a r t i c l e i n f o a b s t ra c t
Articlehistory:
Received4April2018
Receivedinrevisedform15June2018 Accepted4July2018
Availableonline11July2018 Editor:W.-D.Schlatter
AsearchforaheavyneutralHiggsboson,A,decayingintoaZbosonandanotherheavyHiggsboson,H, isperformedusingadatasamplecorrespondingtoanintegratedluminosityof36.1 fb−1fromproton–
protoncollisionsat√s
=13 TeV recordedin2015and2016bytheATLASdetectorattheLargeHadron Collider. The search considersthe Z boson decaying to electrons ormuons and the H boson intoa pairofb-quarks.Noevidencefortheproductionofan Abosonisfound.Considering eachproduction processseparately,the95%confidence-levelupperlimitsonthepp→A→Z Hproductioncross-section times thebranching ratio H→bb are inthe rangeof 14–830 fbfor thegluon–gluonfusion process and 26–570 fb for the b-associated process for the mass ranges 130–700 GeV of the H bosonand 230–800 GeV ofthe Aboson.Theresultsareinterpretedinthecontextoftwo-Higgs-doubletmodels.
©2018PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense (http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP3.
1. Introduction
Afterthediscovery ofaHiggsboson atthe LargeHadronCol- lider (LHC) [1,2], one ofthe mostimportantremaining questions iswhethertherecentlydiscoveredparticleispartofanextended scalar sector or not. Additional Higgs bosons appear in all mod- elswithanextendedscalarsector,such asthetwo-Higgs-doublet model (2HDM) [3,4]. Such extensions are motivated by, and in- cluded in, several new physics scenarios, such as supersymme- try [5], darkmatter [6] and axion [7] models,electroweakbaryo- genesis [8] andneutrinomassmodels [9].
The addition of a second Higgs doublet leads to five Higgs bosonsafterelectroweaksymmetrybreaking.Thephenomenology of such a model is very rich anddepends on many parameters, such as the ratio of the vacuum expectation values of the two Higgsdoublets(tanβ),andtheYukawacouplingsofthescalarsec- tor [4].WhenCPconservationisassumed,themodelcontainstwo CP-even Higgs bosons, h and H with mH >mh, one CP-odd, A, andtwochargedscalars, H±.Therehavebeenmanysearchesfor theheavy neutralHiggs bosonsof the2HDMattheLHC, includ- ing H→W W/Z Z [10–13], A/H→τ τ/bb [14–16], A→Zh [17, 18] and H→hh [19,20].For theinterpretation ofthesesearches itis usually assumedthat theheavy Higgsbosons, H and A,are degenerateinmass,i.e.mA=mH.
E-mailaddress:atlas.publications@cern.ch.
This assumption of mass degeneracy is relaxed in this Letter by assuming mA>mH. Such a case ismotivated by electroweak baryogenesis scenarios in the context of the 2HDM [21–24]. For 2HDM electroweakbaryogenesis to occur, the requirementmA>
mH isfavoured[21] forastrongfirst-orderphasetransitiontotake place in the early universe. The A boson mass is also bounded from above tobe less thanapproximately 800 GeV, whereas the lighterCP-evenHiggsboson,h,isrequiredtohavepropertiessimi- lartothoseofaStandardModel(SM)Higgsbosonandisassumed tobe theHiggsbosonwithmassof125 GeV thatwas discovered atthe LHC [21].Under suchconditionsandforlarge partsofthe 2HDM parameter space, the CP-odd Higgs boson, A, decays into Z H [25,21].Theproductionofthe A bosonintherelevant2HDM parameterspaceproceedsmainlythroughgluon–gluonfusionand b-associatedproductionattheLHC.
This search for A→Z H decays uses proton–proton collision dataat√
s=13 TeV correspondingtoanintegratedluminosityof 36.1 fb−1 recordedby theATLAS detectoratthe LHC.The search considers only Z →, where =e, μ, to take advantage of the clean leptonic final state, and H→bb, because of its large branching ratio. This final state allows full reconstruction of the A boson’s decay kinematics.The reconstruction of the A boson’s invariant massuses theassumedvalue ofthemassof the H bo- son to improve its resolution. The final state is also categorised by thepresence of two or three b-taggedjets to take advantage oftheb-associatedproductionmechanism.TheCMSCollaboration haspublished asimilarsearch at√
s=8 TeV [26].ThisLetterre- ports the result of a search at √
s=13 TeV, which extends the
https://doi.org/10.1016/j.physletb.2018.07.006
0370-2693/©2018PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP3.
previous search by considering explicitly the gluon–gluon fusion andb-associatedproductionprocessesaswellasbothnarrowand widewidthsofthe Aboson.
