ATLAS-CONF-2019-024 08July2019
ATLAS CONF Note
ATLAS-CONF-2019-024
7th July 2019
Z boson production in Pb+Pb collisions at
√sNN = 5.02 TeV measured by the ATLAS detector
The ATLAS Collaboration
The production ofZbosons is measured in the dilepton, electrons and muons, decay channel in Pb+Pb collisions at√
sNN = 5.02 TeV measured with the ATLAS detector. Data from the 2015 LHC run corresponding to an integrated luminosity of 0.49 nb−1are used for the analysis. Zboson production is measured with respect to dilepton rapidity and event centrality.
The measurements in Pb+Pb collisions are compared with similar measurements made inpp collisions at the same center-of-mass energy. The nuclear modification factor, which quantifies the compatibility of the heavy ion andppcollision results using the Glauber model calculation, is found to be consistent with unity in central and mid-central collisions. After rejecting the photon-induced background, the nuclear modification factor in peripheral collisions is above unity, although consistent with it. The nuclear modification factor measured as a function of rapidity agrees with unity while showing a trend consistent with a calculation including the isospin effect. The experimental data are compared with theoretical predictions obtained at the NLO using a proton and nuclear parton distribution functions. The measuredZ boson yields lie about 1–3σbelow the NLO predictions.
© 2019 CERN for the benefit of the ATLAS Collaboration.
Reproduction of this article or parts of it is allowed as specified in the CC-BY-4.0 license.
1 Introduction
The measurement of electroweak (EW) boson production is a key part of the heavy ion (HI) physics program at the Large Hadron Collider. Isolated photons and heavy vector bosons,Z andW, are powerful probes of the hot and dense medium formed in HI collisions. After being created at the initial stage of the collision in high-momentum exchange processes,Z andWbosons decay much faster than the timescale of the medium evolution. Their leptonic decay products are not affected by the strong interaction, hence they carry the information about the initial stage of the collision, the geometry and the dynamics inside the colliding nuclei.
Measurements performed by the ATLAS and CMS experiments withZandWparticles decaying leptonically show that the production rate of non-strongly interacting particles is proportional to the amount of the nuclear overlap, quantified by the nuclear thickness function, TAA[1–4]. Results obtained with isolated high-energy photons [5,6] are also consistent with this observation.
The transverse momentum and rapidity distributions ofZbosons and the pseudorapidity distribution of muons coming fromW bosons measured in Pb+Pb collisions at√
sNN = 2.76 TeV are consistent with perturbative Quantum Chromodynamics (pQCD) simulations of nucleon-nucleon collisions multiplied by thehTAAi [1–4]. Z boson production in Pb+Pb collisions was found to be consistent with next-to- leading-order (NLO) pQCD calculations that do not include nuclear modifications in the treatment of parton distribution functions (PDF). However, a nuclear modification of the PDFs cannot be excluded within the precision of the existing Pb+Pb measurements [1,2,5].
On the other hand, the study of asymmetricp+Pb collisions at√
sNN = 5.02 TeV shows that including nuclear modifications of the PDF gives a better description of the data than using a ‘free’ proton PDF. This is seen by comparing theZboson cross section inp+Pb collisions to pQCD calculations [7], and topp collision data collected at the same energy [8]. In addition, preliminary studies ofZ bosons differentially inp+Pb centrality demonstrate that they are a sensitive test of the Glauber model description of centrality [7].
This note presents results on Z boson production in theZ → µµandZ →eedecay channel in Pb+Pb collisions at√
sNN = 5.02 TeV with the ATLAS detector at the LHC. The data sample was collected in November 2015 and it corresponds to an integrated luminosity of 0.49 nb−1. The observables under study are the number of producedZbosons per minimum-bias event in the fiducial volume defined by detector acceptance and lepton kinematics, measured differentially in rapidity and event centrality. The Pb+Pb data are compared to pQCD calculations, and the nuclear modification factor is measured relative to previously measuredppcross section [9].
The note is organised as follows. Section2gives a brief overview of the ATLAS detector subsystems used in this analysis. Section3describes the dataset, triggers and the offline selection criteria used to select events together with Monte Carlo simulations used to correct the data for detector effects. Section4.1 provides a description of the analysis methodology while Section4.2gives a summary of the considered background sources. The lepton performance measurements are described in Section4.3. The systematic uncertainties are detailed in Section5and the results are presented and discussed in Section6. Section7 gives a summary of the main results and observations.
2 ATLAS detector
The ATLAS detector [10] at the LHC covers nearly the entire solid angle around the collision point. It consists of an inner tracking detector surrounded by a thin superconducting solenoid, electromagnetic and hadronic calorimeters, and a muon spectrometer incorporating three large superconducting toroid magnets.
