cloud is suppressed by reducing the photon reflectivity in the arcs, and by special getter material, TiZrV, in certain sections to reduce secondary emissions. Also, the conditioning of the arc chambers by so-called “beam scrubbing”, i.e. cleaning of the walls using the electron cloud effect itself, helps to reduce the disturbances due to scattered electrons during LHC operation.

At the start-up in 2009, LHC will not yet be operated at the full energy, but only at a centre-of-mass energy of initially 7 TeV, to be further increased to 10 TeV. The dipoles are expected to be commissioned to the full energy in 2011/2012. Initially, not all bunches will be filled and a 43-on-43 or a 156-on-156 bunch scenario is foreseen. The $\beta^*$ in the pilot run will be in the range 3 – 11 m, the number of protons per bunch may reach $5\times 10^{10}$, and no beam crossing-angle is foreseen. Instantaneous luminosities in the order of $10^{31} - 10^{32} \textrm{cm}^{-2}\textrm{s}^{-1}$ can be expected in the first run period, and during 100 days of running an integrated luminosity of $100 \mathrm{pb^{-1}}$ may be collected, followed by a period of another 100 days with twice the luminosity. In the following three years, a low luminosity period with $L=10^{33} \textrm{cm}^{-2}\textrm{s}^{-1}$ is foreseen, corresponding to $1.0 - 2.5 \textrm{fb}^{-1}$ of data in 2011/2012 assuming 150 days of physics running. Initially, there will be 936 bunches per beam with with 75 ns spacing and a $250 \upmu\textrm{rad}$ crossing angle. Eventually, the nominal number of bunches is increased to 2,808 with 25 ns spacing and nominal $285 \upmu\textrm{rad}$ crossing-angle. The number of protons per bunch will stay in the order of $5\times 10^{10}$.

After the full implementation of the collimators and the final completion beam dump system in 2013/2014, the beam intensities can be pushed to $9\times 10^{10}$ protons per bunch and the $\beta^*$ will be squeezed to 0.55 m. With these parameters, the design luminosity of $L=10^{34} \textrm{cm}^{-2}\textrm{s}^{-1}$ is expected to be reached and the LHC will be operated in this mode until 2017/2018 with up to $80 - 100 \textrm{fb}^{-1}$ of data per year.

From the above discussion one can see that possible ways to further upgrade the LHC machine to higher luminosity are, for example, a modified triplet magnet with larger aperture and smaller $\beta^*=0.25 \upmu\textrm{m}$ and/or longer proton bunches with larger bunch spacing. This luminosity upgrade, with a possible increase of the beam energy, is foreseen for a shutdown period in 2018/2019. In the most complete scenario, this requires a replacement of the complete injection chain to attain higher proton energies and better beam brilliance $N_b/\varepsilon^*$ at the interaction points.

The actual proton–proton interaction rate seen by the CMS and ATLAS detectors in the interaction regions is proportional to the luminosity and given by

$$\frac{dN}{dt}=L\sigma_{pp}\;,$$

((5.7))

where the main contribution to the proton–proton cross-section, $\sigma_{pp}$, is from inelastic scattering, $\sigma_{inel}$. It amounts to about 80 mb, or $80\times 10^{-28} \textrm{cm}^2$. This means at the peak luminosity of $10^{34} \textrm{cm}^{-2}\textrm{s}^{-1}$, the number of events per bunch crossing, N _c, is about

$$N_c=\frac{L}{n_b}\frac{\sigma_{inel}}{f_{rev}}=25.6\;.$$

((5.8))

On top of each hard pp scattering event, for example the production of a Z boson in a Drell-Yann process, 25 inelastic pp events are overlaid. This phenomenon of in-time pile-up , the high collision frequency and very good resolution for optimal signal identification and background rejection drives the performance requirements of the ATLAS and CMS detectors.

5.2 The ATLAS and CMS Experiments

The layout of the ATLAS and CMS experiments follows the well-known concepts of particle collider detectors with tracking systems for charged particles close to the interaction point, electromagnetic and hadronic calorimeters at larger distances, completed by muon detectors in the outermost layer. A schematic view of ATLAS and CMS is shown in Figs. 5.4 and 5.5. Both detectors provide maximal hermeticity in their angular coverage. The general design is furthermore motivated by the challenging physics tasks of experiments at the LHC. The proton–proton beam crossing rate is at 40 MHz, each crossing with up to 25 inelastic pp collisions. This results in a high density of charged particle tracks, calorimetric energy deposits and signals in the muon detectors. For example, the flux of charged particles at a radius $r=10 \textrm{cm}$ around the beam axis is about $10 \textrm{MHz}/\textrm{cm}^2$. Apart from providing a good energy and momentum resolution for leptons, jets and global energy flow, the detection systems must in addition be of high granularity and must sustain high radiation levels. They should also have a fast signal response time and be equipped with a fast electronic readout to filter out the interesting physics processes using a multi-layer trigger system.

Fig 5.4

Cut-away view of the ATLAS detector. The dimensions of the detector are 25 m in height and 44 m in length

Fig 5.5

Schematic view of the CMS detector showing the compact design of the different particle detection systems

5.2.1 The ATLAS Detector and Performance

The ATLAS detector [1,8] is located at IP 1 of the LHC ring. Closest to the collision point is the inner detector for tracking of charged particles, which is installed in a cylindrical superconducting solenoid of 5.5 m length and 1.15 m radius. The particles are bent in a 2 T magnetic field and detected by three separate tracking devices.

