Z boson production in Pb+Pb collisions at √SNN=5.02 TeV measured by the ATLAS experiment

Contents lists available atScienceDirect

Physics

Letters

B

www.elsevier.com/locate/physletb

Z boson

production

in

Pb+Pb

collisions

at

√

s

_NN

=

5

.

02 TeV measured

by

the

ATLAS

experiment

.

The

ATLAS

Collaboration



a

r

t

i

c

l

e

i

n

f

o

a

b

s

t

r

a

c

t

Articlehistory:

Received30October2019

Receivedinrevisedform27December2019 Accepted27January2020

Availableonline31January2020 Editor:D.F.Geesaman

The production yield of Z bosons is measured in the electron and muon decay channels in Pb+Pb collisions at√sNN=5.02 TeVwith the ATLAS detector. Data from the 2015 LHC run corresponding toan integratedluminosity of0.49 nb−1 areused forthe analysis.The Z bosonyield,normalised by the totalnumber ofminimum-biasevents andthe meannuclearthicknessfunction,is measuredasa functionofdileptonrapidityand eventcentrality.ThemeasurementsinPb+Pbcollisions arecompared with similar measurementsmadein proton–proton collisionsatthe same centre-of-massenergy.The nuclearmodificationfactorisfoundtobeconsistentwithunityforallcentralityintervals.Theresults arecomparedwiththeoreticalpredictionsobtainedatnext-to-leadingorderusingnucleonandnuclear parton distributionfunctions. The normalised Z boson yields inPb+Pb collisions lie 1–3

σ

abovethe predictions.Thenuclearmodificationfactormeasuredasafunctionofrapidityagreeswithunityandis consistentwithanext-to-leading-orderQCDcalculationincludingtheisospineffect.

©2020TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense (http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP3_.

1. Introduction

The measurement ofelectroweak (EW) boson production is a key part of the heavy-ion (HI) physics programme at the Large HadronCollider(LHC).Isolated photonsandheavyvector bosons,

Z and W , are powerful tools to probe the initial stages of HI collisions. Afterbeing createdat the initial stage of the collision in high-momentumexchange processes, Z and W bosonsdecay muchfaster than the timescale ofthe medium’s evolution.Their leptonicdecayproductsaregenerallyunderstoodtonotbeaffected bythestronginteraction;hencetheycarry informationaboutthe initialstage ofthecollision andthe partonicstructureofthe nu-clei.

Measurementsperformedbythe ATLAS andCMSexperiments with Z and W bosons decaying leptonically show that produc-tion rates of these non-strongly interacting particles are propor-tionalto the amount of nuclearoverlap, quantified by the mean nuclearthicknessfunction,



TAA



[1–6].Resultsobtainedwith iso-latedhigh-energy photons [7,8] are also consistent withthis ob-servation.

The transverse momentum and rapidity distribution of Z

bosons and the pseudorapidity distribution of muonsoriginating fromW bosonsmeasuredinPb+Pbcollisions at

√

sNN

=

2

.

76 TeV have been found to be generally consistent with perturbative quantumchromodynamics(pQCD)calculationsofnucleon–nucleon

 _{E-mailaddress:atlas.publications@cern.ch.}

collisions scaledby



TAA



.Productionof Z bosonsinPb+Pb colli-sionswasfoundtobeconsistentwithnext-to-leading-order(NLO) pQCDcalculationsthatdonotincludenuclearmodificationsinthe treatment ofpartondistributionfunctions(PDFs). However, some nuclear modification of PDFs could not be excluded within the precision of theexisting Pb+Pb measurements [1,2,7]. The recent ALICE result at

√

sNN

=

5

.

02 TeV shows better agreement with nPDFcalculationsatforwardrapidities [9].

On the other hand, the study of asymmetric p+Pb collisions at

√

sNN

=

5

.

02 TeVshowsthatincludingnuclearmodificationsof PDFsgives asubstantially better descriptionof thedatathan us-ingafreeprotonPDF.ThisisseenbycomparingtheZ bosoncross sectionin p+PbcollisionswithpQCD calculations [10–13] and re-cently also the W boson cross section at

√

sNN

=

5

.

02 TeV and

√

sNN

=

8

.

16 TeV[13,14].Inaddition,studiesof Z bosons differen-tiallyinp+Pbcentralitydemonstratethattheyareasensitivetest oftheGlaubermodeldescriptionofnucleargeometry [11].

Thisletter presents resultson Z boson productionyield mea-surement in the Z

→

μμ

and Z

→

ee decay channels in Pb+Pb collisionsat

√

sNN

=

5

.

02 TeVwiththeATLASdetectorattheLHC. ThedatasamplewascollectedinNovember2015andcorresponds to an integrated luminosity of 0.49 nb−1_{.The observables under} study are the yield of produced Z bosons in the fiducial kine-maticregiondefinedbydetectoracceptanceandleptonkinematics normalisedtothenumberofminimum-biasevents,measured dif-ferentially in rapidity and event centrality. The Pb+Pb data are compared with pQCD calculations, and the nuclear modification

https://doi.org/10.1016/j.physletb.2020.135262

0370-2693/©2020TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).Fundedby SCOAP3_.

factorismeasuredrelativetopp crosssectionpreviouslymeasured bytheATLASexperiment [15].

2. ATLASdetector

The ATLAS detector [16] covers nearly the entiresolid angle1

aroundthecollisionpoint.Itconsistsofaninnertrackingdetector surrounded by a thin superconducting solenoid, electromagnetic andhadronic calorimeters,and a muon spectrometer incorporat-ingthreelargesuperconductingtoroidmagnets.

Theinner-detectorsystemisimmersedina2 Taxialmagnetic fieldandprovidescharged-particletrackingintherange

|

η

|

<

2

.

5. Thehigh-granularitysiliconpixeldetectorcoversthevertexregion andtypically provides fourmeasurements per track, the first hit beingintheinsertableB-layer [17,18] inoperationsince2015.Itis followedby thesiliconmicrostrip tracker,whichusually provides eightmeasurementspertrack.Thesesilicondetectorsare comple-mentedby thetransition-radiation tracker,whichenablesradially extendedtrackreconstructionupto

|

η

| =

2

.

0.

The calorimetersystemcovers the pseudorapidity range

|

η

|

<

4

.

9. Within the region

|

η

|

<

3

.

2, electromagnetic (EM) calorime-tryisprovidedbybarrelandendcaphigh-granularityliquid-argon (LAr)samplingcalorimeters,with an additionalthinLAr presam-plercovering

|

η

|

<

1

.

8,to correctforenergylossin material up-streamofthecalorimeters.Hadroniccalorimetryisprovidedbythe scintillator-tilecalorimeter,segmentedintothreebarrelstructures within

|

η

|

<

1

.

7, andtwo LAr hadronicendcap calorimeters.The forward calorimeter(FCal) is a LAr sampling calorimeterlocated on either side of the interaction point. It covers 3

.

1

<

|

η

|

<

4

.

9 andeachhalfis composedofone EMandtwo hadronicsections. TheFCal isused tocharacterise thecentralityof Pb+Pbcollisions asdescribedbelow.Finally,zero-degreecalorimeters(ZDC)are sit-uatedatlargepseudorapidity,

|

η

|

>

8

.

3,andareprimarilysensitive tospectatorneutrons.

The muon spectrometer comprises separate trigger and high-precisiontrackingchambersmeasuringthedeflectionofmuonsin amagneticfieldgeneratedbysuperconductingair-coretoroids.The precision chamber systemcovers the region

|

η

|

<

2

.

7 with three layers of monitored drift tubes, complemented by cathode-strip chambersintheforwardregion,wherethebackgroundishighest. Themuontriggersystemcoverstherange

|

η

|

<

2

.

4 with resistive-platechambersinthebarrel,andthin-gapchambersintheendcap regions.

Atwo-level trigger systemis usedto selectevents ofinterest forrecording [19].Thelevel-1(L1)triggerisimplementedin hard-wareandusesasubset ofthedetectorinformationto reducethe eventrate.Thesubsequent,software-basedhigh-leveltrigger(HLT) selects events for recording. Both the electron and muon event selection used in this analysis combine L1 and HLT decision al-gorithms.

3. Datasetsandeventselection

Allof theanalysed data were recordedin periodswith stable beam, detector, and trigger operations. Candidate events are re-quiredtohaveatleastoneprimaryvertexreconstructedfromthe inner-detectortracks.Inaddition,atriggerselectionisapplied, re-quiringamuon oran electroncandidateabove a pT thresholdof

1 _{ATLASusesaright-handed coordinatesystemwith itsoriginat thenominal}

interactionpoint(IP)inthecentreofthedetectorandthez-axisalongthebeam pipe.Thex-axispointsfromtheIPtothecentreoftheLHCring,andthe y-axis

pointsupwards.Cylindricalcoordinates(r,φ)areusedinthetransverseplane,φ beingtheazimuthalanglearoundthez-axis.Thepseudorapidityisdefinedinterms ofthepolarangleθasη= −ln tan(θ/2).

