Contents lists available atScienceDirect
Physics Letters B
www.elsevier.com/locate/physletb
Search for resonances in the mass distribution of jet pairs with one or two jets identified as b-jets in proton–proton collisions at √
s = 13 TeV with the ATLAS detector
.TheATLASCollaboration
a r t i c l e i n f o a b s t ra c t
Articlehistory:
Received29March2016
Receivedinrevisedform13May2016 Accepted20May2016
Availableonline26May2016 Editor:W.-D.Schlatter
Searches forhigh-mass resonancesin thedijet invariant mass spectrumwith oneortwo jetsidenti- fiedasb-jetsareperformedusinganintegratedluminosityof3.2 fb−1ofproton–protoncollisionswith acentre-of-massenergyof√s
=13 TeV recordedbythe ATLASdetectorattheLargeHadronCollider.
No evidenceofanomalousphenomenaisobservedinthedata,whichareusedtoexclude,at95% cred- ibility level, excitedb∗ quarks with massesfrom 1.1 TeV to 2.1 TeV andleptophobic Z bosons with massesfrom1.1 TeV to1.5 TeV.ContributionsofaGaussian signalshape witheffectivecross sections rangingfromapproximately0.4to0.001 pbarealsoexcludedinthemassrange1.5–5.0 TeV.
©2016TheAuthor.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense (http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP3.
1. Introduction
Manyextensionsto theStandardModel(SM)predict theexis- tenceofnewmassiveparticlesthatcoupletoquarksorgluons.If producedinproton–proton(pp)collisionsattheLargeHadronCol- lider(LHC),thesenewbeyond-the-SM(BSM)particlescoulddecay into quarks (q) or gluons (g), creating resonant excesses in the two-jet (dijet) invariant massdistributions [1–6].If thenew par- ticlecouples to the b-quark and decays into bb,¯ bq or bg pairs, a dedicatedsearchfordijetresonanceswithoneorbothjetsiden- tifiedasoriginatingfromab-quark(“b-jet”)couldgreatlyincrease thesignalsensitivity.
Priorresonancesearchesindijeteventscontainingb-jets were performed by the CDF [7] and CMS [8,9] experiments, probing the mass ranges 200–750 GeV and 1–4 TeV respectively. Excited heavy-flavourquarks have been investigated in alternative decay modesaswell [10].No BSMphenomenahavebeenobservedyet.
Theincrease in centre-of-massenergyof the pp collisions atthe LHC from √
s=7 and 8 TeV to 13 TeV provides a new energy regime in which to search for such a heavy resonance. This is particularly true for heavy states coupling to b-quarks from the protonsea,whencomparedtostatesproducedbyvalencequarks.
Thepartonluminositytocreatea2 TeV objectincreasesbyanad- ditionalfactorof 2–3forbb and¯ bg overqq and¯ qg pairs,when increasingthecentre-of-massenergyfrom8 TeV to13 TeV.Theto- talproductionratefordijetBSMsignalscanbecomelargeenough
E-mailaddress:atlas.publications@cern.ch.
toallowagoodsignalsensitivityevenwitharelativelysmalldata sample. Inthispaperthesearch foranewnarrowresonancede- cayingto b-quarks withthe ATLAS detector, using 3.2 fb−1 inte- grated luminosity of proton–proton collisions at √
s=13 TeV, is reported.Themassrange1.1–5.0 TeV isprobed.
The results are interpreted in the context of two benchmark processesshowninFig. 1: anexcited heavy-flavourquark b∗ and anewgaugeboson Z.Excitedquarksareaconsequenceofquark compositeness modelsthat wereproposed to explainthe genera- tionalstructure andmasshierarchyofquarks[11,12]. The Z bo- son arisesinmanyextensionstothe SMwithan additionalU(1) group. Two Z models are considered, one with SM-like fermion couplingsintheSequentialStandardModel(SSM)andaleptopho- bic Zmodel[13,14].Allbenchmarkmodeldecaysareexpectedto resultina narrowresonancesuperimposed onasmoothly falling dijet invariant mass distribution. This search divides the events into samples with one or two jets identified as b-jets to en- hancethesignalsensitivitytothebenchmarkmodelsb∗→bg and Z→bb.¯ Inaddition,theresultsare interpretedinthe contextof possibleGaussian-shapedsignalcontributionstothedijetinvariant mass spectrawhereone orboth jetsare identified asb-jets.The results,presented in termsof the crosssection times acceptance timesbranching ratio (σ×A×BR),are quoted forcontributions withwidthsofupto15% oftheresonancemass.