The silicon pixel detector is composed of three barrel layers at radii between 50.5 and 122.5 mm and three endcap disks at 495–650 mm distance to the nominal interaction vertex. The size of the 47,232 pixels is $50\times 400 \upmu\textrm{m}^2$ and the intrinsic spacial resolution is $10 \upmu\textrm{m}$ and $115 \upmu\textrm{m}$ in $r-\phi$ and z/r direction, respectively. The subsequent tracking is done by the silicon tracker (SCT) which has 4 cylindrical silicon microstrip layers in the barrel section and $2\times 9$ microstrip disks in the endcaps. The outermost silicon layer is at 514 mm radius and the most distant disk at $|z|=2,727 \textrm{mm}$. With a strip width of $80 \upmu\textrm{m}$ in the barrel and $57-94 \upmu\textrm{m}$ in the endcap, a spacial resolution of 17 and $580 \upmu\textrm{m}$ in $r-\phi$ and z/r direction can be achieved. The combined angular coverage of the pixel and SCT reaches to $|\eta|<2.5$. Further details of the geometrical layout are shown in Fig. 5.6

Fig 5.6

Geometrical layout in the $r-z$ plane of one quarter of the ATLAS inner detector with the pixel, SCT and TRT sub-detectors [1]

The inner detector is completed by a transition radiation tracer (TRT) which uses straw tubes filled with a $\textrm{Xe}/\textrm{CO}_2/\textrm{O}_2$ gas mixture as active medium. The tubes have a 4 mm diameter and a length of 144 cm in the barrel and 37 cm in the endcap. They provide additional 20–35 space points along the particle tracks in the fiducial volume $|\eta| <2.0$. The combined momentum resolution for single charged tracks in the inner detector, using pixel, SCT and TRT, is

$$\frac{\sigma_{p_T}}{p_T}=\sqrt{\left(5\cdot 10^{-4} p_T\right)^2+0.01^2}$$

((5.9))

in the fiducial volume $|\eta| <2.5$.

The TRT is furthermore used to separate electrons and pions. Photon radiation is produced at the transition from the straw tube plastic and the gas, an effect that depends on the relativistic $\gamma$ of the particle, which, for the same energy, is about 280 times larger for electrons than for pions. The photons created by traversing electrons are in the X-ray energy range and produce additional high threshold signals in the straw tube detectors. The high threshold probability is used as a discriminant against pions. For particles of 25 GeV, a $\pi^\pm$ rejection factor between 10 and 100 at an electron efficiency of 90% is obtained.

A serious challenge for the measurement of electromagnetic particles is the number of interaction lengths in the silicon tracker in front of the calorimeter. About 20–60% of the high-energy photons undergo conversion into an $\textrm{e}^+\textrm{e}^-$ pair, while electrons loose between 30 and 60% of their energy due to bremsstrahlung. This is compensated by taking the two effects into account during reconstruction. Figure 5.7 gives examples for the track reconstruction efficiencies for electrons and pions which are affected by bremsstrahlung as compared to muons which are minimum ionising. Also shown is the high 80% efficiency to reconstruct photon conversions. The precision of the position of the conversion vertex is about 5–7 mm in radial direction.

Fig 5.7

(a) Track reconstruction efficiency in the inner detector for 5 GeV electrons, pions and muons [1]. The lower tracking efficiency for electrons and pions is due to bremsstrahlung effects when traversing the material of the inner detector and the beam pipe. It is increasing with $\eta$. Minimum ionising muons are less affected. (b) Efficiency to reconstruct photon conversions of photons with $p_T =20 \textrm{GeV}$ in the central $|\eta| <2.1$ detector region, as a function of the conversion radius [1]. The overall efficiency is a combination of double and single track conversions

Vertexing precision is as well important in the identification of primary and secondary vertices. The primary vertex is reconstructed in $H\to \gamma\gamma$ events with 96% efficiency and the correct vertex is selected with 79% probability, both evaluated for a luminosity of $L=10^{33} \textrm{cm}^{-2}\textrm{s}^{-1}$. Vertex reconstruction efficiency and correct vertex assignment are both close to 100% for $t\bar{t}$ events where additional tracks are produced in the hard scattering process. The primary vertex resolution is about $20 - 35 \upmu\textrm{m}$ in the plane transverse to the beam and $40-70 \upmu\textrm{m}$ in longitudinal direction. The secondary vertex reconstruction efficiency depends even stronger on the event topology. For $t\bar{t}$ events it is typically higher than 60%.

The energy measurement of electrons and photons is performed in the electromagnetic calorimeter that consists of a Liquid Argon barrel and endcap calorimeter in the pseudo-rapidity ranges $|\eta| <1.475$ and $1.375<|\eta| <3.2$, respectively. The calorimeter is a LAr-lead sampling calorimeter with accordion-shaped kapton electrodes and lead absorbers. The accordion structure is chosen in order to achieve a homogeneous energy response in $\phi$ without detection gaps. The ATLAS calorimetric devices, including the LAr calorimeters, are shown in Fig. 5.8.