The inner-detector system (ID) is immersed in a 2 T axial magnetic field and provides charged particle tracking in the range|η| <2.5.1
The high-granularity silicon pixel detector covers the vertex region and typically provides four measurements per track, the first hit being normally in the insertable B-layer[11–13] in operation since 2015. It is followed by the silicon microstrip tracker (SCT) which usually provides eight measurements per track. These silicon detectors are complemented by the transition radiation tracker (TRT), which enables radially extended track reconstruction up to|η|=2.0.
The calorimeter system covers the pseudorapidity range |η| < 4.9. Within the region |η| < 3.2, electromagnetic calorimetry is provided by barrel and endcap high-granularity liquid-argon (LAr) sampling electromagnetic calorimeters, with an additional thin LAr presampler covering|η| < 1.8, to correct for energy loss in material upstream of the calorimeters. Hadronic calorimetry is provided by the scintillating- tile calorimeter, segmented into three barrel structures within|η| < 1.7, and two LAr hadronic endcap calorimeters. The forward calorimeter (FCal) is a liquid-argon sampling calorimeter located on either side of the interaction point. It covers 3.1< |η| <4.9 and each half is composed of one EM and two hadronic sections. The FCal is used to characterise the centrality of Pb+Pb collisions as described below. Finally, zero-degree calorimeters (ZDC) are situated at large pseudorapidity,|η| >8.3, and are primarily sensitive to spectator neutrons.
The muon spectrometer (MS) comprises separate trigger and high-precision tracking chambers measuring the deflection of muons in a magnetic field generated by superconducting air-core toroids. The precision chamber system covers the region|η| <2.7 with three layers of monitored drift tubes, complemented by cathode strip chambers in the forward region, where the background is highest. The muon trigger system covers the range|η| <2.4 with resistive plate chambers in the barrel, and thin gap chambers in the endcap regions.
A two-level trigger system is used to select events of interest for recording [14]. A level-1 (L1) trigger is implemented in hardware and uses a subset of detector information to reduce the event rate. A subsequent, software-based high-level trigger (HLT) selects events for recording. Both the electron and muon event selection used in this analysis combine L1 and HLT decision algorithms.
3 Data Sets and Event Selection
All of the analyzed data were recorded in periods with stable beam, detector, and trigger operations.
Candidate events are required to have at least one primary vertex reconstructed from the inner detector tracks. In addition, a trigger selection is applied, requiring a muon or an electron candidate with apT threshold of 8 GeV or 15 GeV, respectively. The electron trigger candidate is further required to pass a
1ATLAS uses a right-handed coordinate system with its origin at the nominal interaction point (IP) in the centre of the detector and thez-axis along the beam pipe. Thex-axis points from the IP to the centre of the LHC ring, and they-axis points upwards. Cylindrical coordinates(r, φ)are used in the transverse plane,φbeing the azimuthal angle around thez-axis. The pseudorapidity is defined in terms of the polar angleθasη=−ln tan(θ/2).
set of loose criteria for the electromagnetic shower shapes [15]. The trigger algorithm implements an event-by-event estimation and subtraction of the underlying event contribution to the energy deposited in each calorimeter cell [16]. For both electron and muon candidates, further requirements are applied to suppress electromagnetic background contributions as will be described in Section4.2.
Offline reconstructed muon candidates must satisfypT >20 GeV and|η|< 2.5 and pass the requirements of medium identification [17]. Selected electron candidates are required to havepT > 20 GeVand|η| <2.47.
Candidates within the transition region between barrel and endcap calorimeters (1.37 <|η|< 1.52) are rejected. In addition, loose likelihood-based identification is applied, developed for the Pb+Pb data conditions and based on a general strategy described in Ref. [18]. At least one of the leptons in the event is required to match the corresponding trigger object.
Events with aZboson candidate are selected by requiring exactly two opposite-charge muons or electrons, at least one of which is matched to a lepton selected at trigger level. The dilepton invariant mass must satisfy the requirement 66<m`` <116 GeV. A total of 5347Z boson candidates are found in the muon channel and 4047 in the electron channel. Figure1shows the invariant mass distribution of the selected signal candidates and all considered background sources for both decay channels integrated in event centrality.