8 GeV or 15 GeV, respectively. The electron-trigger candidate is further required to satisfy a set of loose criteriafor the electro-magneticshowershapes [20].Thetriggeralgorithmimplementsan event-by-eventestimationandsubtractionoftheunderlying-event contributiontothetransverse energydepositedineach calorime-ter cell [21].For both theelectron andmuon candidates, further requirementsare appliedtosuppresselectromagneticbackground contributions,asdescribedinSection4.2.

MuoncandidatesreconstructedofflinemustsatisfypT

>

20 GeV and

|

η

|

< 2.5andpasstherequirementsof‘medium’identification optimised for2015analysisconditions [22]. Offlineselected elec-tron candidatesarerequiredtohave pT

>

20 GeVand

|

η

|

<

2

.

47, although candidates within the transition region between barrel andendcapcalorimeters(1

.

37

<

|

η

|

<

1

.

52) arerejected. In addi-tion, likelihood-based identification is applied, developed for the Pb+Pb dataconditionsand basedon ageneralstrategy described inRef. [23].

Events with a Z boson candidate are selected by requiring exactly two opposite-charge muons or electrons, at least one of whichismatchedtoaleptonselectedattriggerlevel.Thedilepton invariant massmust satisfy therequirement66

<

m

<

116 GeV

consistent withprevious ATLASmeasurements. Atotal of5347 Z

bosoncandidatesarefoundinthemuonchanneland4047inthe electronchannel.

In order to estimate the geometric characteristics of HI colli-sions,itiscommontoclassifytheeventsaccordingtotheamount of nuclearoverlapin thecollision. Thequantity used toestimate the collisiongeometryis calledthe‘collisioncentrality’. The cen-tralitydeterminationisbasedonthetotaltransverseenergy mea-suredbybothFCaldetectorsineachevent,



EFCal_T .Thisquantityis thenmappedtogeometricquantities,suchastheaveragenumber ofparticipatingnucleons,



Npart



,andthemeannuclearthickness function,



TAA



,whichquantifiestheamountofnuclearoverlapin acentralityclassandisevaluatedusingaGlauber calculation [24,

25].The mappingisbasedonspecific studiesofan eventsample withoutadditionalPb+Pbcollisionswithinthesameor neighbour-ing bunch crossings(pile-up) collected with minimum-bias (MB) triggers. A special treatment is employed for events in the 20% mostperipheral interval, wherediffractiveand photonuclear pro-cesses contribute significantly to the MB event sample. This re-quires extrapolating fromthe total number ofMB eventsin this regionandemployingaspecialrequirementontheZ bosonevent topology,asdescribed inSection 4.2.Table 1summarises the re-lationshipbetweencentrality,



Npart



,and



TAA



ascalculatedwith GlauberMCv2.4 [6,26],whichincorporatesnucleardensities aver-agedoverprotonsandneutrons.ThetotalnumberofMBeventsin the0–80%centralityinterval is

(

2

.

99

±

0

.

04

)

×

109,whichisthen distributedindifferentcentralityintervalsaccordingtotheir size. ThequoteduncertaintyonthenumberofMBeventsincludes vari-ationson the



E_TFCalvalue corresponding tothe 0–80%centrality intervalestimatedwiththeGlaubermodel.Thissampleisobtained by selecting eventspassingMB triggers andexcluding theevents withapile-upcontribution,wherethetotalsampledintegrated lu-minositycorrespondstothesignalselection [25].

SimulatedsamplesofMonteCarlo (MC)eventsareusedto eval-uatetheselectionefficiencyforsignaleventsandthecontribution ofseveralbackgroundprocessestotheanalyseddataset.Allofthe sampleswereproducedwiththe Geant4-basedsimulation [27,28] of the ATLAS detector. Dedicated efficiency and calibration stud-ies withdataare usedto derive correctionfactors toaccount for residualdifferencesbetweenexperimentandsimulation.

The processes ofinterestcontaining Z bosonswere generated with the Powheg-Box v1 MC program [29–32] interfaced to the Pythia _{8.186 parton shower model [33]. The CT10 PDF set [34]} was usedinthe matrixelement,whiletheCTEQ6L1 PDFset [35]

Table 1

Centralityintervalsandtheircorrespondinggeometricquantitieswithsystematicuncertainties,from Ref. [6,26].

Centrality [%] Npart TAA[mb−1] Centrality [%] Npart TAA[mb−1]

0–2% 399.0±1.6 28.30±0.25 20–25% 205.6±2.9 9.77±0.18 2–4% 380.2±2.0 25.47±0.21 25–30% 172.8±2.8 7.50±0.17 4–6% 358.9±2.4 23.07±0.21 30–40% 131.4±2.6 4.95±0.15 6–8% 338.1±2.7 20.93±0.20 40–50% 87.0±2.4 2.63±0.11 8–10% 317.8±2.9 18.99±0.19 50–60% 53.9±2.0 1.28±0.07 10–15% 285.2±2.9 16.08±0.18 60–80% 23.0±1.3 0.39±0.03 15–20% 242.9±2.9 12.59±0.17 80–100% 4.80±0.36 0.052±0.006 0–100% 114.0±1.1 5.61±0.06

was usedwiththeAZNLO [36] setof generator-parametervalues (tune)forthemodellingofnon-perturbativeeffectsinthe initial-state partonshower.The Photos++v3.52program [37] was used forfinal-statephotonradiationinEWprocesses.

Asampleoftop-quarkpair (t

¯

t)productionwasgeneratedwith the Powheg-Box v2 generator, which uses NLO matrix-element calculations [38] together with the CT10f4 PDF set [39]. The parton shower, fragmentation and underlying event in nucleon– nucleon collisions were simulated using Pythia 6.428 [40] with the CTEQ6L1 PDF set and the corresponding Perugia 2012 tune (P2012) [41]. Thetop-quarkmasswas setto172.5 GeV. The Evt-Gen v1.2.0 program [42] was used to model bottom and charm hadron decays for all versions of Pythia. The total Z boson and top-quarkyields in MC samplesare normalised usingthe results ofNLOQCDcalculations.

ThesignalMCsampleswereproducedwithdifferentnucleon– nucleoncombinations(pp, pn,nn) weightedtoreflecttheisospin composition oflead nuclei. For lead, A

=

208 and Z

=

82, so all sampleswithtwoneutronshaveaweightof

(

[

A

−

Z

]/

Z

)

2

=

36.7%, eventswith two protons have a weight of 15.5% and the unlike nucleoncombinations(pn,np)eachhaveweightsof23.9%.

Once produced, the simulated events were overlaid with MB eventstakenduringthePb+Pbrun.Theoverlayofdataeventswas donesuch thattheMC simulationaccurately reflectsdetector oc-cupancyconditionspresentinthePb+Pbrun.TheMBeventsused for the overlay were sampled such that the centrality distribu-tion,basedonthetotaltransverseenergydepositedintheforward calorimeters,approximatesthatof Z bosonevents,whichare gen-erallybiasedtomore-centralcollisions.Thesimulatedeventswere finally reconstructed by the standard ATLAS reconstruction soft-ware.

4. Analysis

4.1.Measurementprocedure

The differential Z boson production yield per MB event is measured withina fiducial phasespace definedby p

T

>

20 GeV,

|

η



|

<

2

.

5 and66

<

m

<

116 GeV.Theyieldsinboththeelectron

andmuonchannelarecalculatedusing

Nfid_Z

=

NZ

−

BZ

Z

trig

·

CZ

,

(1)

whereNZ and BZ are thenumberofselected eventsindataand

theexpectednumberofbackgroundevents,respectively, and

Z

trig isthe triggerefficiency per Z bosoncandidate measured indata anddescribedinSection4.3.

Acorrection forthe reconstruction efficiency,momentum res-olution and the final-state radiation effects is applied with the bin-by-bin correction factor CZ which is obtainedfrom MC

sim-ulationas

CZ

=

NMC_Z ,sel NMC_Z ,fid

.

Here, NMC_Z ,sel is the number of events passing the signal selec-tion at the detectorlevel. The number ofselected events is cor-rected for measured differences between data andsimulation in lepton reconstruction and identificationefficiencies. The denomi-nator NMC_Z ,fid iscomputedbyapplyingthefiducialphase-space re-quirementstothegenerator-levelleptonsoriginatingfromZ boson

decays.ThemeasurementiscorrectedforQEDfinal-stateradiation effectsbyusingthegenerator-levelleptonkinematicsbefore pho-tonradiation.

The value of CZ in the fiducial acceptance averaged over all

centralitiesisdeterminedfromMCsimulationafterreweightingas explained inSection 3.It is 0

.

659 and 0

.

507 inthe Z

→

μ

+

μ

−

and Z

→

e+e− decaychannels, respectively. Theuncertainty due tothesizeofthesimulatedsampleisatthelevelof0.1%foreach decaychannelandisnotthedominantuncertainty.