Fig 5.8

The ATLAS Liquid Argon calorimeters consist of the electromagnetic barrel and endcap calorimeters, a hadronic endcap and a forward calorimeter. They are installed in metal cryostats onto which the front-end electronics is mounted. The ATLAS calorimetry is completed by a hadronic Tile calorimeter in the barrel and forward region

The two halves of the LAr barrel are installed in the same cryostat as the solenoid magnet and are separated by a small 4 mm gap. The barrel electrodes are segmented in radial direction in three compartments each having different signal cell sizes. The innermost layer is composed of strips covering $\Delta\eta\times\Delta\phi$ segments of $0.025/8\times 0.1$ in the centre, $|\eta| <1.40$, becoming more course, $0.025\times 0.025$, in the $1.40<|\eta| <1.475$ region. The second layer covers $0.025\times 0.025$ and $0.075\times 0.025$ segments in the two angular regions, while the third layer has a granularity of $0.050\times 0.025$. The strip layer is meant to sample the beginning of the electromagnetic shower with high resolution to be able to resolve adjacent showers, for example from $\pi^0\to \gamma\gamma$ decays, converted photons or bremsstrahlung photons close to electron clusters. The middle layer is generally containing the peak of the electromagnetic energy deposition along a photon or electron shower, while the back layer is measuring the shower tail. An example of the LAr cells in the central barrel section is given in Fig. 5.9.

Fig 5.9

Segmentation in $\Delta\eta\times\Delta\phi$ and radial depth of the LAr calorimeter cells in the barrel. Also indicated is the size of the calorimeter trigger towers which combine larger angular areas [1]

The $\Delta\eta\times\Delta\phi$ segmentation of the endcap calorimeter is similar to the barrel, with cell sizes between $0.025/8\times 0.1$ to $0.1\times 0.1$ in the first layer, between $0.025\times 0.025$ to $0.1\times 0.1$ in the second layer, both varying with $\eta$, and eventually $0.050\times 0.050$ in the back layer. The LAr calorimeter has in total about 180 thousand cells that are read out.

The barrel calorimeter has a depth of at least 22 radiation lengths, X ₀, increasing to maximal 36 X ₀, while the endcap provides between 24 X ₀ and 38 X ₀ of absorbing material. Since showering of electromagnetic particles usually starts already in the upstream material of the inner detector, a presampler is installed in front of the barrel calorimeter and in parts of the endcap. The energy deposition of photons and electrons in the 11 mm LAr gap in the barrel and the two 2 mm gaps in the endcap are proportional to their upstream energy loss. In this way the complete longitudinal shower development can be reconstructed.

Using the energy measurements in the presampler, E _PS, in the strip, middle and back compartments, E _strip, E _middle, E _back, the electromagnetic cluster energy can be reconstructed as a weighted sum:

$$E=s(\eta)\left(c(\eta)+w_0(\eta)E_{PS}+E_{strips}+E_{middle}+w_3(\eta)E_{back}\right)\;.$$

((5.10))

The overall scale factor s, the offset, c, and the weights, w ₀ and w ₃, are determined in the calibration procedure and are $\eta$ dependent. The weight w ₀ applied to the presampler measurement is in the order of 20 and 60 for the barrel and endcap, respectively. This compensates the energy loss in front, while the weight w ₃ takes the leakage of the shower at the back of the calorimeter into account. The obtained energy resolution for electrons and photons is shown in Fig. 5.10. It follows a parameterisation

$$\left(\frac{\sigma}{E}\right)^2= \left(\frac{S}{\sqrt{E}}\right)^2+ C^2\;,$$

((5.11))

with a stochastic term S, and a constant term C. In the most central $\eta$ region the values are $S=10.0\%$, and $C=0.7\%$. The latter is mainly determined by the long-range uniformity of the calorimeter and by the calibration at cell level. It also assumes perfect knowledge of the material distribution in front of the calorimeter.

Fig 5.10

(a) Expected energy resolution for electrons in different $\eta$ ranges as a function of energy [1]. (b) Relative energy resolution for 100 GeV electrons as a function of $\eta$. The resolution is rather uniform except in the transition region between barrel and endcap [1]

The calorimetry of ATLAS is completed by a hadronic tile calorimeter in the barrel and extended barrel regions at $|\eta| <1.0$ and $0.8<|\eta| <1.7$, respectively. The calorimeter is built of steel as absorber and scintillating tiles are the active material. The inner and outer radius are 2.28 and 4.25 m. At $\eta=0$, the detector thickness corresponds to 7.4 hadronic interaction lengths, $\lambda$. The tile calorimeter is radially segmented in four layers and the cell sizes in $\eta$ and $\phi$ direction are $0.1\times 0.1$. The combined LAr and tile calorimeter performance for pions was determined in test beam measurements. For a calorimeter slice that corresponds to the region at $\eta=0.25$ a pion energy resolution with a stochastic term of 52% and a constant term of 3% was found, following equation 5.11.

In the forward region, a LAr hadronic endcap calorimeter (HEC) at $1.5<|\eta| <3.2$ is built of 25 and 50 mm thick copper plates interleaved with LAr gaps. The granularity of the calorimeter is $0.1\times 0.1$ for $|\eta| <2.5$ and twice as large in the remaining $\eta$ range. The LAr forward calorimeter (FCAL) further extends the angular coverage to $3.1<|\eta| <4.9$. It consists of three modules with in total $10 \lambda$ depth. The one closest to the IP is intended for electromagnetic measurements and made of copper, while the other two are made of tungsten and measure hadronic particles. The FCAL energy resolution for electrons follows the functional form of Eq. 5.11 with $S=(28.5\pm 1.0)\%$ and $C=(3.5\pm 0.1)\%$, which is obtained in testbeam measurements. The pion energy resolution is about a factor 2–3 worse, depending on the sophistication of the reconstruction algorithm. The stochastic term of the global jet energy resolution is therefore expected to be 50% for $|\eta| <3.2$ and 100% for $3.1<|\eta| <4.9$, while the constant energy resolution term is 3 and 10%, respectively.