In order to understand the geometry of heavy ion collisions it is common to classify the events according to the amount of nuclear matter participating in the collision. The quantity used to estimate the collision geometry is called the ‘collision centrality’. The centrality determination is based on the total transverse energy measured by both FCal detectors in each event,ΣEFCal
T . This quantity is then mapped to geometric quantities, such as the average number of participating nucleons,hNparti, and the mean nuclear thickness function - which quantifies the amount of the nuclear overlap in a centrality class - hTAAi, calculated using the Glauber model [19,20]. The mapping is based on specific studies done on an event sample without pile-up collected with minimum-bias (MB) triggers. A special treatment is employed for the 20%
most peripheral interval of events, where diffractive and photonuclear processes significantly contribute to the MB event sample. This includes extrapolating the total number of MB events in this region and employing a special requirement on the Z boson event topology, as described in detail in Section4.2.
Table1summarises the relationship between centrality, hNparti, andhTAAifor the particular analysis bins calculated with Glauber MC v2.4 [21]. The total number of minimum-bias events in 0–80% centrality interval is(2.95±0.05) ·109which is then distributed in different centrality intervals according to their size.
Centrality hNparti hTAAi[mb−1] Centrality hNparti hTAAi[mb−1] 0–2% 399.0±1.6 28.30±0.25 25–30% 172.8±2.8 7.50±0.17 2–4% 380.2±2.0 25.47±0.21 30–40% 131.4±2.6 4.95±0.15 4–6% 358.9±2.4 23.07±0.21 40–50% 87.0±2.4 2.63±0.11 6–8% 338.1±2.7 20.93±0.20 50–60% 53.9±2.0 1.28±0.07 8–10% 317.8±2.9 18.99±0.19 60–80% 23.0±1.3 0.39±0.03 10–15% 285.2±2.9 16.08±0.18 80–100% 4.80±0.36 0.052±0.006 15–20% 242.9±2.9 12.59±0.17 0–100% 114.0±1.1 5.61±0.06 20–25% 205.6±2.9 9.77±0.18
Table 1: Centrality intervals and their corresponding geometric quantities with systematic uncertainties. From Ref.
[21].
Samples of Monte Carlo (MC) simulated events are used to evaluate the selection efficiency for signal events and the contribution of several background processes to the analysed dataset. All of the samples
are processed with the Geant4-based simulation [22,23] of the ATLAS detector. Dedicated efficiency and calibration studies with data are used to derive correction factors to account for residual differences between experiment and simulation, as is subsequently described.
The processes of interest containingZ bosons were generated with the Powheg-Box v2 MC program [24]
interfaced to the Pythia 8.186 parton shower model [25]. The CT10 PDF set [26] was used in the matrix element, while the CTEQ6L1 PDF set [27] was used with the AZNLO [28] set of generator-parameter values (tune) for the modelling of non-perturbative effects in the initial-state parton shower. The Photos++
v3.52 program [29] was used for photon radiation in EW processes.
The sample of top-quark pair (t¯t) production was generated with the Powheg-Box v2 generator, which uses NLO matrix element calculations together with the CT10f4 PDF set [30]. The parton shower, fragmentation and underlying event in nucleon-nucleon collisions were simulated using Pythia 6.428 [31] with the CTEQ6L1 PDF set and the corresponding Perugia 2012 tune (P2012) [32]. The top-quark mass was set to 172.5 GeV. The EvtGen v1.2.0 program [33] was used to model bottom and charm hadron decays for all versions of Pythia. The totalZboson and top-quark yields in MC samples are normalised using the results of NLO QCD calculations.
The signal MC samples are produced with different nucleon-nucleon combinations (pp,pn,nn) weighted by the combinatoric factors coming from the nucleon composition of lead nuclei. For lead,A=208 and Z =82 hence all the samples with two neutrons have weight of 36.7%, events with two protons 15.5% and different nucleon combinations have weights of 23.9% (pn,np).
Once produced, the simulated events are overlaid with minimum-bias events taken during the Pb+Pb run.
The overlay of data events is done such that the MC simulation accurately reflects detector occupancy conditions present in the Pb+Pb run. The minimum-bias events used for the overlay are sampled such that the centrality distribution, based on the total transverse energy deposited in the forward calorimeters, approximates that ofZ boson events which are biased to more central events. The simulated events are finally reconstructed by the standard ATLAS reconstruction software.
4 Analysis
4.1 Measurement procedure
The differential Z boson production yield per minimum-bias event is measured within a fiducial phase space defined withp`
T >20 GeV, |η`| < 2.5 and 66<m`` < 116 GeV. The yields in both, the electron and muon channel are calculated using:
NZfid= NZ−BZ Z
trig·CZ , (1)
whereNZandBZare the number of selected events in data and the expected number of background events, respectively. TheZ
trig is the trigger efficiency perZ boson candidate measured directly in the data and described in Section4.3.
A correction for the reconstruction efficiency, momentum smearing and the final state radiation effects is applied with the bin-by-bin correction factorCZ (CZ <1), which is obtained from the MC simulation as:
CZ = NZMC,sel NZMC,fid .