The rapidity, momentum and centrality dependence of CZ is

calculatedfromthesimulationas

CZ

(

pT

,

y

,



EFCalT

)

=

F

(

pT

,

y

)

G

(

y

,

EFCalT

),

(2) where F

(

pT

,

y

)

isthecentrality-averagedefficiencycalculatedper y and pT interval of the dilepton system and G

(

y

,



EFCalT

)

is a parabolicparameterisationofacorrectionfactoraccountingforthe centrality andrapidity dependences ofthe efficiency. In each ra-pidity bin,the factor G is obtainedfroma fit ofthe ratioof the efficiencyinaparticularcentralitybintothevalueaveraged over allpossiblecentralityvalues.

Nuclearmodificationisquantifiedbymeasuringtheratioofthe

Z boson production rate, scaled by the mean nuclear thickness function,tothe Z bosonproductioncrosssection inpp collisions,

aquantityknownasthenuclearmodificationfactor:

RAA

(

y

)

=

1



TAA



Nevt dN_PbZ_+Pb

/d y

d

σ

Z pp

/d y

,

where



TAA



isthe nuclearthicknessfunction in agiven central-ityclass,

(

1

/

Nevt

)

dN_PbZ_+Pb

/

d y isthedifferentialyieldof Z bosons perinelasticMBeventmeasuredinPb+Pbcollisions andd

σ

Z

pp

/

d y

is the differential Z boson cross section measured in pp

colli-sions [15]. A deviation from unity in RAA indicates the nuclear modification of the observable. The value of RAA is expected to begreaterthanunitybyabout2.5%,basedonMCsimulation,due tothehigher Z bosonproductioncrosssectioninproton–neutron andneutron–neutroninteractionswhicharepresentinPb+Pb col-lisions and amount to 84.5% of the total hadronic cross section. Thisislaterreferredtoasthe‘isospineffect’andisnotaccounted forinthedefinitionofRAA.

Fig. 1. Centrality-integrateddetector-levelinvariantmassdistributionof(left)dimuonand(right)dielectronpairstogetherwiththeZ→τ+τ−,topquark,multi-jetandEM backgroundcontributions.Onlythestatisticaluncertaintiesofthedataareshown.

4.2. Backgrounddetermination

There are two background source categories studied in this analysis. Thefirst includesthe samebackgroundsources that are studied in pp collisions [15] and the second includes additional backgroundsourcesspecifictothePb+Pbcollisionsystem.

Background contributions in the first category are expected from Z

→

τ

+

τ

−,top-quarkpairproductionandmulti-jet events. The first two contributions are evaluated from dedicated simu-lation samples,whereas the multi-jet background contributionis derivedusingadata-drivenapproach.The Z

→

τ

+

τ

− background is found to be 0.05% of all signal candidates in the muon chan-neland0.06%in theelectronchannel. Thetop-quarkbackground amountsto0.08%inthemuon channeland0.05%intheelectron channel. Thebackground contributionfrom W bosondecays and

W +jetproductionisfoundtobenegligible.

The multi-jet background originates from jets, misidentified hadronsand, intheelectron channel,fromconvertedphotons. In themuonchannel,itscontributionisestimatedfromthe distribu-tionofthesame-sign Z bosoncandidatesinrapidityandcentrality. Due to the low charge misidentification rate in the muon spec-trometer,their invariantmassdistributiondoesnotexhibitapeak inthe Z bosonmassregion.Inthemuonchannelthisbackground amountsto 0.5% ofallsignal candidates. Intheelectron channel, there is a significant contribution from charge misidentification, fakesandconversions.Theelectronsame-signpairstherefore can-notbeusedtoestimatethemulti-jetbackground.Thiscontribution totheselectedeventsample intheelectronchannel isestimated using a background template obtained from the data in Z

bo-son rapidity and event centrality. The template is derived from a subset ofthe signal sample that corresponds to electrons from jets,i.e.electronswithaverypoorreconstructionquality [23] that alsosatisfy pTcone20

/

2pT

>

0

.

05



EFCalT

+

0

.

025,where pTcone20 isthetotaltransversemomentummeasuredinsidetheconeofsize

R

=

0

.

2 aroundthe electrontrack.The templateis normalisedto thenumberofsame-sign datacandidatesinthe low-massregion of60

<

mee

<

70 GeVafterthesignalMC subtraction.Duetothe

smallnumberofsignal candidatessatisfyingthiscondition, same-signelectronpairswiththesamekinematicrequirementsarealso added to this background template. The shape of the obtained multijettemplateisshowninFig.1.Thisbackgroundamountsto 2%ofallsignalcandidatesintheelectronchannel.

ThebackgroundcontributionsspecifictoPb+Pbcomefromtwo mainsources.Thefirstisduetopile-upwhenmorethanasingle Pb+Pb collision is recorded simultaneously orin a nearby bunch crossing.Thesecondistheproductionofadditional Z boson

can-didatesbyphoton-inducedreactionsproducedbytheintense elec-tromagneticfieldsgeneratedbythecollidingions(belowreferred toas‘electromagneticbackground’).Pile-updistortsthetransverse energy measured in theFCal and causesreconstructed Z bosons

to be assigned to an incorrect centrality interval. Pile-up events fromothercollisionsinthesamebunchcrossing(in-timepile-up) increase the



E_TFCal, shifting the Z boson candidate to a more-central interval. Alternatively, if a pile-up collision precedes the trigger event(out-of-time pile-up),its contribution tothe



EFCal_T

canbe negative,duetothetimeresponseoftheelectronicsignal shapers used in the calorimeters [43]. In this case, the Z boson

candidate isshifted toa more-peripheralinterval. Both processes dependontheinstantaneousluminosityduringdatataking.Atany time during the HI run the number of hadronicinteractions per bunchcrossingwaslessthan0.01.Topreservetheaccuracyofthe total yield measurement, no pile-up removal procedure was ap-plied tothe selectedevents.However, duetothe fact thatthe Z

bosonproductionscaleslinearlywith



TAA



[1],theincreaseinthe FCal transverse energy in in-time pile-up events transfers candi-dates from lesspopulated tomore-populated centralityintervals, thus havinga verysmalleffectandchangingtheaveragenumber ofcountsinthemostcentralcollisionsbyaninsignificantamount. Contrarytothat,thereductioninthe



EFCal_T transfersout-of-time pile-up events from more-populated to less-populated centrality intervals, thus making alarger relative contributionto the more-peripheral events.Theeffecthasbeenstudiedusingseveral inde-pendentdata-drivenandsimulation-basedapproaches.Thelargest contribution to the most peripheral 80–100% centrality interval dueto thistype of thepile-up islessthan 2%, i.e.lessthan one count,andissignificantlylessinanyothercentralityrange.

A non-negligible relative contribution to the dileptons in the

Z bosonmass rangeinperipheral centralityintervalsisexpected fromelectromagneticbackgroundsources.Ontheotherhand,the expectedrateofsignaleventsinthoseperipheralcentralitybinsis low.Twophoton-inducedprocessesare expectedtocontribute to the background: photon–photon scattering,

γ γ

→ 

+

_

− _[44–46]

andphoton–nucleusscattering

γ

+

A

→

Z

→ 

+



−[47].Although measurements of exclusive high-mass dilepton production have beenperformedin pp andPb+Pb collisionsby ATLAS [48] andin

pp collisionsbyCMS [49] aphoto-nuclearZ bosonproductionhas notyetbeenobservedinHIcollisions.Whentheimpactparameter of thephoton-induced processesis largerthan twice the nuclear radiussuchprocessesarereferredtoasultra-peripheralcollisions, andinthesethey arenotobscuredbyhadronicinteractions.Both physics processes arecharacterisedby large rapiditygaps onone orbothsidesofthedetector(regionswithnoparticleproduction

recordedinthedetector),whichareusedinthisanalysisto mea-sureandsubtract thesebackgrounds.Therapidity gapestimation isimplementedusingasimilartechniqueasdevelopedinRef. [50]. Inthe50–100%(peripheral)centralityintervals,thereisan ad-ditionalrequirementofaZDCsignal coincidenceinorderto sup-press the electromagnetic background contributions. The energy measuredin eitherdetectorisrequiredto be atleast1 TeV, cor-responding to 40% of the energy deposition of a single neutron. WithoutusinganyZDCcoincidence requirementinthe event se-lection, 34 events witha rapidity gapgreater than 2.5 units are foundin thetwo decaychannels. Since theestimatednumberof hadronic Z boson candidates with such a gap is below 0.05, all oftheseeventsareconsideredtobeproducedby photon-induced reactionsandare removedfromthesample. Events withoutgaps can have a photon-induced dilepton pair as well, ifthe rapidity gapisfilledbyparticlesoriginatingfromasimultaneousnucleon– nucleon interaction.These events would appearin the centrality intervalsdefinedby the



EFCal_T deposition fromthe hadronic in-teraction.

FollowingRef. [51],photon-inducedreactionsoccurring simul-taneouslywithhadroniccollisionscanbeidentifiedusingboththe angularand momentum correlationsof final-statedilepton pairs. One variable used to quantify these correlations is the dilepton acoplanarity,definedas

α

≡

1

− |φ

+

− φ

−

|/

π

,where

φ

± are the azimuthal angles of the two opposite-charge leptons. The same observable is used in this analysis to extract the contribution of photon-induced reactions to the measured Z boson produc-tion.BasedontheMCsignalsimulationandmeasurementsinthe 0–50%centralityinterval,(13

.