2. TheATLASdetector
The ATLAS experiment [15] at the LHC is a multi-purpose particle detector with a forward–backward symmetric cylindrical http://dx.doi.org/10.1016/j.physletb.2016.05.064
0370-2693/©2016TheAuthor.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).Fundedby SCOAP3.
Fig. 1. Leading-order Feynman diagrams for the two processes considered: gb→b∗→bg and qq¯→Z→bb.¯
geometry and a near 4π coverage in solid angle.1 It consists of an inner tracking detector surrounded by a thin superconduct- ingsolenoidprovidinga 2Taxialmagneticfield,electromagnetic and hadronic calorimeters, and a muon spectrometer. The inner trackingdetectorcoversthepseudorapidityrange|η|<2.5.Itcon- sists of,in ascending order ofradius from the beam-line, silicon pixel, silicon microstrip, and transition radiation tracking detec- tors.Thepixeldetectorsarecrucialforb-jetidentification.Forthe second LHC data-taking period, a new inner pixel layer, the In- sertableB-Layer(IBL)[16,17],was addedatameansensorradius of 3.2 cm from the beam-line. Lead/liquid-argon (LAr) sampling calorimeters provide electromagnetic (EM) energy measurements withhighgranularity.A hadron(steel/scintillator-tile) calorimeter covers the central pseudorapidity range (|η|<1.7). The end-cap and forward regions are instrumented withLAr calorimeters for EMandhadronicenergymeasurementsupto |η|=4.9.Thefirst- leveltriggerisimplementedinhardwareandusesasubsetofthe detector information to reduce the input rate from the nominal LHC collision rate to an acceptancerate of 100 kHz. This is fol- lowedbyasoftware-basedtriggerthatreducestherateofevents recordedto1kHz.
3. Dataandsimulatedeventsamples
ThedatausedinthisanalysiswerecollectedbytheATLASde- tector in pp collisions at the LHC with a centre-of-mass energy of 13 TeV during 2015. Events were recorded using a jet-based triggerrequiringatleastonejetwithatransversemomentum pT ofatleast360 GeV.Thefulldatasetcorresponds toan integrated luminosity of3.2 fb−1 withan associateduncertainty of5%after applyingquality criteriatothe data.Themeasurement ofthe in- tegratedluminosityisderived,followingamethodologysimilarto thatdetailedinRef.[18],fromacalibrationoftheluminosityscale usingapairofx– y beam-separationscans.
MonteCarlo(MC)simulatedeventsamplesare usedto model theexpectedsignalsandstudythecompositionofSMbackground processes. The QCD dijet process is simulated with Pythia8 [19]
usingthe A14tuned parameter set[20] forthe modellingofthe partonshower, hadronization andunderlying event. The leading- order(LO) partondistributionfunction(PDF)setNNPDF2.3[21] is usedforthegenerationofevents.Therenormalizationandfactor- izationscales are setto theaverage transversemomentum pT of the two leading jets. The EvtGen decay package [22] is used for bottomandcharmhadrondecays.
Thethreesignalsamplesaregeneratedwith Pythia8 usingthe A14set of tuned parameters andthe NNPDF2.3PDF set. Forthe b∗ model,thecompositenessscaleissettotheexcited-quarkmass
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).Angulardistanceismeasuredinunitsof
R≡
(η)2+ (φ)2.
and 85% ofdecays are tobg. The remaining decay modesare to a SMgauge boson( Z boson,W boson orphoton)andab-quark.
IntheSSM Zmodel,the Z bosonhasthesamecouplingstoSM fermionsastheSM Z bosonandthebottomquarkdecaybranch- ing ratio BR(Z→bb¯) is13.8%. Theleptophobic Z model differs byhavingvanishingcouplingstoleptons.Thecorrespondingvalue ofBR(Z→bb¯)is18.9%.Forboth,onlydecaystob-quarkpairsare simulated.Theintrinsicdecaywidthis ∼0.6% oftheresonance mass for the b∗ model and ∼3% of the mass forthe SSM Z boson.