Jets are reconstructed from calorimeter towers with $\Delta\eta\times\Delta\phi=0.1\times 0.1$ cell size or from so-call topological clusters. The latter are reconstructed from combined high resolution calorimeter cells and noise is already subtracted. The basic jet clustering algorithms are the seeded iterative cone [15] and the k _T algorithm [16]. The former combines the input objects in angular cones of a fixed sizes $\Delta R=\sqrt{\Delta\eta^2+\Delta\phi^2}$, with $\Delta R=0.4$ and 0.7 for narrow and wide jets with a seed energy of $E_T=1 \textrm{GeV}$. It is the most widely used algorithm in ATLAS, but it is neither infrared nor collinear safe and may lead to inconsistencies with fixed order QCD calculations. More performing algorithms like SISCone [17] or anti-k _T [18], which theoretically preferred since they are infrared and collinear safe, are under study. Details of the standard cone jet resolution are shown in Fig. 5.11. The stochastic term is in the order of 60% and the constant term around 3%. The noise contribution is 0.5 GeV in the barrel and increases to 1.5 GeV in the endcap. The reconstruction efficiency approaches 100% at jet p _T values of more than 40 GeV. The jet direction is determined to better than $\Delta R=0.2$ for jets with $p_T>100 \textrm{GeV}$.

Fig 5.11

(a) Expected jet energy resolution for seeded iterative cone jets with different cone size and energy range [1]. (b) Resolution of the reconstructed transverse missing energy as a function of the sum of the transverse energy in the event [1]. Different event samples are used to evaluate the resolution at various transverse energy scales

Also important is the reconstruction of the missing transverse energy, ${E_T^{miss}}$, which provides signatures for particles that escape detection, like neutrinos or non-interacting super-symmetric particles. It is calculated as the E _T sum of the calorimetric towers and also reconstructed muons are taken into account. The corresponding resolution is shown in Fig. 5.11 as a function of $\sum E_T$ using different simulated physics processes. It can be parameterised at high $\sum E_T$ by $\sigma_{{E_T^{miss}}}\approx 0.57\cdot\sqrt{\sum E_T}$. The direction of the missing E _T vector in the $x-y$ plane is determined with a precision better than $\Delta \phi=0.8$ at low ${E_T^{miss}}$ reducing to below 0.1 at ${E_T^{miss}}>150 \textrm{GeV}$.

The outermost detector layer of ATLAS is composed of four different muon detection systems : monitored drift tubes (MDT) in barrel and endcap for the precise measurement of the muon momenta, thin-gap chambers (TGC) in the endcap for triggering, cathode-strip chambers (CSC) for the innermost endcap region, and eventually resistive plate chambers (RPC) for triggering and momentum determination in the barrel. The general geometrical arrangement of the muon chambers can be seen in Fig. 5.4 and a side view in the $r-z$ plane is shown in Fig. 5.12. The angular coverage of the MDT is $|\eta| <2.7$ and there are more than 1,000 chambers with 340 thousand channels installed. The innermost MDT layer only reaches to $|\eta| <2.0$. CSC chambers cover this endcap region $2.0<|\eta| <2.7$ with high neutron background and high particle rate. The more than 500 RPC trigger chambers are installed at the middle and outer MDL layers in the fiducial region $|\eta| <1.05$. Triggering is extended to $1.05<|\eta| <2.7$ in the forward region by 3,588 TGC chambers.

Fig 5.12

Schematic view in the $r-z$ plane of one quarter of the ATLAS muon detectors [1], with MDT chambers in barrel and endcap inner, middle and outer layer (BIL, BML, BOL, EIL, EML, EOL), the CSC and TGC detectors in the endcap region, as well as the RPC trigger layers attached to the barrel middle and outer MDTs

The magnetic bending field is provided by three air-coil superconducting toroid magnets. Eight barrel toroid coils create a field of 0.15–2.5 T, while the endcap field is between 0.2 and 3.5 T, depending on the azimuthal and polar angle. The endcap toroid also has an eight-fold symmetry and is rotated by $\pi/8$ with respect to the barrel coils. The analysing power is better quantified in terms of the field integral along the possible muon trajectory, which is displayed in Fig. 5.13. It shows that the field integral is mostly between 2 and 7 Tm, except in the barrel-endcap transition region where the momentum reconstruction power is much reduced.

Fig 5.13

(a) Integral of the magnetic field seen by muons along their trajectory [1]. To visualise the effect of the eight-fold symmetry of the toroid system the integral is evaluated at $\phi=0$ and $\phi=\pi/8$. (b) Momentum resolution for $p_T=100 \textrm{GeV}$ muons in different $\eta$ regions [1]. The stand-alone resolution is mostly better than 4%, exccept in the barrel-endcap transition region and in the very forward $\eta$ range. If the muon chamber reconstruction is combined with inner detector tracks the resolution is generally improved, in particular in the transition region

The resolution for high momentum muons is shown in Fig. 5.13. At very high momentum of $p_T=1 \textrm{TeV}$ the relative resolution is expected to be on the order of 10%.