3

±

0

.

4)%of Z bosoncandidates pro-ducedbyhadroniccollisionshaveacoplanaritybelow0.01.Onthe otherhand,amongthe34eventswitharapidity gapgreaterthan 2.5units which were rejectedfrom the sample aspure photon-inducedevents,26arefoundtohave

α

<

0

.

01,correspondingtoa fractionof76%.Thisdemonstratesthattheacoplanarityissensitive tophoton-inducedreactionsinthe Z bosoninvariant-massregion. Thisallowsthephoton-inducedbackgroundtobeestimatedinall centralityintervals by comparing the number of Z boson candi-datesinagivencentralityinterval withthenumberofcandidates with

α

<

0

.

01.

Itisestimatedthat inthe80–100%centrality interval,besides the events witha large rapidity gap, 7

±

3 out of 28 remaining candidatesoriginatefromphoton-inducedreactions.Inthe60–80% interval this number is 15

±

5 out of 182, and in the 50–60% interval it is 18

±

8 out of 258,where the uncertainty includes statisticalandsystematicuncertaintiesinthebackground,butnot inthe numberofeventcandidates.Inmore-centralcollisions,the methodofestimatingthephoton-inducedbackgroundislimitedby thestatisticalprecisionofthesampleandthiscontributionis con-sistentwithzero.Rapidity distributionsin centralityintervalsare normalisedtotheestimatednumberofsignaleventsandan addi-tionalsystematicerrorisassignedineach intervalto accountfor theEMbackgroundsubtractionprocedure.

Fig. 1 shows the invariant mass distribution of backgrounds from all considered sources for both decay channels integrated ineventcentrality. The EM backgroundshape shownin Fig. 1is derivedfromtheeventscontaininglargerapiditygapsandis nor-malised tothe total estimatednumber ofbackground candidates quotedabove.

4.3.Detectorperformancecorrections

After subtracting background contributions, the number of Z

bosoncandidatesiscorrectedforthetriggerefficiencyand detec-tionefficiency, accordingto Eq. (1). All thecorrection factors are derived directly from the current data set used in the analysis,

with the exception of the lepton momentum calibration correc-tions that are derived from pp collision data, and extrapolated to thePb+Pbdataset conditions.The triggerefficiencyper recon-structed Z candidate

Z

trig is derived from the efficiency of the single-leptontrigger

viatherelation

trigZ

=

1

−(

1

−

1

)(

1

−

2

)

,

wheretheindicesrefer tothetwo leptonsforming thecandidate pair.In orderto obtain

Z

trig asafunction ofthedilepton pT and y whichisfurtherapplied asacorrection perdileptoncandidate, kinematicdistributions of thedecayproducts are takenfromMC simulation.

Muon and electron trigger efficiencies are measured using a tag-and-probemethod [19,22,23] asafunctionof

η

and

φ

.Thetag leptonis requiredtobereconstructed withhighquality andvery lowprobabilityofbackgroundcontaminationandtobematchedto a leptonselectedattriggerlevel. Theprobelepton, satisfyingthe analysisreconstructionandidentificationrequirements,ispairedto ittogiveaninvariantmassintherange66

<

m

<

116 GeV.The

background contribution to this measurement is estimated from the numberofsame-sign pairs andamounts to0.8% and3.5% in themuonandelectronchannels,respectively.

Thesingle-muontriggerefficiencyintheendcapregion ofthe detector (1

.

05

<

|

η

|

<

2

.

4) ismeasured tobe around85%, andin the barrel region (

|

η

|

<

1

.

05) it variesbetween 60% and80%. A significant dependence of the efficiency on the muon azimuthal angle

φ

wasmeasuredandthusthetriggerefficiencycorrectionis derived asafunction ofboth

φ

and

η

.Thesingle-electrontrigger efficiencyis measured tobe around 95% inthe endcapregion of thecalorimeter (1

.

52

<

|

η

|

<

2

.

47)anditincreasesslightlyto97% inthebarrel (

|

η

|

<

1

.

37).Asignificantdependenceontheelectron

pT wasmeasured,andtheefficiencyrisesfrom85%to97%inthe rangefrom 20to 100 GeV integratedover

η

.The single-electron trigger efficiency is thus derived as a function of pT and

η

.The average

Z

trig is (94

.

2

±

0

.

2)% in the muon channel and(99

.

74

±

0

.

03)%inthe electronchannel,constantwiththeeventcentrality within1%.

Selectionrequirements includingthemuonreconstruction and identificationare imposed onmuon candidates usedin the anal-ysis. The efficiency of the selection criteria is measured using a tag-and-probe method in Z

→

μ

+

μ

− events [22] and compared withsimulation.Ratiosoftheefficiencies determinedindataand simulationareappliedasscalefactors(SF)tocorrectthesimulated events.Sincethemeasured efficienciesare foundto have negligi-bledependenceonthemuonmomentumintheselectedkinematic region anda very weak centralitydependence, theSF are evalu-ated onlyasa functionofmuon

η

. Thecentralitydependence of theSFistakenintoaccountintheevaluationofsystematic uncer-tainties.Thecombinedreconstructionandidentificationefficiency formedium-qualitymuonstypicallyexceeds84%inboththedata andsimulationwithgood agreementbetweenthetwo estimates. Thelargestdifferenceisobservedintheendcapregion(

|

η

|

>

1

.

8). Ultimately, the SF valuesforthree different

η

regions are as fol-lows:0

.

97

±

0

.

01 for

|

η

|

<

0

.

8,0

.

99

±

0

.

01 for0

.

8

<

|

η

|

<

1

.

8 and 1

.

04

±

0

.

02 for

|

η

|

>

1

.

8.

Electron candidates used for the analysisare requiredto sat-isfy selection criteriarelatedto reconstruction andidentification. Theefficiencyoftheselection ismeasured usingatag-and-probe methodin Z

→

e+e− events,asdescribed inRef. [23],and com-pared with simulation to derive electron scale factors. Measure-mentsareperformedasafunctionoftheelectron

η

, pTandevent centrality.Theelectronreconstructionandidentificationefficiency is measured to be typically 70% in the endcap (

|

η

|

>

1

.

52) with good agreement betweenthe data and the simulation.The SF is measuredtobe1%awayfromunitywithaprecisionof3%inthat region.Inthebarrelregion(

|

η

|

<

1

.

37)theefficiencyismeasured

tobearound80%whileintheMCsimulationtheefficiencyreaches 85%.Therefore,inthisregionasignificantSFisapplied,measured withaprecisionof3–5%.

Theleptonmomentumscaleandresolutioncorrectionsare de-rivedusingthe pp signalMCsamplesandareappliedtothe sim-ulation for both the electrons and muons. For the reconstructed muons,thesecorrectionsarederivedasafunctionofthemuon

η

and

φ

[22].Thecorrectionfactorsarechosen suchthatthey min-imisethe

χ

2 _{betweenthemuon-pairinvariantmassdistributions} indataandsimulation.Theenergyscaleofreconstructedelectrons iscorrectedbyapplyingtothedataa per-electroncorrection fac-tor. The momentum scale correction factors are derived from a comparison of the electron-pair invariant mass between simula-tionanddata.

5. Systematicuncertainties

In both channels, the trigger efficiency is derived from the tag-and-probe resultsinthedata. Thestatisticallimitation ofthe measured sampledetermines the uncertaintyassociatedwiththe triggercorrection. Althoughtheuncertainties ineach binare rel-ativelylargeinthemuon triggerefficiency,afterpropagating this uncertaintytothedimuonefficiency,whereonlyoneofthemuons isrequiredtofire thetrigger, thetotaluncertaintyis quitesmall, between0.2% and 0.5% when derived asa function ofcentrality and0.1–0.2%whenderivedasafunctionof Z bosonrapidity.The uncertainty ispropagated usingMC pseudo-experiments andthe uncertainties inthe linearfit coefficientsof thetrigger efficiency asafunctionofcentrality.Intheelectronchannelthisuncertainty isatmost0.5%.

The NLO cross section of the background samples of Z

→

τ

+

τ

− and top-quark production is varied by 10% to derive the correspondinguncertainties [15].However,inbothdecaychannels themulti-jet backgrounddominates theuncertainty contribution. Inthemuonchannel,themultijetbackgroundcontributionis var-iedby 10% accordingto thelevelof statisticaluncertaintyinthe numberof same-sign candidates. This source produces 0.01–0.1% uncertaintyinrapiditybinsandupto0.2%uncertaintyincentrality bins.Intheelectronchannel,themulti-jettemplatenormalisation isvariedby 20%,whichcorresponds tothelevelofstatistical un-certaintyinthe numberofsame-sign candidatesinthelow-mass control region. The overall contribution to the systematic uncer-taintyisabout0.5%inrapiditybinsand0.5–2%incentralitybins.