The generatedsamplesare processedwiththe ATLASdetector simulation [23],which is basedon the GEANT4 package [24].To account foradditional pp interactions fromthesame orclose-by bunch crossings, a number of minimum-bias interactions gener- ated using Pythia8 and the MSTW2008LO PDF [25] set are su- perimposedonto thehard scatteringevents.The MCsamples are re-weighted tomatch thecollisions per bunch crossing observed inthedata.
4. Eventreconstructionandselection
Jets are reconstructed fromnoise-suppressed topological clus- ters[26]ofenergydepositedinthecalorimetersusingtheanti-kt algorithm[27]witharadiusparameterof0.4.Jetenergiesanddi- rections are corrected by the jet calibrations derived from√
s= 13 TeV simulation,and pp collisiondatatakenat√
s=8 TeV and
√s=13 TeV, asdescribed inRef. [28]. Jetsare required to have pT>50 GeV.EventswhereanyofthethreeleadingjetswithpT>
50 GeV is compatiblewithnon-collision backgroundorcalorime- ternoiseareremoved. Eventsarepreselectedinthesamewayas inthedijetanalysisofRef.[5],requiringthatthe pToftheleading jetisgreaterthan440 GeV toensurefulltriggerefficiency.Anad- ditionalrequirementisplacedonthejetpseudorapidity,|η|<2.4, to ensure trackercoverage forb-jet identification. The analysisis performed in an unbiased dijet mass range of mjj>1.1 TeV. To reduce thebackgroundfromQCD multijetprocessesandenhance s-channelprocesses,therapidity difference y∗= (y1−y2)/2 be- tween the two leading jets isrequired tobe |y∗|<0.6. Here y1 and y2 are the rapidities of the leading and sub-leading jet re- spectively.
To identifyjetsoriginating fromb-hadrons (b-tagging)a mul- tivariate algorithm that combines information about the impact parameters of inner detector tracks associated with the jet, the presence of displaced secondary vertices, and the reconstructed flight paths of b- and c-hadrons associated withthe jet [29,30]
is employed.The b-taggingworkingpoint with85% efficiency,as determined when integrating over all jetsin a simulatedsample oftt events,¯ ischosen becauseitgivesthehighestsignalsensitiv- ity.Astheaveragejetenergiesinthisanalysisarelargerthanint¯t events andthe b-taggingefficiencydrops withjet pT, theper-jet efficiencies arebelow85% andareroughly 50% forjetswitha pT of1 TeV.
Theb-jetidentificationalgorithmisappliedtothetwoleading jets, andeventsare categorizedasinclusive, single b-tagged“1b”
ordoubleb-tagged“2b”,inordertoenhancethesensitivityofdif-
Fig. 2. Theper-eventb-taggingefficienciesaftertheeventselectionasafunctionof thereconstructedinvariantmassforsimulatedsampleswithsixdifferentb∗andZ resonancemasses.
ferentsignalcompositions.The“1b”categoryisdefinedinclusively, includingeventsfromthe“2b”category.
Theper-event b-taggingefficiencies asfunctionsoftherecon- structed invariant mass are shown in Fig. 2. Efficiencies are for benchmark models with different b∗ and Z resonance masses, afterthe eventselection is applied. The tagging efficiencyfor Z eventsintheinclusive“1b”category ishigherthanforb∗ events becausethisprocesshasmoreb-quarksinthefinalstate.Athigh mass,thegluon fromthe decayof theb∗ hasa higherprobabil- itytoproduceabb-pair,¯ whichcausestheeventtaggingefficiency tobecomparableforthe Z andb∗.Thetaggingefficiencyinthe
“2b”categoryisabout2.5 timesloweratlowmassandafactor 10 lower at high mass compared to the inclusive “1b” category for thesame Z events.The average light-flavour jet rejectionfactor forjetspassingthekinematicselectionisapproximately30forjet transversemomentaupto∼1 TeV.