Here,NZMC,selis the number of events passing the signal selection at the detector level. The number of selected events is corrected for the observed differences between data and simulation in lepton reconstruction and identification efficiencies. The denominatorNZMC,fidis computed applying the fiducial phase space requirements to the generator-level leptons originating fromZboson decays. The measurement is corrected for QED final-state radiation effects by applying the fiducial phase space requirements to the lepton momenta before photon radiation.
The full statistical sample of the MC simulation allows determining theCZin the whole fiducial acceptance for all centralities as 0.7274±0.0007 and 0.5497±0.0004 in theZ → µ+µ−andZ →e+e−decay channels, respectively.
The rapidity, momentum and the centrality dependence of theCZis calculated from the simulation as:
CZ(pT,y,ΣEFCal
T )= F(pT,y)G(y,ΣEFCal
T ) (2)
whereF(pT,y)is the efficiency calculated per bin inyandpT of the dilepton system andG(y,ΣEFCal
T )is a parabolic parametrisation of a correction factor accounting for the centrality dependence of the efficiency.
In each rapidity bin, the factorGis obtained from a fit of the ratio of the efficiency in a particular centrality bin to the value averaged over all possible centrality values.
Nuclear modification can be quantified by measuring the ratio of theZ boson production rate scaled by the mean nuclear thickness function to theZ boson production cross section inppcollisions, called the nuclear modification factor:
RAA(y)= 1 hTAAiNevt
dNPb+Pb/dy
dσp p/dy , (3)
where hTAAi is the nuclear thickness function in a given centrality class, (1/Nevt)dNPb+Pb/dy is the differential yield ofZbosons per inelastic minimum-bias event measured in Pb+Pb collisions and dσp p/dy is the differentialZ boson cross section measured in ppcollisions [9]. A deviation from unity inRAA indicates the nuclear modification of the observable. Note thatRAAis expected to be greater than unity by about 2.5% (based on the MC simulation) due to the higher Z boson production cross section of proton-neutron and neutron-neutron interactions which are present in Pb+Pb collisions. This is later denoted as the "isospin effect" and is not accounted for in the definition ofRAA.
4.2 Background determination
There are two background source categories studied in this analysis. The first includes the same background sources that are studied inppcollisions [9] and the second includes additional background sources specific to the Pb+Pb collision system.
The background contributions in the first category are expected fromZ →τ+τ−, top quark pair production and multi-jet events. The first two contributions are evaluated from dedicated simulation samples, whereas the the multi-jet background contribution is derived with data-driven approach. TheZ →τ+τ−background is found to be 0.05% of all signal candidates in the muon channel and 0.06% in the electron channel. The top-quark background amounts to 0.08% in the muon channel and 0.05% in the electron channel. The background contribution fromWboson decays and theW-jet production is found to be negligible.
[GeV]
µ
mµ
−2
10
−1
10 1 10 102
103
104
Events/GeV
Preliminary ATLAS
=5.02 TeV, 0.49 nb-1
sNN
Pb+Pb, µ-
µ+
→ Z
0-100% centrality
Data→µ+µ-
Z
EM background Multi-jet
τ-
τ+
→ Z Top quarks
60 70 80 90 100 110 120
[GeV]
µ
mµ
0.51 1.52
Pred.Data
[GeV]
mee
−2
10
−1
10 1 10 102
103
104
Events/GeV
Preliminary ATLAS
=5.02 TeV 0.49 nb-1
sNN
Pb+Pb, e-
e+
→ Z
0-100% centrality
Data - +e
→e Z
EM background Multi-jet
τ-
τ+
→ Z Top quarks
60 70 80 90 100 110 120
[GeV]
mee
0.51 1.52
Pred.Data
Figure 1: Detector-level invariant mass distribution of dimuon (left) and dielectron (right) pairs fromZ boson decays together withZ →τ+τ−, top quark, multi-jet and the EM background contributions for any value of the event centrality. Only the statistical uncertainties of the data are shown.
The multi-jet background originates from jets, mis-identified hadrons and in the electron channel, from converted photons. Its contribution was estimated in the muon channel from the distribution of the same-signZboson candidates in rapidity and centrality which does not exhibit a peak in theZboson mass region due to the low charge misidentification rate in the muon spectrometer. This background amounts to 0.5% of all the signal candidates. In the electron channel, there is a significant contribution from charge mis-identifications, fakes and conversions. The electron same-sign pairs therefore cannot be used to estimate the multi-jet background. This contribution to the selected event sample in the electron channel is finally estimated using a template fit to data inZboson rapidity and event centrality. The template is derived from a subset of data that corresponds to the electrons from jets, i.e. electrons passing the same kinematic requirements but with a very poor reconstruction quality. Due to a low number of data candidates available for building the template, no requirement on the pair charge is made hence the template includes electrons which are a part of the signal sample. The shape of the obtained QCD template is shown in Figure1where it was normalised to the number of the same-sign data candidates in the low-mass region of 60<mee < 70 GeV after the signal MC subtraction. This background amounts to 2% of all the signal candidates in the electron channel.