Inthe50–100%centralityinterval,theuncertaintyin subtract-ing the photon-induced background is evaluated by considering two sources. The first source is the compatibility between the acoplanaritycutefficienciesforhadronicZ bosonproduction eval-uated fromdataandsimulation. Anuncertainty of0.4% accounts forthis difference. Forthe second source, the uncertainty inthe background rejection efficiency of the acoplanarity cut is evalu-atedfromthecandidateswithlargerapiditygaps.Thisuncertainty has two contributions. One is the statistical uncertainty of the eventsamplewithlargerapidity gaps,whichamounts to7%.The othercontributionisduetotheobserveddifferencesbetweenthe acoplanaritydistributionsforelectronsandmuons,whichamounts to about 8%. In the 0–50% centrality interval, where the back-groundsubtractionisnotperformed,anuncertaintyof0.4%, evalu-atedfromthedifferencebetweenthedataandsimulation acopla-naritydistributions,isintroducedtoaccountforapossibleresidual EMbackgroundcontribution.

Uncertaintiesinthedeterminationofleptonreconstructionand identificationefficiencyscalefactors,aswellasthe parameterisa-tionofthecentralitydependenceofthetotalcorrectionaffectthe measurementsthroughthecorrectionfactorsCZ.

In the muon channel,the scale factors inthe three

η

regions described in Section 4.3 are modified by their errors to derive the corresponding systematic uncertainty of CZ. In addition, the

impact of the measured SF dependenceof the final CZ value on

theeventcentralityisalsoevaluated.Thetotalrelativeuncertainty fromthesetwosourcesrangesfrom3.1%atmidrapidity(

|

y

|

<

0

.

5) to4.5%atforwardZ bosonrapiditiesandgivesacontribution, con-stant as afunction ofthe eventcentrality, of

∼

3.4% for Z boson

yields.

Themaincontributiontothesystematicuncertaintyinthe elec-tron channel comes from the uncertainties inmeasuring the re-constructionefficiencyscalefactor.Uncertaintiesrelatedtothis ef-ficiencyareclassifiedaseithercorrelatedoruncorrelated,andare propagated accordinglytothefinal measurementuncertainty.The correlateduncertaintycomponentoftheSFisobtainedbyvarying the requirements on the tag electron identification andisolation and on the invariant mass of the tag-and-probe pair. The statis-tical, uncorrelated, components of the scale factor uncertainties are propagatedtothemeasurements viaMCpseudo-experiments, whilethesystematiccomponentsarepropagated asasingle vari-ation fully correlated across all electron

η

intervals. This source givesa2.5–5%uncertaintyasafunctionofrapidityandaround3% forallcentralityintervals.

Theeffectofthecalibrationandenergyscalecorrection uncer-taintyofelectronsandmuonsisnegligible.

An additional uncertainty of the bin-by-bin correction is due to theparameterisationoftherapidityandcentralitydependence of CZ described in Eq. (2) and it stems primarily from the

sta-tistical uncertainty ofthe MC data set.To estimate uncertainties associated with these assumptions the parameters of the func-tion G

(

y

,



EFCal_T

)

are varied by the errors of the parameters of theparabolicfitincludingcovariancebetweentheparameters.The data are corrected with these variations, and the difference be-tween theseresults and the standard correction are taken asan estimate ofthe systematic uncertainty. Inthe muon channel the uncertainty associated withthis source rangesfrom 0.4% to 1.4% in rapidity bins and is constant at

∼

0.5% as a function of cen-trality, with the exception of the mostperipheral bin where the uncertainty is

∼

1%. Inthe electron channel, theuncertainty asa function ofthe rapidity ranges from0.4% to1.6% andis

∼

1% for mostcentralityintervals,althoughinthemostperipheralbinthis contributionrisesto

∼

2%.

For both channels and the combined result, the uncertainties of thegeometric parameters (



TAA



and



Npart



) listed inTable 1 range from about 1% in central collisions to about 12% in pe-ripheral collisions. These are treated as fully correlated between the channels. Finally, the total uncertainty for the pp

measure-ment [15] usedforthe RAAcalculationis2.3%.

6. Results

The rapidity distributions of the Z boson yield for the muon and electron decay channels, normalised by the number of MB events andmean nuclear thicknessfunction



TAA



, are shownin the upper-leftpanel ofFig.2.The upper-rightpanelofthe figure showsthe



Npart



dependenceofthenormalised Z bosonyieldin the fiducialacceptancewhere thesystematicuncertainties of the



Npart



valuesarescaledbya factorofthreeforclarity.The mea-surementsperformedinthetwochannelsarecombinedusingthe BestLinearUnbiasedEstimate(BLUE)method [52],accountingfor thecorrelationsofthesystematicuncertaintiesacrossthechannels andmeasurementbins.ThecombinedresultisshowninFig.2 to-gether withthe combinedstatisticalandsystematicuncertainties. Thelevelofagreementbetweenthechannelsshowninthelower

Fig. 2. NormalisedZ bosonyieldsmeasuredinthemuonandelectrondecaychannelstogetherwiththecombinedyieldasafunctionof(left)rapidityand(right)Npart.

Lowerpanelsshowtheratioofindividualchannelstothecombinedresult.Theerrorbarsintheupperpanelsshowthetotaluncertaintyformuonsandelectronsandthe statisticaluncertaintyforthecombineddata.Inthelowerpanels,theerrorbarsshowthestatisticaluncertainty.Theshadedband(left)andboxes(right)showthesystematic uncertaintyofthecombinedresultinbothpanels.ThewidthofeacherrorboxintherightpanelcorrespondstothesystematicuncertaintyofNpart,scaledbyafactorof

threeforclarity.Thepointscorrespondingtomuonandelectrondecaychannelsareshiftedhorizontallyineachpanelrelativetothebincentre.

Fig. 3. Theupperpanelsshowtherapiditydependenceof(left)thenormalisedZ bosonyieldsand(right)oftheRAAcomparedwiththeoreticalpredictions.Thelowerpanels

showtheratioofthetheoreticalpredictionstothedata.TheexpectedcontributionoftheisospineffecttotheRAAisshownintheupper-rightpanelbythedashedline.The

errorbarsonthedatapointsindicatethestatisticaluncertaintiesandtheshadedboxesshowthesystematicuncertainties.Theerrorbarsonpredictionsshowthetheoretical uncertainty.ThepointscorrespondingtonuclearPDFpredictionsareshiftedhorizontallyrelativetothebincentreforclarity.

panelsofthefigureisquantifiedas

χ

2

_/

_N

dof

=

1

.

7

/

5 asafunction ofrapidityand

χ

2

_/

_N

dof

=

21

.

6

/

14 asafunctionofcentrality. The measured Z boson yields are compared with theoretical predictionsobtainedusingamodifiedversion of DYNNLO 1.5 [53,

54] optimisedforfastcomputations.Thecalculationisperformed at O

(

α

S

)

in QCD and at leading order in the EW theory, with parameters set according to the Gμ scheme [55]. The input pa-rameters(theFermiconstantGF,themassesandwidthsofW and Z bosons,andtheCKMmatrixelements)aretakenfromRef. [56]. The DYNNLOpredictionsarecalculatedusingthefree protonPDF setCT14NLO [57] typicallyusedtocomparewiththe pp dataand, additionally,the nuclearPDF sets nCTEQ15 NLO [58] and EPPS16 NLO [59], which are averaged over each Pbnucleus. In addition, the parton-level NLO prediction from the MCFM code [60], in-terfaced to the CT14 NLO PDF set, is calculated. This takes into accountthe isospineffect,dueto differentpartoniccompositions ofprotonsandneutronsinthePbnuclei,whichisneglectedinthe DYNNLOcalculations.Therenormalisationandfactorisationscales, respectively denoted by

μ

r and

μ

f, are set to the value of lep-tonpairinvariantmass.Theuncertainties ofthesepredictionsare derived asfollows.Theeffects ofPDF uncertainties are evaluated fromthevariationscorrespondingtoeachNLOPDFset. Uncertain-tiesduetothescalesaredefinedbytheenvelopeofthevariations obtained by changing

μ

r and

μ

f by a factor of two from their

nominalvaluesandimposing0

.

5

≤

μ

r

/

μ

f

≤

2.Theuncertainty in-ducedbythestrongcouplingconstant isestimatedbyvarying

α

S by

±

0

.

001 aroundthecentralvalue of

α

S

(

mZ

)

=

0

.

118,following

the prescription ofRef. [57];the effect ofthese variations is es-timatedbycomparingthe CT14nlo_as_0117 and CT14nlo_as_0119 PDFsets [57] to CT14NLO.ImperfectknowledgeoftheprotonPDF and the scale variations are the main contributions to the total theory uncertainties. Incalculating the RAA predictions, only the nuclear PDF uncertainties contribute since the CT14 NLO uncer-taintiescancel.