Correction factors are applied to the simulated event sam- ples to compensate for differencesbetween data and simulation in b-tagging efficiencies and mis-identification rates. These cor- rections were derived from comparisons of samples of b-quark- enrichedeventsindataandsimulation[31].Theaveragecombined signal acceptanceandefficiencyis around 20%for theb∗ bench- markin the “1b” category and drops withincreasing massfrom 9%at1.5 TeV to2%at5.0 TeV forthe Zsignalsforthe“2b”cate- gory.
5. Dijetmassspectrum
The dijet mass spectrum is predominantly composed of jets arising from QCD interactions. Fig. 3 shows the comparison be- tween data and Pythia8 multijet MC simulation. The simulated distributionsarenormalizedtothe numberofeventsobservedin the data in each category separately. The bin widths are chosen toapproximatethemjj resolutionasderivedfromsimulatedQCD processes,whichrangefrom3% at1.0 TeV to2% at5.0 TeV.Good agreementbetweentheshapesofthe Pythia8 multijetpredictions andthedataisfound. Theinclusivedistribution,notrestrictedin theinnertrackingdetectoracceptance,wasanalysedinRef.[5].
Thedijet backgroundestimation doesnot rely onthe simula- tionasitisobtaineddirectlyfromafittothemjjdistribution.The followingparameterizationansatzisadoptedtofitthedistribution inthemjjrangefrom1.1 TeV up tothelastdatapointofthein- clusive,“1b”and“2b”massdistributionsseparately,
f(z)=p1(1−z)p2zp3, (1)
Fig. 3. Theinvariantmassdistributionoftheinclusivedijet(dots),“1b”(squares) and “2b” (triangles) categories in data. The inclusive distribution is similar to Ref.[5],butanadditionalrequirementisplacedonthejetpseudorapidity,|η|<2.4.
TheMCdistributionsarenormalizedtothedatainthethreecategoriesseparately:
asolidlineforinclusivedijets,adashedlinefor“1b”andasmalldashedlinefor
“2b”categories.ThelowerpanelsshowtheratiobetweendataandMCsimulation forallthreecategories.
wherepiarefreeparametersandz=mjj/√
s.Thisansatzwasused inprevious searches [5]andisfound toprovidea satisfactory fit to leading-order Pythia8 multijetMC simulationat √
s=13 TeV.
EmployingWilks’theorem[32],alog-likelihoodstatisticisusedto confirmthat noadditionalparametersareneededtomodelthese distributionsforadatasetaslargeastheoneusedforthisanaly- sis.
TheresultsofthefitsareshowninFig. 4.Thefitsofthisansatz to the data without considering systematic uncertainties return p-values of 0.73, 0.90 and 0.66 for the inclusive, “1b” and“2b”
categoriesrespectively.The p-valuewascalculatedasagoodness- of-fitmeasure usinga χ2 test statistic determined frompseudo- experiments.
ThelowerpanelsofFig. 4showthesignificancesofbin-by-bin differencesbetweenthedataandthefit.TheseequivalentGaussian significancesarecalculatedfromthePoissonprobability,consider- ingonlystatisticaluncertainties.
The statistical significance of any localized excess in the di- jet mass distribution is quantified using the BumpHunter algo- rithm[33].Thealgorithmcomparesthebinnedmjjdistributionof thedatatothefittedbackgroundestimate,consideringcontiguous massintervalsinallpossiblelocations,fromawidthoftwobinsto one-halfofthedistribution.Foreachinterval inthe scan,itcom- putesthesignificanceofanyexcessfound.Thealgorithmidentifies the intervals 1493–1614 GeV in the “1b” and3596–3827 GeV in the “2b” sample, indicated by the two vertical lines in Fig. 4, as the mostdiscrepantintervals. The statisticalsignificance ofthese
Fig. 4. Dijetmassspectraoverlaidwiththefitstothebackgroundfunctiontogetherwiththeresultsfrom BumpHunter andbenchmarksignalsscaledbyafactorof50.The mostdiscrepantregionisindicatedbythetwobluelines.Thelowerpanelsshowthesignificancesperbinofthedatawithrespecttothebackgroundfit,intermsofthe numberofstandarddeviations,consideringonlythestatisticalfluctuations.Thedistributionsareshownforthe(a)“1b”and(b)“2b”categories.(Forinterpretationofthe referencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
outcomes is evaluated using the ensemble of Poisson outcomes across all intervals scanned, by applying the algorithm to many pseudo-data samples drawn randomly from the background fit.