The background contributions in the second category come from two main sources. The first is that theZ boson candidate can be assigned to the wrong centrality interval due to the pile-up (more than a single Pb+Pb collision recorded simultaneously). The second is the production of additionalZ boson candidates by photon-induced reactions produced by the intense electromagnetic fields generated by the colliding ions (below referred to as the "electromagnetic background"). The pile-up affects the transverse energy measured in the FCal and causes reconstructedZ bosons to be assigned to a wrong centrality interval. The pile-up events from the same collision (in-time pile-up) increase theΣEFCal
T , shifting theZboson candidate to a more central interval. Alternatively, if the pile-up collision precedes the trigger event (out-of-time pile-up), its contribution to theΣEFCal
T can be negative, due to the time response of the electronic signal shapers used in the calorimeter [34]. In this case, theZboson candidate is shifted to a more peripheral interval. Both processes depend on the instantaneous luminosity during the data taking. At any time during the HI run the number of interactions per bunch crossing was less than 0.04. To preserve the accuracy of
the total yield measurement the pile-up removal procedure is not applied in the event selection. However, due to the fact that theZboson production is expected to grow linearly withTAA[1], the increase in the FCal transverse energy in in-time pile-up events transfers candidates from less populated to more populated centrality intervals, thus having a very small effect and changing the average number of counts in the most central collisions by significantly less than one. Contrary to that, the reduction in theΣEFCal
T transfers out-of-time pile-up events from more populated to less populated centrality interval, thus expected to have a larger effect to peripheral intervals. The effect is studied in the analysis using several independent data driven and MC simulation based approaches. The possible contribution to the most peripheral centrality interval due to this type of the pile-up is less than 2%, i.e. less than one count and significantly less in any other centrality range.
A non-negligible contribution to yields in peripheral centrality bins is expected from electromagnetic background sources. On the other hand, the expected rate of signal events in those peripheral centrality bins is low. There are two kinds of photon induced processes expected to contribute to the background.
One is the photon-photon scattering,γγ→`+`−[35–37] and the other is the photon-nucleus scattering γ+ A → Z →`+`− [38]. While measurements of the exclusive high mass dilepton production were performed in ppand Pb+Pb collisions by ATLAS [39, 40] and in ppby CMS [41], a mechanism of photo-nuclearZproduction has not yet been observed in HI collisions. The impact parameter of the photon involved processes is typically larger than twice the nuclear radius and therefore such processes are studied in so-called ultra-peripheral collisions, where they are not obscured by hadronic interactions. Both physics processes are characterised by large rapidity gaps on one or both sides of the detector, which are used in the analysis to measure and subtract these backgrounds. The rapidity gap analysis is implemented using the same technique as developed in [42].
In the 50-100% peripheral centrality intervals, an additional selection is made on the ZDC detector signal coincidence in order to suppress electromagnetic background contributions. The energy measured in ZDC detectors is required to be at least 1 TeV on both sides, corresponding to 40% of the energy deposition of a single neutron. Without using the ZDC coincidence requirement on the event selection, there are 34 events with a rapidity gap greater than 2.5 units found in both decay channels. Since the estimated number of hadronicZboson candidates with such gap is below 0.05, all these events are considered to be produced by photon-induced reactions and are removed from the sample. Events without gaps can have a photon-induced dilepton pair as well, if the rapidity gap is filled by particles coming from a nucleon-nucleon interaction that occured simultaneously. Such event would appear in the centrality intervals defined by the ΣEFCal
T deposition mainly coming from these simultaneous interactions.
Following a paper recently published by ATLAS [43], the contribution from photon-induced reactions can be statistically disentangled from the simultaneous contribution due to hadronic collisions by measuring angular and momentum correlations of final state dilepton pairs. One variable used to quantify these correlations is the dilepton acoplanarity, defined as
Aco≡1− |φ+−φ−|
π ,
whereφ±are the azimuthal angles of the two produced leptons. The same observable is used in this analysis to further quantify a contribution of photon-induced reactions in the measuredZboson production.