InFig.3thenormalised Z bosonyieldiscomparedbetweenthe combinedmeasurementandthetheoreticalpredictionscalculated with the CT14, nCTEQ15 and EPPS16 NLO PDF sets, with uncer-taintiesassignedaspreviously described.Allcalculationslie1–3

σ

belowthedatainall rapidityintervals,integratedoverevent cen-trality. Calculationsusing nuclear PDF sets deviate fromthe data morestronglythancalculationsbasedonlyonthe CT14NLO PDF set. Asimilar observationfor theCT14 PDF was madein the pp

collision system [15] wheresystematicdeviations fromthe mea-sured values are observed for calculations made at the NNLO. When comparing the measured RAA with calculations, shown in therightpanelofFig.3,residualdeviationsfromthedataare ob-served. Thetrend observedin dataisconsistent withtheisospin effect only,expected fromthe differentvalence quark content of

Fig. 4. NormalisedZ bosonyieldversusrapiditymeasuredinthreecentrality in-tervals:peripheral30–100%(bluecircles),mid-central10–30%(greensquares)and central0–10%(reddiamonds).ThedifferentialZ bosoncrosssectionsmeasuredin

pp collisionsareshownbyopencircles [15].Theerrorbarsonthedatapoints in-dicatethestatisticaluncertaintiesandtheshadedboxesshowthetotalsystematic uncertainties.ThelowerpanelshowstheRAAandthecontributionfromtheisospin

effectcalculatedwithCT14NLOPDF(dashedline).Theshadedboxesatunity in-dicatethecombineduncertaintyfromthepp dataaddedinquadraturetotheTAA

uncertainty.Thepointsineachcentralityintervalareshiftedhorizontallyrelativeto thebincentreforclarity.

Fig. 5. TheupperpanelshowsthenormalisedZ bosonyieldasafunctionofNpart.

Theintegratedfiducialcrosssectionmeasuredinthepp systemisshownatNpart=

2 (opencircle).Thelowerpanelshowsthenuclearmodificationfactor.Theerror barsonthedatapointsindicatethestatisticaluncertaintiesandtheshadedboxes showthetotalsystematicuncertainties.Theblueshadedbandindicatesthetotal uncertaintyofthe pp data.ThefigureshowstheresultscalculatedwithGlauber MCv2.4 [24] andv3.2 [61].ThedashedandthefulllineindicateCT14NLOPDF calculationsthataccountforisospineffectsforthetwoGlauberMCversions.The widthoftheerrorboxforeachcentralityintervalcorrespondstothesystematic uncertaintyinNpart,scaledupbyafactorofthreeforclarity.

protonsandneutronsinthePbnucleus,shownintheupper-right panel of Fig. 3 by the dashed line. These results are in contrast withrecent measurements of EW bosons in p+Pb collisions [11,

14] whichfavournPDFsetstodescribethedata.

Fig.4shows thenormalised Z bosonyieldasafunctionof ra-pidityforthreecentralityintervals.Theresultsareconsistentwith eachotherwithintheirrespectivestatisticaluncertainties.Thesize of the current data sample precludes making a more definitive statement aboutanypossible modificationof the Z boson rapid-itydistributionwithcentrality.

Fig.5showsthecentralitydependenceofthenormalised Z

bo-sonyieldandofRAAcomparedwithresultsfrompp collisionsand GlauberMC.Thepoint correspondingtothe pp crosssection [15] is shown in the plot at Npart

=

2. The results are derived from Glauber MC v2.4 and a newer version v3.2 following the same procedureasdescribedinRef. [6].Theresultsarefoundtobe con-sistentwitheachotherwithinexperimentaluncertainties.Thenew Glauber MC calculation [61] implements a more advanced treat-mentofthe nucleardensityprofileandan updated experimental

Fig. 6. RatiosofW+(circles)andW−(squares)yieldsmeasuredinthesamedata set [6] totheyieldsofZ bosonsversusNpartcomparedwiththeratiosmeasured

inpp data [15] (shadedbands)scaledbyisospinfactorsobtainedfromtheCT14

NLOPDFcalculation.Theerrorbarsonthedatapointsindicatethestatistical un-certaintiesandtheshadedboxesshowthetotalsystematicuncertainties.Thepoints areshiftedhorizontallyrelativetothebincentreforclarity.

value of the nucleon–nucleon inelastic cross section. From these two updates, the formerdirectly affects the normalised Z boson

yieldderivedinthisanalysiswhiletheupdatedcrosssectionvalue has no appreciable effect. The new model also leads to a set of reducedsystematicuncertaintiescomparedwiththeprevious ver-sion.

Intheestimationoftheisospineffectcontributionshownwith dashedandfulllinesinFig.5,theGlauberMCv3.2modelaccounts fortheslightlylargerradiusoftheneutrondistributioncompared totheprotonsinthePbnucleus,oftencalledtheneutronskin ef-fect.However,duetotheweakdependenceof Z bosonproduction ontheisospincontentofthecollidingbaryons,thepredictionsof thetwoGlauberMCversionsgiveessentiallythesameresult.

Thenormalisedyields areconsistentwiththe pp crosssection atallmeasuredcentralitiesandshowonlyaweakdependenceon

Npart.Thevaluesof RAA,showninthelowerpanel,areconsistent with unity within the total uncertainty. When the isospin effect is taken intoaccount as shownwith the dashed line, the model seems to agree better with the data at low Npart values rather than athighvalues.Toquantify thedependenceof RAA on Npart, the dataare fitto a linearfunction.Including statisticaland sys-tematic uncertainties thedecrease in the RAA value betweenthe most-peripheral(80–100%)andmost-central(0–2%) centrality in-tervalsisfoundtobe

(

10

±

7

)

% and

(

6

±

6

)

% forGlauberMCv2.4 andv3.2,respectively.

The Z boson measurement is used to compare the Npart de-pendence of W [6] and Z boson productionby calculatingtheir yieldratiosasshowninFig.6,wheretheuncertaintiesofthetwo measurementsareconservativelytreatedasuncorrelated.Thedata pointsarecomparedwiththeratiomeasuredinpp collisions [15] thatisscaledbytheisospinfactorscalculatedusingtheCT14NLO PDF set. The measurements for both channels are found to be consistent withthescaled pp measurement andshow a constant behaviourasafunctionofcentrality.

The trend of the points shown in Fig. 5 for Z bosons is dif-ferent fromthetrendobserved bythe ALICECollaboration inthe measurementswithchargedhadronswithhightransverse momen-tum [62].It wasrecently showninRef. [63],that the RAA in pe-ripheralnucleus–nucleuscollisionscandeviatefromunityduetoa biasedclassificationoftheeventgeometryforeventscontaininga hardprocess.Inthatanalysis,thevalueofRAAwithoutanynuclear effectswasdeterminedbyusingtheHG-Pythia model [63],which cancreateanensembleofeventswherethe Hijing [64] event gen-erator is used to determine the numberof hard sub-interactions foreachevent,andtheparticleproductionisdeterminedsolelyby

Fig. 7. Nuclearmodificationfactor RAA incentralityintervalscomparedwiththe

HG-Pythia model [63] scaled bytheisospinfactorsobtainedfromtheCT14NLO PDFcalculation.Theerrorbarsonthe datapointsindicate thestatistical uncer-taintiesandtheshadedboxesshowthetotalsystematicuncertainties.Themodel accountsforabiasedclassificationoftheeventgeometryforeventscontaininga hardprocess.Thebins areorderedincentralitypercentile, startingfromcentral eventsonthelefttowardsmoreperipheralon theright.The Z boson measure-mentextendstothemostperipheral80–100%centralityinterval.Thecomparisonis showninthe0–80%centralityintervalwherethedifferentsmearingoftheATLAS centralityestimator(EFCal

T )isfoundtohaveanegligibleeffect.

superimposingacorrespondingnumberof Pythia 6.4 [40] events. Themodelisabletoqualitativelyexplain theALICEmeasurement intheperipheralregion.

Fig.7showstheRAA forW± [6] and Z bosonscomparedwith theHG-Pythia model. To comparethe model withthe measure-mentsofmassiveelectroweakbosonstheresultsarecorrectedfor the isospin effect using the CT14 NLO PDF. All three data sets showtrendsthatareconsistentbetweenthespecies,butthat dif-ferfromtheircorrespondingmodelpredictions.Thissuggeststhat theapparent suppressionmechanism [63] thatexplainstheALICE data [62] fortheyieldsofhigh-pT chargedparticlesdoesnothave thesameeffectontheyieldsofmassiveelectroweakbosons.

7.Summary

The Z boson production yield per minimum-bias collision, scaled by the mean nuclear thickness function



TAA



isreported in Pb+Pb collisions at a nucleon–nucleon centre-of-mass energy

√

sNN

=

5

.

02 TeV.Themeasurementisbasedondatatakenbythe ATLAS detector at the LHC corresponding to an integrated lumi-nosityof0.49nb−1_{.Normalisedyieldsarereportedintheelectron} andmuon decaychannels, differentially in rapidity and collision centralityin themass window 66

<

m

<

116 GeV.The fiducial

regionisdefinedusingthelepton kinematicsanddetector accep-tance.Theelectronchannelandmuonchannelresultsarefoundto agreewithinthemeasurementprecisionandarecombinedforthe finalresult.