Without including systematic uncertainties, the probability that fluctuationsof thebackground modelwould produceexcesses at leastassignificantasthoseobservedinthedata,anywhereinthe distribution,is greater than 60% in the “1b” and“2b” categories.
Thus, thereis noevidence oflocalizedcontributions to themass distributionfromBSMphenomena.
6. Systematicuncertainties
Uncertaintiesintheparametersofthefittedbackgroundfunc- tion Eq. (1) are evaluated by fitting the ansatz to pseudo-data drawn via Poisson fluctuations around the fitted background model.Theuncertainty inthepredictionineach mjj binis taken tobe therootmean square ofthefunction value for10000 gen- eratedpseudo-experiments.Toestimatean uncertaintyduetothe choice of background parameterization, one additional degree of freedom, zp4log(z), is appended as a multiplicative factor to the nominalansatz(Eq.(1)),andthedifferencebetweentheestimated parametersfromthetwofitsistakenasanuncertainty.
Theuncertaintyinthejetenergyscaleisestimatedusingvari- ous methods in8 TeV data, corrected to the newcentre-of-mass energy by taking the difference between the 8 TeV and 13 TeV runsintoaccount using MCsimulation [28]. Thejet energyscale uncertaintyusedinthisanalysisreliesonasetofthreenuisance parameters[34].Foruntaggedjetsitiswithintherange1–5%for jettransversemomentagreaterthan200 GeV.
The relative additional uncertainty in the energy scale of b-taggedjetsisestimatedusingtheMCsamplesandverifiedwith datafollowingthemethoddescribedinRef.[35].Theratiortrk of the sum oftrack transverse momenta inside the jet to the total jettransversemomentummeasuredinthecalorimeterisusedfor thisestimate. Thedoubleratioofrtrk fromdataandsimulationis formedandcomparedforinclusivejetsandb-jets.The estimated
relativeadditionaluncertaintyforjetswith200<pT<800 GeV is foundtobelessthan2.6%,andthisvalueissubsequentlyusedin the higher pT regions. This relative uncertaintyis applied in ad- ditionto thenominaljetenergyscaleuncertainty. Themaximum uncertaintyforb-taggedjetsisestimatedto be6% andisconser- vativelyappliedtoall pTregions.
Theuncertaintyinthejetenergyresolutionisestimatedusing thesamemethodastheuntaggedjetenergyscaleuncertaintyand reliesonanadditionalGaussiansmearingofthereconstructedjet energies inMC simulation.Forjetswith pT>50 GeV,theuncer- taintyislessthan2%.
Theuncertaintyintroduced bytheapplicationoftheb-tagging algorithmisthelargestsystematicuncertaintyintheanalysis.The uncertainty in the measured tagging efficiency of b-jets is esti- mated by studying tt events¯ in 13 TeV data for jet pT up to 200 GeV [31].The uncertainties in the measured rateof mistag- gingc-jets andlight-flavourjetsare estimatedin8 TeV data.The uncertainties are extrapolatedto 13 TeV, takinginto account the additionofthenewIBLsystemaswellasreconstructionandtag- gingimprovements.An additionaltermisincludedtoextrapolate themeasured uncertaintiestothehigh-pT regionofinterest.This term is calculated from simulated events by considering varia- tions on the quantitiesaffecting the b-tagging performance such astheimpactparameterresolution,percentageofpoorlymeasured tracks, descriptionofthedetectormaterial,andtrackmultiplicity per jet. Thedominanteffecton theuncertaintywhen extrapolat- ing at high-pT isrelated tothe differenttagging efficiencywhen smearing the tracks impact parameters based on the resolution measuredindataandsimulation.Thedifferenceintheimpactpa- rameterresolutionisduetoeffectsfromalignment,deadmodules and additionalmaterial not properly modelled inthe simulation.
The impact oftheb-tagging efficiencyuncertaintyincreaseswith jet pT andreaches50%above2 TeV.