Based on the MC simulation and measurements at 0-50% centrality, 13.3±0.4% ofZboson candidates produced by hadronic collisions have an acoplanarity below 0.01. On the other hand, among the 34 events with a rapidity gap greater than 2.5 units which were rejected from our sample as pure photon-induced events, 26 are found to have Aco<0.01, corresponding to a fraction of 76.5%. This demonstrates that the
acoplanarity is highly selective to the photon-induced reactions in theZ boson invariant mass region. This allows for correcting the photon-induced background in all centrality intervals by comparing the number of Z boson candidates in a given centrality interval to the number of candidates with Aco<0.01.
It is estimated that in the 80–100% centrality interval, besides the events with a large rapidity gap, 7±3 out of 28 remaining candidates are coming from the photon-induced reactions. In the 60–80% interval this number is 15±5 out of 182, and in 50–60% interval it is 18±8 out of 258, where the uncertainty includes statistical and systematic uncertainties on the background but not on the number of event candidates. The EM background shape denoted in Figure1is derived from the events containing large rapidity gaps and is normalised to the total estimated number of background candidates quoted above.
In more central collisions, the method of estimating the photon-induced background is limited by the statistical precision of the sample and its contribution is consistent with zero. Rapidity distributions in measured centrality intervals are normalised to the estimated number of signal events and an additional systematic error is assigned in each bin to account for the EM background subtraction procedure.
4.3 Detector performance corrections
After subtracting background contributions, the number ofZ-boson candidates is corrected for the trigger efficiency and detection efficiency, according to Eq.1. All the correction factors are derived directly from the current data set used in the analysis. The dilepton trigger efficiencyZ
trigis derived from the efficiency of the single lepton trigger`via the relationZ
trig=1− (1−`1)(1−`2), where indices are the two leptons forming the candidate pair. In order to obtain theZ
trigas a function of the dileptonpTandywhich is further applied as a correction per dilepton candidate, kinematic distributions of the decay products are taken from the MC simulation.
Muon and electron trigger efficiencies are measured in the data with the tag-and-probe method [15,17,44]
as a function ofηandφ. The tag lepton is required to be reconstructed and identified with high purity and the probe lepton is paired to it to give an invariant mass in the range 66<m`` <116 GeV. The background contribution to this measurement is estimated from the number of same-sign pairs and amounts up to 0.8%
and 3.5% in the muon and electron channels respectively.
The single muon trigger efficiency in the endcap region of the detector (1.05 < |η| <2.4) is measured to be around 85%, and in the barrel region (|η| <1.05) it varies from 60% to 80%. A significant dependence of the efficiency on the muon azimuthal angleφwas measured and thus the trigger efficiency correction is propagated as a function of bothφandηto the final correction. The single electron trigger efficiency is measured to be at the level of 95% in the end-cap part of the calorimeter (1.52 < |η| < 2.47) reaching 97% in the barrel (|η| < 1.37). A significant dependence on the electron pT was measured where the efficiency rises from 85% to 97% in the range from 20–100 GeV integrated overη. The single electron trigger efficiency is thus propagated in dependence ofpT andηfor the final correction.
TheZ
trigis measured to be on average 94.4±0.3% in the muon decay channel and 99.74±0.03% in the electron channel, constant with the event centrality.
Selection requirements including the muon reconstruction and identification are imposed on muon candidates used in the analysis. The efficiency of the selection criteria is measured in data with the tag-and-probe method in Z → µ+µ−events [15,17] and compared with the simulation. Ratios of the efficiencies determined in data and simulation are applied as scale factors (SF) to correct the simulated
events. Since the measured efficiencies are found to have negligible dependence on the muon momentum in the selected kinematic region and a very weak centrality dependence, theSFare evaluated only as a function of muonη. The centrality dependence of theSFis taken into account for evaluation of systematic uncertainties.
The combined reconstruction and identification efficiency for medium-quality muons typically exceeds 84%
in both the data and simulation with good agreement between the two estimates. The largest discrepancy is observed in the endcap region (|η| >1.8) resulting with theSFof 4% away from unity. In the analysis, theSF is applied in three differentη regions as follows: 0.970±0.01 for|η| < 0.8, 0.987±0.01 for 0.8 < |η|< 1.8 and 1.04±0.02 for|η| >1.8.
Electron candidates used for the analysis are required to satisfy selection criteria related to reconstruction and identification. The efficiency of the selection is measured in data with the tag-and-probe method in Z →e+e−events, as described in Ref. [44], and compared with the simulation to derive electron scale factors. Measurements are performed as a function of the electronη,pTand event centrality. The electron reconstruction and identification efficiency is measured to be typically 70% in the endcap (|η| >1.52) with a good agreement between the data and the simulation. TheSF is measured to the precision of 3% in that region. In the barrel region (|η| <1.37) the efficiency in data is measured to be around 80% while in the MC simulation the efficiency reaches to 85%. Therefore, in this region a significantSF is applied, measured with a precision of 3–5%.