For many of the projected physics analyses, in particular for measurements of cross-sections, the knowledge of the absolute, and eventually integrated LHC beam luminosity is needed. The main luminosity detector of ATLAS is the Cerenkov integrating detector LUCID [19,1] which is installed at 17 m distance to the ATLAS interaction point. It will measure the number of elastic pp interactions at every bunch crossing, which is proportional to the LHC luminosity. An absolute calibration of the luminosity is provided by the ALFA [19,20] system which consists of scintillating-fibre trackers located inside Roman Pots at a distance of 240 m from the interaction point.

The measurements with the ALFA detector require dedicated runs with special beam optics and low luminosity, $L= 10^{27} \textrm{cm}^{-2}\textrm{s}^{-1}$. The LUCID detector will be operated in parallel so that the two will be inter-calibrated. LUCID is then used to extrapolate this measurement up to the design luminosity with a final expected accuracy of 2–3%.

5.2.2 The ATLAS Trigger and Data Acquisition System

Important for physics measurements is the trigger system . The pp bunch crossing rate is at 40 MHz and the inelastic cross-section about 80 mb. That means that at high luminosities of $10^{34} \textrm{cm}^{-2}\textrm{s}^{-1}$ there are about 25 interactions per bunch crossing. The interesting physics processes have however much lower cross-sections in the picobarn to below femtobarn range. In general, high transverse momentum electrons, photons, muons, taus, and jets, as well as large missing transverse momentum are typical signatures of the physically interesting hard scattering processes.

The ATLAS trigger system is divided into three layers, where the first one (L1) is implemented in custom hardware and the second and third level triggers, L2 and Event Filter (EF), are based on software algorithms running on large PC farms. The software triggers are also called high-level triggers (HLT).

The first level trigger has two main inputs: the calorimeter trigger and the muon trigger. The corresponding trigger signals are read out with an electronics chain parallel to the standard readout. The data from the detector front-end is already concentrated so that a fast trigger decision with a maximum trigger latency of $2.5 \upmu\textrm{s}$ is provided. After this time interval the readout buffers on the detectors must have received a L1 signal to transfer the detector data to the Data Acquisition system (DAQ). The front-end buffer sizes are chosen according to this timing structure.

The calorimeter trigger receives information from about 7,000 analogue trigger towers with mostly $\Delta\eta\times\Delta\phi=0.1\times 0.1$ granularity. After digitisation, a Cluster Processor (CP) and a Jet/Energy-sum Processor (JEP) identify physics signatures. The CP looks for electron, photon and $\tau$ candidates above programmable p _T thresholds with possible isolation from other detector activities, while the JEP forms jets, and calculates the quantities $\sum E_T$ and ${E_T^{miss}}$, which also have to fulfil energy and multiplicity thresholds.

The muon trigger system in the barrel region is identifying low p _T muons from hits in both innermost RPC layers that are mounted inside and outside of the middle MDT stations. The p _T measurement is performed using pre-defined tracking roads whose width selects different p _T values: more narrow roads correspond to higher p _T thresholds. For high p _T tracks in the innermost RPC layers, a third measurement in the outer RPC layers is used for refining the threshold using a similar algorithm based on tracking roads. In the endcaps, the TGC chambers provide the trigger input. They stand higher rates than RPCs and operate with 99% efficiency up to rates of $20 \textrm{kHz}/\textrm{cm}^2$. Coincidence hits in the outer TGC chambers are treated independently in r and $\phi$ direction. They are finally merged and combined with the innermost TGC chamber hits. Six geometrical windows for the hits along the muon track correspond to pre-defined p _T thresholds, similarly to the barrel algorithm.

The L1 trigger decisions are taken by the central trigger processor at a rate of 75 kHz. On a L1-accept signal the detector data are transfered via optical fibres from the front-end through the Read-Out Drivers (ROD) to the DAQ system and the subsequent L2 algorithms are started. Each L1 object defines a so-called region of interest (ROI) which is an angular cone around the $\eta$ and $\phi$ direction of the identified particle or ${E_T^{miss}}$ candidate. The L2 algorithms are seeded from the ROI and only accesses data within this region, which is about 1–2% of the full event information. The L2 algorithms search for more refined physics signatures, so-called trigger chains. These chains correspond to step-wise hypothesis tests to verify if the detected signature corresponds to the programmable L2 selection criteria. In each step, the hypotheses become more restrictive. Trigger signatures are rejected as early as possible, whenever a hypothesis step is not fulfilled any more. Typical trigger signatures are

• electrons reconstructed as electromagnetic calorimetric clusters matched to inner detector tracks,

• photons identified as electromagnetic calorimeter clusters without matched track,

• tau leptons with hadronic decay signatures in the calorimeters and matched tracks,

• cone jets formed from calorimeter towers, including b-tag information,

• muons reconstructed from combined tracks in the muon chamber and the inner detectors,

• transverse calorimetric energy sums, and

• missing transverse energy.

For all trigger objects, energy and p _T threshold can be freely defined as well as isolation from nearby hadronic activity, which is important for background rejection especially at high luminosities. The L2 trigger decision must be taken within 40 ms.