The normalised Z boson yields measured in Pb+Pb collisions are 1–3

σ

higher than NLO pQCD predictions withboth free and nuclearPDF sets,where thedifference increasestowardsforward

Z bosonrapidity.CalculationsusingnuclearPDFsetsdeviatefrom thedatamore stronglythan calculationsbasedonlyon theCT14 NLOPDFsetwhichisincontrastwithrecentEWboson measure-ments performed in the p+Pb system. The nuclear modification factoris measured differentiallyas a function of Z boson rapid-ityandeventcentrality.It isfoundtobeconsistent withunityin centralityandtoagreewiththepredictionbasedontheCT14PDF setthat takesisospininto account.This behaviouris also consis-tentwiththeATLASmeasurementperformedwithW bosons.The yieldratiosW

/

Z arefoundtobeconstantasafunctionof central-itywithinthe uncertainties ofthemeasurements. Unlikehigh-pT chargedhadronsmeasured by theALICECollaboration, W and Z

bosonsshownoindicationofyieldsuppressioninperipheral colli-sions.

Acknowledgements

We thank CERN forthe very successfuloperation of the LHC, aswell as thesupport staff fromour institutionswithout whom ATLAScouldnotbeoperatedefficiently.

WeacknowledgethesupportofANPCyT,Argentina;YerPhI, Ar-menia; ARC, Australia; BMWFW and FWF, Austria; ANAS, Azer-baijan; SSTC, Belarus; CNPq andFAPESP, Brazil; NSERC, NRC and CFI,Canada; CERN; CONICYT, Chile;CAS, MOSTandNSFC, China; COLCIENCIAS, Colombia; MSMT CR, MPO CR and VSC CR, Czech Republic;DNRFandDNSRC,Denmark;IN2P3-CNRS,CEA-DRF/IRFU, France; SRNSFG, Georgia; BMBF, HGF, and MPG, Germany; GSRT, Greece;RGC,Hong KongSAR, China;ISFandBenoziyoCenter, Is-rael; INFN, Italy; MEXT and JSPS, Japan; CNRST, Morocco; NWO, Netherlands; RCN,Norway;MNiSW andNCN, Poland;FCT, Portu-gal;MNE/IFA,Romania;MESofRussiaandNRCKI,Russian Feder-ation;JINR;MESTD,Serbia;MSSR,Slovakia;ARRSandMIZŠ, Slove-nia; DST/NRF, South Africa; MINECO, Spain; SRC andWallenberg Foundation,Sweden;SERI,SNSFandCantons ofBernandGeneva, Switzerland;MOST,Taiwan;TAEK, Turkey;STFC,UnitedKingdom; DOE and NSF, United States of America. In addition, individual groupsandmembershavereceivedsupportfromBCKDF,CANARIE, CRCandComputeCanada,Canada;COST,ERC,ERDF,Horizon2020, andMarie Skłodowska-CurieActions,European Union; Investisse-mentsd’AvenirLabexandIdex,ANR,France;DFGandAvH Foun-dation,Germany;Herakleitos,ThalesandAristeiaprogrammes co-financed by EU-ESF and the Greek NSRF, Greece; BSF-NSF and GIF,Israel;CERCAProgrammeGeneralitatdeCatalunya,Spain;The RoyalSocietyandLeverhulmeTrust,UnitedKingdom.

The crucial computingsupport from all WLCG partnersis ac-knowledged gratefully, in particular from CERN, the ATLAS Tier-1 facilities at TRIUMF (Canada), NDGF (Denmark, Norway, Swe-den),CC-IN2P3(France),KIT/GridKA(Germany),INFN-CNAF(Italy), NL-T1(Netherlands),PIC(Spain),ASGC(Taiwan),RAL(UK)andBNL (USA),theTier-2facilitiesworldwideandlargenon-WLCGresource providers.Majorcontributorsofcomputingresourcesare listedin Ref. [65].

References

[1]ATLASCollaboration,MeasurementofZ bosonproductioninPb+Pbcollisions at √sN N=2.76 TeV with the ATLAS detector, Phys. Rev.Lett. 110 (2013)

022301,arXiv:1210.6486 [hep-ex].

[2]ATLASCollaboration,Measurementoftheproductionandleptoncharge asym-metryofW bosonsinPb+Pbcollisionsat√sN N=2.76 TeVwiththeATLAS

detector,Eur.Phys.J.C75(2015)23,arXiv:1408.4674 [hep-ex].

[3]CMSCollaboration,StudyofZ bosonproductioninPbPbcollisionsat√sN N=

2.76TeV,Phys.Rev.Lett.106(2011)212301,arXiv:1102.5435 [nucl-ex]. [4]CMSCollaboration,StudyofZ productioninPbPbandppcollisionsat√sNN=

2.76 TeVinthedimuonanddielectrondecaychannels,J.HighEnergyPhys.03 (2015)022,arXiv:1410.4825 [nucl-ex].

[5]CMS_√ Collaboration,StudyofW bosonproductioninPbPbandpp collisionsat

sN N=2.76 TeV,Phys.Lett.B715(2012)66,arXiv:1205.6334 [nucl-ex].

[6]ATLAS Collaboration,MeasurementofW± bosonproduction inPb+Pb colli-sionsat √sNN=5.02 TeVwiththeATLASdetector,Eur.Phys.J.C79(2019) 935,arXiv:1907.10414 [nucl-ex].

[7]ATLAS Collaboration, Centrality, rapidityand transversemomentum depen-denceofisolatedpromptphotonproductioninlead-leadcollisionsat√sNN= 2.76 TeVmeasuredwiththeATLASdetector,Phys.Rev.C93(2016)034914, arXiv:1506.08552 [hep-ex].

[8]CMSCollaboration,Measurementofisolatedphotonproductioninpp andPbPb collisionsat√sN N=2.76 TeV,Phys.Lett.B710(2012)256,arXiv:1201.3093

[nucl-ex].

[9]ALICECollaboration,MeasurementofZ0_{-bosonproductionatlargerapidities} inPb-Pbcollisionsat√sNN=5.02 TeV,Phys.Lett.B780(2018)372,arXiv: 1711.10753 [nucl-ex].

[10]LHCbCollaboration,ObservationofZ productioninproton-leadcollisionsat LHCb,J.HighEnergyPhys.09(2014)030,arXiv:1406.2885 [hep-ex]. [11]ATLASCollaboration, Z bosonproductioninp+Pbcollisionsat√sN N=5.02

TeVmeasuredwiththeATLASdetector,Phys.Rev.C92(2015)044915,arXiv: 1507.06232 [hep-ex].

[12]CMSCollaboration,StudyofZ bosonproductioninpPbcollisionsat√sN N=

5.02 TeV,Phys.Lett.B759(2016)36,arXiv:1512.06461 [hep-ex].

[13]ALICECollaboration,W andZ bosonproductioninp−Pbcollisionsat√sNN= 5.02 TeV,J.HighEnergyPhys.02(2017)077,arXiv:1611.03002 [nucl-ex]. [14]CMSCollaboration,ObservationofnuclearmodificationsinW±boson

produc-tioninpPbcollisionsat√sNN=8.16 TeV,arXiv:1905.01486 [hep-ex],2019. [15]ATLASCollaboration,MeasurementsofW andZ bosonproductioninpp

colli-sionsat√s=5.02 TeVwiththeATLASdetector,Eur.Phys.J.C79(2019)128, arXiv:1810.08424 [hep-ex],Erratum:Eur.Phys.J.C79(2019)374.

[16]ATLASCollaboration,TheATLASexperimentattheCERNlargehadroncollider, J.Instrum.3(2008)S08003.

[17] ATLASCollaboration,ATLASinsertableB-layertechnicaldesignreport, ATLAS-TDR-19,https://cds.cern.ch/record/1291633,2010,Addendum:https://cds.cern. ch/record/1451888,2012.

[18]B.Abbott,etal.,ProductionandintegrationoftheATLASinsertableB-layer,J. Instrum.13(2018)T05008,arXiv:1803.00844 [physics.ins-det].

[19]ATLAS Collaboration,PerformanceoftheATLAS triggersystemin2015,Eur. Phys.J.C77(2017)317,arXiv:1611.09661 [hep-ex].

[20]ATLAS Collaboration, Performanceofelectronand photontriggersinATLAS duringLHCrun2,arXiv:1909.00761 [hep-ex],2019.

[21]ATLASCollaboration,Measurementofthejetradiusandtransversemomentum dependenceofinclusivejetsuppressioninlead-leadcollisionsat√sN N=2.76

TeVwiththe ATLASdetector,Phys.Lett.B719(2013)220,arXiv:1208.1967 [hep-ex].

[22]ATLASCollaboration,MuonreconstructionperformanceoftheATLASdetector inproton–protoncollisiondataat√s=13 TeV,Eur.Phys.J.C76(2016)292, arXiv:1603.05598 [hep-ex].

[23]ATLAS Collaboration,ElectronreconstructionandidentificationintheATLAS experimentusingthe2015and2016LHCproton-protoncollisiondataat√s= 13 TeV,Eur.Phys.J.C79(2019)639,arXiv:1902.04655 [physics.ins-det]. [24]M.L.Miller, K.Reygers,S.J. Sanders,P.Steinberg,Glaubermodelingin

high-energy nuclear collisions, Annu.Rev.Nucl. Part. Sci.57 (2007) 205,arXiv: nucl-ex/0701025.