The lepton momentum scale and resolution corrections are derived using the signal MC samples and are applied to the simulation for both electrons and muons. For the reconstructed muon objects, these corrections are derived as a function of the muonηandφ[17]. The correction factors are chosen such that they minimise the χ2between the muon-pair invariant mass distributions in data and simulation. The energy scale of reconstructed electrons is corrected by applying a per-object correction factor to the data derived from a comparison of the electron-pair invariant mass between the simulation and the data. This procedure was found to be sensitive to the pile-up distribution in data due to different settings used for the signal readout from the EM calorimeters. Therefore, a dedicated set of energy scale correction factors was derived for the data sets used in this analysis and in theppreference measurement [9].
5 Systematic uncertainties
This section describes different sources of systematic uncertainties affecting the measurement. It summarises the contributions from the trigger efficiency uncertainty, background subtraction uncertainty, the uncertainty on the correction factorCZ, and finally, the uncertainties on the geometrical parameters calculated in the Glauber model.
In both channels, the trigger efficiency is derived completely from the tag-and-probe results in the data.
The statistical limitation of the measured sample determines the uncertainty associated with the trigger correction. Although the uncertainties in each bin are relatively large on the muon trigger efficiency as seen in Figure8, after propagating this uncertainty to the dimuon efficiency, where we only require one of the muons to fire the trigger, the total uncertainty is quite small, around 1% when derived as function of centrality and 1–2% when derived as a function of theZ rapidity. The uncertainty is propagated using MC pseudo-experiments. In the electron channel, due to higher efficiency of the electron trigger, this uncertainty amounts to maximally 0.5%.
The NLO cross section of the background samples ofZ →τ+τ−and top quark was varied by 10% to derive the corresponding uncertainties [9]. However, in both channels the multi-jet background dominates the uncertainty contribution. In the muon channel, the combinatorial background normalisation was varied by 10% which amounts to 0.5% uncertainty on the finalZ boson yield. In the electron channel, the multi-jet template normalisation is varied by 20% which corresponds to the level of the statistical uncertainty for the number of same-sign candidates in the low-mass control region. The overall contribution to the systematic uncertainty is about 0.5% in rapidity bins and 0.5–2% in centrality bins.
In the 50–100% centrality interval, the photon-induced background subtraction uncertainty is evaluated by taking into account the uncertainty on the acoplanarity cut efficiency for the hadronicZboson production evaluated from the data and the simulation and the uncertainty on the background rejection efficiency evaluated from the candidates with large gaps. The first component, which is at the level of 0.4%, accounts for the difference between the efficiency measured in the data and the simulation. The uncertainty on the background rejection efficiency of the acoplanarity cut has two sources. One is the statistically limited size of the event sample with large rapidity gaps which amounts to 7%. Another contribution comes from accounting for the difference in the acoplanarity distributions for electrons and muons which is about 8%. In the 0–50% centrality intervals where the background subtraction is not performed, an uncertainty of 0.4% evaluated from the difference between the data and the MC in the acoplanarity distribution is introduced to account for possible residual EM background contribution.
Uncertainties in the determination of lepton reconstruction and identification efficiency scale factors, as well as the parametrisation of the centrality dependence of the total correction affect the measurements through the correction factorsCZ.
In the muon channel, the scale factors in the threeη regions described in Section4.3are modified by their errors to derive the corresponding systematic uncertainty on theCZ. In addition, the impact of the measuredSF dependence on the event centrality on the finalCZ is also evaluated. The total relative uncertainty from these two sources on theZboson yield is from 3.1% at midrapidity (|y| < 0.5) to 4.5% in forwardZboson rapidities and gives a contribution, constant with respect to the event centrality, of∼3.4%
inZ boson yields.
The main contribution to the systematic uncertainty in the electron channel comes from the uncertainties in measuring the reconstruction scale factor. Uncertainties related to efficiency are classified as either correlated or uncorrelated, and are propagated accordingly to the final measurement uncertainty. The correlated uncertainty component of the SF was obtained by varying the requirements on the tag electron identification and isolation and on the invariant mass of the tag and probe pair. The statistical, uncorrelated, components of the scale factor uncertainties are propagated to the measurements via MC pseudo-experiments, while the systematic components are propagated as a single variation fully correlated across all electronηintervals. This source gives a 3–5% uncertainty as a function of rapidity and around 3% for all centrality intervals.
In thepp reference measurement [9] which uses the same lepton energy calibration corrections, it is measured that the contribution of the muon momentum calibration and scale uncertainty amounts to 0.03%. Similarly, the electron momentum calibration and scale uncertainty contribute 0.004% and 0.065%, respectively. Therefore, these uncertainties were neglected in this measurement.