The last trigger level, the event filter (EF), has access to the complete event data. These data are provided at a maximum frequency of 3.5 kHz. The EF is able to perform event selections very similar to offline data analysis so that more complicated signatures can be implemented. The EF algorithms are tuned such that the final event output rate is at most 100 Hz. The events are grouped in different categories or data streams. Typical event rates at luminosities of $10^{33} \textrm{cm}^{-2}\textrm{s}^{-1}$ in the different streams are

• electron stream: $31\pm 8 \textrm{Hz}$

• muon stream: $34\pm 9 \textrm{Hz}$

• jet stream: $38\pm 6 \textrm{Hz}$

• ${E_T^{miss}}$ and $\tau$ stream: $32\pm 8 \textrm{Hz}$

• B-physics stream: $10\pm 6 \textrm{Hz}$

In this way, the input to the offline data analyses can be reduced to the selected trigger streams so that total data processing time can be optimised for the various physics analyses.

5.2.3 The CMS Detector and Performance

The CMS detector [2] has a length of 21.6 m and a diameter of 14.6 m and is installed at IP 5 of the LHC ring. The central tracker, as well as the electromagnetic and the hadronic calorimeters are inside a solenoidal magnetic field of 4 T in which the charged particle tracks are bent. The solenoid is made of a superconducting coil of 5 m radius and 13 m length. The iron return yoke on the outside of the solenoid is equipped with muon chambers. CMS has thus a much more compact design than the ATLAS detector.

The CMS inner tracking system is composed of a silicon pixel and microstrip detectors. The central barrel region is covered by 3 pixel layers and 9 microstrip layers. The barrel region is completed with 2 pixel disks and 3 microstrip disks oriented perpendicular to the beam axes. The tracker endcap consists of 9 microstrip disk layers, which are installed parallel to the barrel disks. In the central barrel region, the innermost strip layers and two strip layers of the outer barrel are made of two-layer modules with a stereo angle of $100 \upmu\textrm{rad}$ to provide measurement in the $r-\phi$ and $r-z$ planes. There are in total 66 million pixels and 9.6 million silicon strips, which cover an angular region up to $|\eta| <2.4$. The occupancy in the $100\times 150 \upmu\textrm{m}^2$ large pixels is $10^{-4}$ per beam crossing, while in the inner $10 \textrm{cm}\times 80 \upmu\textrm{m}^2$ large strips the occupancy is 2–3%. The single-point resolution in the inner stereo layers is $23-34 \upmu\textrm{m}$ in $r-\phi$ and $230 \upmu\textrm{m}$ in z direction. In the outer layers it is about a factor 2 larger. The expected track resolution obtained for single muon events is shown in Fig. 5.14. For low momentum tracks below 10 GeV, the p _T resolution will be better than 0.7% in the central region and decreases to 2% for $\eta=2.4$. Higher energy tracks of 100 GeV can be measured with 1.5–6% resolution. Another performance parameter, important e.g., for the lifetime tag of B meson decays, is the resolution on the impact parameter distance to the vertex. It is better than $200 \upmu\textrm{m}$ for low energy tracks and reaches $10 \upmu\textrm{m}$ for more straight tracks above 100 GeV, also shown in shown in Fig. 5.14. Some performance parameters of the ATLAS and CMS tracking systems [21] are compared in Table 5.2. The CMS tracker achieves a better momentum resolution as well as better impact parameter (IP) resolutions at high p _T due to the higher magnetic field and the smaller pixel sizes, respectively.

Fig 5.14

Expected momentum and transverse impact parameter resolution of single muon tracks of three different energies, simulated for the CMS tracking system in different $\eta$ regions [23]

Table 5.2 Comparison of performance parameters of the inner tracker systems of the ATLAS and CMS experiments [21]

The CMS electromagnetic calorimeter (ECAL) is composed of a barrel and two endcaps with 61,200 and $2\times 7,\!324$ lead tungstate ($\textrm{PbWO}_{4}$) scintillating crystals, respectively. The crystals have a short radiation length, $X_0=8.9 \textrm{mm}$, fast light response and are radiation hard. The barrel section has an inner radius of 129 cm and covers an $\eta$ range up to 1.479, while the endcaps are at 3.14 m distance to the IP and cover $1.479<|\eta| <3.0$. The crystal length in barrel and endcap correspond to about $25 X_0$, so that electromagnetic showers are contained in the calorimeter up to high energies. In the endcap region a preshower device made of 2 lead absorber disks and 2 planes of silicon strips with 1.9 mm pitch provides additional rejection power against $\pi_0$ background. The electron resolution of an ECAL module is shown in Fig. 5.15. It is parameterised as

$$\left(\frac{\sigma}{E}\right)^2= \left(\frac{S}{\sqrt{E}}\right)^2+ \left(\frac{N}{E}\right)^2+ C^2\;,$$

((5.12))

Fig 5.15

(a) Electron energy resolution of a CMS ECAL module [23] measured in a beam test. “Hodo cuts” refers to a measurement using a beam hodoscope to restrict the incoming beam to the centre of the 3x3 electromagnetic cluster, which results in a better resolution since effects of energy loss at the crystal edges are avoided. (b) Jet transverse energy resolution for the three $\eta$ regions of the hadron calorimeter, barrel, endcap and forward [23]. The jets were reconstructed using an iterative cone algorithm with $R=0.5$. The reconstructed transverse energy, $E_T^{rec}$, is compared to the MC generated energy, $E_T^{MC}$

with a stochastic term S, noise N, and a constant term C. The expected performance parameters are also given in Fig. 5.15. The overall resolution for a 100 GeV electron is about 0.5%. Table 5.3 compares the CMS ECAL to the ATLAS LAr calorimeter [21], where the latter shows a slightly worse performance in terms of energy resolution.