[25]ATLASCollaboration,Promptandnon-prompt J/ψ andψ(2S)suppressionat hightransversemomentumin5.02 TeVPb+PbcollisionswiththeATLAS ex-periment,Eur.Phys.J.C78(2018)762,arXiv:1805.04077 [nucl-ex]. [26]ATLASCollaboration,Measurementofphoton–jettransversemomentum

cor-relationsin5.02TeVPb+Pband pp collisionswithATLAS,Phys. Lett.B789 (2019)167,arXiv:1809.07280 [nucl-ex].

[27]S.Agostinelli,etal.,GEANT4- asimulationtoolkit,Nucl.Instrum.MethodsA 506(2003)250.

[28]ATLAS Collaboration,TheATLAS simulationinfrastructure,Eur.Phys.J. C70 (2010)823,arXiv:1005.4568 [physics.ins-det].

[29]P.Nason,AnewmethodforcombiningNLOQCDwithshowerMonteCarlo algorithms,J.HighEnergyPhys.11(2004)040,arXiv:hep-ph/0409146 [hep -ph].

[30]S.Frixione,P.Nason,C.Oleari,MatchingNLOQCDcomputationswithparton showersimulations:thePOWHEGmethod,J.HighEnergyPhys.11(2007)070, arXiv:0709.2092 [hep-ph].

[31]S.Alioli,P.Nason,C.Oleari,E.Re,AgeneralframeworkforimplementingNLO calculationsinshowerMonteCarloprograms:thePOWHEGBOX,J.High En-ergyPhys.06(2010)043,arXiv:1002.2581 [hep-ph].

[32]S.Alioli,P.Nason,C.Oleari,E.Re,NLOvector-bosonproductionmatchedwith showerinPOWHEG,J.HighEnergyPhys.07(2008)060,arXiv:0805.4802 [hep -ph].

[33]T.Sjöstrand,S.Mrenna,P.Z.Skands,AbriefintroductiontoPYTHIA8.1,Comput. Phys.Commun.178(2008)852,arXiv:0710.3820 [hep-ph].

[34]J.Gao,etal.,CT10next-to-next-to-leadingorderglobalanalysisofQCD,Phys. Rev.D89(2014)033009,arXiv:1302.6246 [hep-ph].

[35]J.Pumplin, etal.,Newgeneration ofpartondistributionswith uncertainties fromglobalQCDanalysis,J.HighEnergyPhys.07(2002)012,arXiv:hep-ph/ 0201195 [hep-ph].

[36]ATLASCollaboration,Measurementofthe Z/γ∗bosontransversemomentum distributioninpp collisionsat√s=7 TeV withtheATLASdetector,J.High EnergyPhys.09(2014)145,arXiv:1406.3660 [hep-ex].

[37]N.Davidson,T.Przedzinski,Z.Was,PHOTOS interfaceinC++:technicaland physics documentation,Comput. Phys.Commun.199(2016)86, arXiv:1011. 0937 [hep-ph].

[38]S.Frixione,P.Nason,G.Ridolfi,Apositive-weightnext-to-leading-orderMonte Carloforheavyflavourhadroproduction,J.HighEnergyPhys.09(2007)126, arXiv:0707.3088 [hep-ph].

[39]H.-L.Lai,etal.,Newpartondistributionsforcolliderphysics,Phys.Rev.D82 (2010)074024,arXiv:1007.2241 [hep-ph].

[40]T.Sjöstrand, S.Mrenna, P.Z.Skands,PYTHIA6.4physicsandmanual,J.High EnergyPhys.05(2006)026,arXiv:hep-ph/0603175 [hep-ph].

[41]P.Z.Skands,TuningMonteCarlogenerators:thePerugiatunes,Phys.Rev.D82 (2010)074018,arXiv:1005.3457 [hep-ph].

[42]D.J.Lange,TheEvtGenparticledecaysimulationpackage,Nucl.Instrum. Meth-odsA462(2001)152.

[43]ATLASCollaboration,ElectronandphotonenergycalibrationwiththeATLAS detector using 2015-2016 LHC proton-protoncollision data, J. Instrum. 14 (2019)P03017,arXiv:1812.03848 [hep-ex].

[44]C.A.Bertulani,G.Baur,Electromagneticprocessesinrelativisticheavyion col-lisions,Phys.Rep.163(1988)299.

[45]F.Krauss,M.Greiner,G.Soff,Photonandgluoninducedprocessesinrelativistic heavyioncollisions,Prog.Part.Nucl.Phys.39(1997)503.

[46]G.Baur, Coherent photon-photoninteractions invery peripheralrelativistic heavyioncollisions,Eur.Phys.J.D55(2009)265,arXiv:0810.1400 [nucl-th]. [47]V.P.Goncalves,M.V.T.Machado,Diffractivephotoproductionof Z0 _bosonsin

coherentinteractions atCERN-LHC,Eur.Phys.J.C56(2008)33, arXiv:0710. 4287 [hep-ph],Erratum:Eur.Phys.J.C61(2009)351.

[48]ATLAS Collaboration, Measurement ofexclusive γ γ→ +_− _{production in} proton-protoncollisions at√s=7 TeVwiththe ATLASdetector, Phys.Lett. B749(2015)242,arXiv:1506.07098 [hep-ex].

[49]CMS Collaboration, Exclusive photon-photon production of muon pairs in proton-protoncollisionsat √s=7 TeV,J.HighEnergyPhys.01(2012)052, arXiv:1111.5536 [hep-ex].

[50]ATLASCollaboration,RapiditygapcrosssectionsmeasuredwiththeATLAS de-tectorinpp collisionsat √s=7 TeV,Eur.Phys.J.C72(2012)1926,arXiv: 1201.2808 [hep-ex].

[51]ATLAS Collaboration, Observation of centrality-dependent acoplanarity for muonpairsproducedviatwo-photonscatteringinPb+Pbcollisionsat√sNN= 5.02 TeVwiththeATLASdetector,Phys.Rev.Lett.121(2018)212301,arXiv: 1806.08708 [nucl-ex].

[52]A.Valassi,Combiningcorrelatedmeasurementsofseveraldifferentphysical quantities,Nucl.Instrum.MethodsA500(2003)391.

[53]S.Catani,M.Grazzini,Next-to-next-to-leadingordersubtractionformalismin hadroncollisionsanditsapplicationtoHiggsbosonproductionattheLarge Hadron Collider, Phys. Rev. Lett. 98 (2007) 222002, arXiv:hep-ph/0703012 [hep-ph].

[54]S.Catani,L.Cieri,G.Ferrera,D.deFlorian,M.Grazzini,Vectorbosonproduction athadroncolliders:afullyexclusiveQCDcalculationatnext-to-next-to-leading order,Phys.Rev.Lett.103(2009)082001,arXiv:0903.2120 [hep-ph]. [55]W.F.L.Hollik,Radiativecorrectionsinthestandardmodelandtheir rôlefor

precisiontestsoftheelectroweaktheory,Fortschr.Phys.38(1990)165. [56]ParticleDataGroup,K.A.Olive,etal.,Reviewofparticlephysics,Chin.Phys.C

38(2014)090001.

[57]S.Dulat,et al.,New partondistributionfunctionsfroma globalanalysisof quantumchromodynamics,Phys.Rev.D93(2016)033006,arXiv:1506.07443 [hep-ph].

[58]K.Kovarik,etal.,nCTEQ15- globalanalysisofnuclearpartondistributionswith uncertaintiesintheCTEQframework,Phys.Rev.D93(2016)085037,arXiv: 1509.00792 [hep-ph].

[59]K.J.Eskola,P.Paakkinen,H.Paukkunen,C.A.Salgado,EPPS16:nuclearparton distributionswithLHCdata,Eur.Phys.J.C77(2017)163,arXiv:1612.05741 [hep-ph].

[60]J.M.Campbell,R.K.Ellis,W.T.Giele,Amulti-threadedversionofMCFM,Eur. Phys.J.C75(2015)246,arXiv:1503.06182 [physics.comp-ph].

[61]C.Loizides,J.Kamin,D.d’Enterria,ImprovedMonteCarloGlauberpredictions atpresentandfuturenuclearcolliders,Phys.Rev.C97(2018)054910,arXiv: 1710.07098 [nucl-ex],Erratum:Phys.Rev.C99(2019)019901.

[62]ALICECollaboration,Analysisoftheapparentnuclearmodificationin periph-eralPb-Pbcollisionsat 5.02TeV,Phys. Lett.B793(2019)420,arXiv:1805. 05212 [nucl-ex].

[63]C.Loizides,A.Morsch,Absenceofjetquenchinginperipheralnucleus–nucleus collisions,Phys.Lett.B773(2017)408,arXiv:1705.08856 [nucl-ex]. [64]X.-N.Wang,M.Gyulassy,HIJING:aMonteCarlomodelformultiplejet

produc-tioninpp,pAandAAcollisions,Phys.Rev.D44(1991)3501.

[65] ATLAS Collaboration, ATLAS computing acknowledgements, ATL-GEN-PUB-2016-002,https://cds.cern.ch/record/2202407.