An additional uncertainty on the bin-by-bin correction is due to the parameterisation of the correlation of the dependence ofCZon rapidity and centrality described in Equation2and it stems primarily from the limited statistics of the MC data set. To estimate uncertainties associated with these assumptions the parameters of the functionG(y,ΣEFCal
T )are varied by the parabolic fit errors on the parameters. The data
0 0.5 1 1.5 2 2.5
20 40 60 80 100
Z -1 -1 [pb]y/dN d〉T〈NevtAA 120
Preliminary ATLAS
Pb+Pb, 0.49 nb -1
=5.02 TeV sNN
0-100% centrality
→ll Z
µ µ
→ Z
→ ee Z
<116 GeV mll
66<
>20 GeV
l
pT
|<2.5 η l
|
0 0.5 1 1.5 2 2.5
| y
|
0.9 1 1.1
CombinedChannel 0 100 200 300 400
300 400 500 [pb] ZN -1〉 AAT〈 -1 evtN
Preliminary ATLAS
Pb+Pb, 0.49 nb -1
=5.02 TeV sNN
<116 GeV mll
66<
>20 GeV
l
pT
|<2.5 η l
|
→ll Z
µ µ
→ Z
→ee Z
0 100 200 300 400
part 〉
〈N 0.8
1 1.2
CombinedChannel
Figure 2: NormalisedZboson yields measured in muon and electron decay channels together with combined yield plotted as a function of rapidity (left) andNpart(right). Lower panels show the ratio of individual channels to the combined result. The error bars in the upper panels show the total uncertainty for muons and electrons and the statistical uncertainty for the combined data. In the lower panel, the error bar shows the statistical uncertainty. The shaded band (left) and boxes (right) show the systematic uncertainty of the combined result in both panels. The width of the error band in the right panel corresponds to the systematic uncertainty inNpart, scaled by 3 for clarity.
is corrected with these variations, and the difference between these results and the standard correction are taken as an estimate of the systematic uncertainty. In the muon channel the uncertainty associated with this source ranges from 0.5–1.5% in rapidity bins and is constant at∼ 0.5% as a function of centrality, with the exception of the most peripheral bin where the uncertainty is∼1%. In the electron channel, the uncertainty as a function of the rapidity ranges from 0.5–1% and∼1% in centrality bins, whereas in the most peripheral bin this contribution rises to∼2%.
For both channels and the combined result, the uncertainties on the geometric parameters (hTAAiand hNparti) listed in Table1range from about 1% in central collisions to about 12% in peripheral collisions.
They are treated as fully correlated between the channels. Finally, the measured luminosity uncertainty for theppmeasurement [9] used for theRAA calculation is 1.9% and is considered correlated between the channels.
6 Results
The rapidity distributions of theZ yield normalised by the number of MB events and nuclear thickness functionhTAAiare shown in the left upper panel of Figure2for the muon and electron decay channels.
The right panel of the figure shows the centrality dependence of the normalisedZboson yield in the full fiducial acceptance. The measurements performed in both channels are combined using the Best Linear Unbiased Estimate (BLUE) method [45], accounting for the correlations of the systematic uncertainties across the channels and measurement bins. The combined result is shown in Figure2with circles together with the combined statistical and systematic uncertainties. The level of agreement between the channels shown in the lower panels of the figure is quantified and it yields χ2/Nd.o.f. = 2.59/5 as a function of rapidity andχ2/Nd.o.f. =21.6/14 for the result as a function of centrality.
![Table 1 summarises the relationship between centrality, hN part i , and hT AA i for the particular analysis bins calculated with Glauber MC v2.4 [21]](https://thumb-eu.123doks.com/thumbv2/1library_info/3999961.1540395/4.892.165.749.795.981/table-summarises-relationship-centrality-particular-analysis-calculated-glauber.webp)
![Figure 6: Ratio of W + (circles) and W − (squares) yields measured in the same data set [55] to the yields of Z bosons versus hN part i compared with the ratios measured in pp data [9] scaled by the isospin factor obtained from the CT14 NLO PDF calculation](https://thumb-eu.123doks.com/thumbv2/1library_info/3999961.1540395/15.892.303.610.531.775/figure-circles-squares-measured-compared-measured-obtained-calculation.webp)
![Figure 7: Nuclear modification factor R AA in centrality intervals compared with the HG-Pythia model [56] scaled by the isospin factor obtained from the CT14 NLO PDF calculation](https://thumb-eu.123doks.com/thumbv2/1library_info/3999961.1540395/16.892.301.611.150.397/figure-nuclear-modification-centrality-intervals-compared-obtained-calculation.webp)