Table 5.3 Comparison of general performance parameters of the calorimeter and muon detectors of the ATLAS and CMS experiments [21]

The hadronic calorimeter (HCAL) must operate in a magnetic field and is optimised to provide a maximum of absorption for hadronic showers before they reach the magnet coil. Therefore a brass/scintillator sampling technique is chosen. The HCAL covers an angular range of $|\eta| <3.0$. Since there is still some hadronic leakage outside the magnet coil, an additional layer of scintillators is installed between magnet and return yoke so that the total number of hadronic interaction lengths, $\lambda$, is between 7 and 11. The hadron barrel ($|\eta| <1.4$) is segmented in $\Delta\eta\times\Delta\phi=0.087\times 0.087$ large towers, while the endcap ($1.3<|\eta| <3.0$) is more coarse in $\phi$ with segmentations of $5^\circ$ and $10^\circ$, respectively. In the forward region $3.0<|\eta| <5.0$, a steel and quartz-fibre calorimeter provides extended coverage for jet and missing E _T measurement.

The resolution of the reconstructed transverse energy of jets, E _T, is displayed in Fig. 5.15. It is about 15% for jets of $E_T=100$ GeV and is below 10% for high energy jets. In the barrel region the resolution is parameterised by

$$\left(\frac{\sigma\left(E_T^{rec}/E_T^{MC}\right)}{\left\langle E_T^{rec}/E_T^{MC} \right\rangle}\right)^2= \left(\frac{1.25}{\sqrt{E_T^{MC}}}\right)^2 +\left(\frac{5.6}{E_T^{MC}}\right)^2 +0.033^2\;,$$

((5.13))

for jets reconstructed with an iterative cone algorithm with a cone size of $R=0.5$. The calorimetric measurement of the missing transverse energy, ${E_T^{miss}}$, is important for searches for new particles and in the reconstruction of events which involve neutrinos in the final state. Figure 5.16 shows the corresponding expected resolution for minimum-bias and soft QCD events at low-luminosity. The ATLAS and CMS hadronic calorimeter performances [21] are again compared in Table 5.3.

Fig 5.16

(a) Missing transverse energy resolution as a function of the sum of the transverse energy, $\sum E_T$, for minimum-bias and soft QCD events at low luminosity [23]. (b) Resolution for muons in the central detector, reconstructed with the muon system only, the tracker only and the combined measurement [23]

The most outer layers of the CMS detector are the muon systems . In the barrel section at $|\eta| <1.2$ drift tube (DT) chambers are installed in four radial layers between $r=4 \textrm{meter}$ and $r=7 \textrm{meter}$. The endcaps ($|\eta| <2.4$) are equipped with three large disks of cathode strip chambers (CSC), which have a higher tolerance to neutron induced background and stand a higher particle rate and a higher magnetic field. In addition, resistive plate chambers (RPC) are used in both regions, which also operate at high rates. They provide a good time resolution needed for bunch crossing detection, but they have a reduced position resolution with respect to the DTs and CSCs. In the barrel, the DT and RPC systems provide up to 44 space points for muon track reconstruction. The muon chamber information is completed by the track measurement in the inner detector. In Fig. 5.16 the individual and combined momentum resolutions of the tracking systems are compared. At low transverse momentum, the resolution is mainly determined by the inner tracker, while the muon system contributes equally at high momenta. A similar behaviour is found in the forward region. For 1 TeV muons the momentum resolution is about 5 and 7% in barrel and endcaps, respectively. The corresponding angular resolutions in $\phi$ are 1 and 10 mrad. When compared to ATLAS, as summarised in Table 5.3, the CMS stand-alone muon spectrometer has a less good resolution, which is however overcompensated in the combined reconstruction by the inner tracker.

The CMS trigger is a 3-layer system to select potentially interesting physics events which appear at rates much lower than the overwhelming inelastic pp collisions. At nominal bunch spacing and high luminosity, a reduction factor of $10^7$ from about 1 GHz interaction rate to 100 Hz event recording frequency is needed. The trigger requirements of CMS and ATLAS are therefore very similar.

The first trigger layer of CMS is implemented in custom electronics while the high-level trigger (HLT) is software based. Only the calorimeters and the muon system provide trigger information to the hardware trigger. The ECAL trigger constructs trigger primitives from the sum of the transverse energy deposited in a calorimeter tower. The tower size in the barrel is $5\times 5$ crystals, and in the endcap $5\times 5$ so-called super-crystals. The HCAL trigger works similarly, with trigger towers that follow the HCAL geometry. The global trigger receives the trigger primitives and filters events that pass a pre-defined set of thresholds. The global trigger decision is prepared and sent back to the detector within $3.2 \upmu\textrm{s}$, which then starts the transfer of the buffered data. The level-1 muon trigger is based on DT, CSC and RPC measurements. Each subsystem provides a local trigger primitive with position, direction, bunch crossing and quality information. A track finding algorithm combines the primitives and assigns a p _T measurement to the track. The global muon trigger (GMT) selects the four best tracks according to p _T and quality and transmits it to the global trigger, which applies the required momentum and isolation thresholds. The global trigger can also combine different trigger primitives and has access to the calorimetric energy sums, $\sum E_T$ and ${E_T^{miss}}$. The level-1 accept rate is up to 100 kHz.

After a positive level-1 decision data are available in dual port memories to the DAQ system. The data blocks of the different detectors are accessible for the HLT processors which further analyse the events. The global strategy is to reject unwanted events as early as possible. Regional reconstruction is preferred in the early filtering stages because it can provide sufficient information to discard events quickly. In several virtual filter levels the complexity increases, eventually exploiting the completely reconstructed event, including full particle tracking. All HLT processors are running the same software code which maximises flexibility of the system. Eventually, after successful HLT filtering, event data are written to mass storage for offline analysis.

In summary, the ATLAS and CMS detectors will provide similar performances using different detection technologies. CMS has slight advantages in tracking charged particles in the inner detector and in the measurement of electron and photon energies. ATLAS, on the other hand, performs better in jet and energy flow determination, and also the stand-alone measurement of muon momenta in the muon chambers is more precise. This is however compensated in the combined reconstruction by the performance of the inner tracking system. The final performance will however also depend on the proper understanding of the sub-detectors and trigger systems, their efficient running, and the combined reconstruction of high p _T objects that will be optimised with data from pp collisions.

5.3 Prospects for the LHC Start-Up Phase

The sub-detectors and trigger systems of ATLAS and CMS were already operational at the LHC start in September 2008. The very first event in ATLAS is shown in Fig. 5.17. It is the interaction of a single proton beam hitting the upstream collimator and producing a shower of particles along the beam direction in the ATLAS detector. Before and after this event, several million of cosmic ray triggers were recorded for calibration and alignment. The sub-systems are completed, commissioned and, apart from temporary and system-specific faults related to normal operation, fully functional. The instantaneous LHC luminosity will increase with time such that the usage for measurements with the ATLAS and CMS detectors can be divided into different stages [22]. With $10-100 \mathrm{pb^{-1}}$ of data collected in the very early phase (2010), detector calibration, trigger performance studies, trigger adjustment, and material studies will be performed. Known physics processes, like Drell-Yan Z- and W-Boson production, are used as standard candles for these tasks. Table 5.4 summarises some of the ATLAS performance goals for $e/\gamma$ energy scale and uniformity, for jet energy scale, as well as for tracking and muon alignment.

Fig 5.17

First LHC beam event recorded by ATLAS

Table 5.4 Expected ATLAS calibration and alignment performance at the start of data taking and with first data samples

In the subsequent phase, with up to about $1 \textrm{fb}^{-1}$ of data (expected in 2012), calibration and alignment will be further refined. Here, background processes for Higgs and SUSY searches need to be studied. Inclusive searches for SUSY particles, respectively their decays, will be possible in the low-mass SUSY parameter space with sensitivity to production cross-sections down to $\approx 0.5 \textrm{fb}$.

Once the amount of well understood data goes beyond $1 \textrm{fb}^{-1}$, the sensitivity extends to more rare processes, like the production SM and SUSY Higgs bosons as well as heavy new particles in the TeV range. Detailed information are collected in [23,24] .

First LHC collision data will be used to verify and improve the calibration and alignment that has already been achieved with the corresponding dedicated hardware calibration systems. As an example, $Z\to \textrm{e}^+\textrm{e}^-$ events are planned to be used to inter-calibrate the different ATLAS calorimeter regions with a relative uniformity of 0.5% between regions of size $\Delta\eta\times\Delta\phi=0.2\times 0.4$. Together with the local uniformity obtained by cell-by-cell calibration, this is necessary to reduce the constant resolution term to below 0.7%. Figure 5.18 shows that only $100 \textrm{pb}^{-1}$ of data are needed to achieve this goal. The Z decay events serve also as a reference sample from which the absolute electromagnetic energy scale can be derived. Local energy scale factors are adjusted until the shape of the di-electron invariant mass distribution corresponds well to the Breit-Wigner line-shape folded with a resolution parameterisation, as expected for $Z\to \textrm{e}^+\textrm{e}^-$ production. Since the Z mass is known to $2.1 \textrm{MeV}$ from LEP measurements [6], the electron energies can be calibrated with high precision.

Fig 5.18

(a) Constant energy resolution term from long-range non-uniformities in the calorimeter as planned to be measured from the line-shape of $Z\to \textrm{e}^+\textrm{e}^-$ events [24]. (b) Muon reconstruction efficiency determined from simulated $Z\to \mu^+\mu^-$ events and the corresponding background in $100 \textrm{pb}^{-1}$ of ATLAS data [24]. The result from the “tag &probe” method compares very well with the expectation directly derived from Monte Carlo information. The inefficiencies at $|\eta|\approx 0$ and $|\eta|\approx 1.2$ are due to the small gap between two muon barrel systems and the barrel-endcap transition region

The $Z\to \ell^+\ell^-$ decays are also an ideal tool to measure lepton reconstruction, identification and trigger efficiencies, as well as resolutions directly from data [25]. The events are triggered and selected by requiring a high-p _T lepton to tag the event and a second object in an invariant mass interval close to the Z boson mass. This object is used as a probe to derive the various efficiencies. Figure 5.18 shows, as an example, the ATLAS muon identification efficiency, as it could be determined from $100 \textrm{pb}^{-1}$ of data. The relative background is small, less than $0.1\%$, and originating from $bb\to\mu\mu+X$ production, $W\to\mu\nu$ and $Z\to\tau\tau$ decays, as well as $t\bar{t}$ production. Similar measurements are foreseen for electrons and taus. The results of these studies will be compared to the estimated detector performance to eventually derive corresponding systematic uncertainties. They will be the base for searches for the Standard Model Higgs boson or for new physics beyond the Standard Model.