2. ATLASdetector
TheATLAS detector isa general-purpose particle detector,de- scribed in detail in Ref. [27]. It includes an inner detector sur- roundedby a 2 T superconducting solenoid, electromagneticand hadronic calorimeters and a muon spectrometer with a toroidal magneticfield.Theinnerdetectorconsistsofahigh-granularitysil- iconpixeldetector,includingtheinsertableB-layer [28] installedin 2014,asiliconmicrostripdetector,andastraw-tubetracker.Itpro- videsprecision tracking ofcharged particles with pseudorapidity
|η|<2.5.1 Thecalorimetersystemcoversthepseudorapidityrange
|η|<4.9.Itiscomposedofsamplingcalorimeterswitheitherliq- uid argon or scintillator tiles as the active medium. The muon spectrometer provides muon identification and measurement for
|η|<2.7. A two-level trigger system [29] is employed to select eventsforofflineanalysis,whichreducedtheaveragerecordedcol- lisionratetoabout1 kHz.
3. Dataandsimulation
The data used inthis search were collected during 2015 and 2016from √
s=13TeV proton–proton collisions and correspond toanintegratedluminosityof36.1 fb−1,whichincludesonlydata- takingperiods whereallrelevant detectorsubsystems were oper- ational.Thedatasamplewas collectedusingasetofsingle-muon andsingle-electrontriggers. The lowest-pT triggerthresholds de- pendonthedata-takingperiodandareintherangeof20–26 GeV forthesingle-muontriggersand24–26 GeV forthesingle-electron triggers.
Simulated signal events with A bosons produced by gluon–
gluon fusion were generated at leading order with MadGraph5_aMC@NLO2.3.3[30,31] usingPythia8.210 [32] with a set of tuned parameters called the A14 tune [33] for parton showering. For the generation of A bosons produced in associ- ation with b-quarks, MadGraph5_aMC@NLO 2.1.2 [31,34,35] was usedfollowing Ref. [36] together withPythia 8.212 andtheA14 tune for parton showering. The gluon–gluon fusion production usedNNPDF2.3LO [37] asthepartondistributionfunctions (PDF), whiletheb-associatedproductionusedCT10nlo_nf4 [38].Thesig- nalsamplesweregeneratedforAbosonswithmassesintherange of230–800 GeV andwidthsupto20%ofthemassandfornarrow- widthHbosonswithmassesintherangeof130–700 GeV.
Background events from the production of W and Z bosons in association with jets were simulated with Sherpa 2.2.1 [39]
using the NNPDF3.0NNLO PDF set [40]. Top-quark-pair produc- tionwassimulatedwithPowheg-Boxv2 [41–43] andtheCT10nlo PDF set [38], while the electroweak single-top-quark production was simulated with Powheg-Box v1 and the fixed four-flavour PDFset CT10nlo_f4 [38].The partonshower was performedwith Pythia 6.428 [44] using the Perugia 2012 set of tuned parame- ters [45]. The production of top-quark pairs in association with avector boson was simulatedusing MadGraph5_aMC@NLO 2.2.3 and the NNPDF3.0NLO PDF set, whereas Pythia 8.186 was used
1 ATLASusesaright-handedcoordinatesystemwithitsoriginatthe nominal interactionpoint(IP)inthecentreofthedetectorandthez-axisalongthebeam pipe.Thex-axispointsfromtheIPtothecentreoftheLHCring,andthe y-axis pointsupward.Cylindricalcoordinates(r,φ)areusedinthetransverseplane,φ beingtheazimuthalanglearoundthebeampipe.Thepseudorapidityisdefinedin termsofthepolarangle,θ,asη= −ln tan(θ/2).Transversemomentaarecomputed fromthethree-momenta,p, aspT= |p|sinθ.
for the parton shower with the A14 tune. Production of W W, Z Z and W Z pairs was simulated using Sherpa 2.2.1 and the NNPDF3.0NNLOPDFset.Finally,SMHiggsbosonproductioninas- sociationwitha Z bosonwasgeneratedwithPowheg-Boxv2and the NNPDF3.0NLO PDF set, whereas the partonshower was per- formedwithPythia8.186usingtheAZNLOtune [46].
The modellingof bottom- and charm-hadron decays was per- formedwiththeEvtGen v1.2.0package [47] forall samplesapart from those simulated with Sherpa. The simulated events were overlaid withinelasticproton–protoncollisionsto accountforthe effect of multiple interactions occurring in the same and neigh- bouring bunch crossings (‘pile-up’). These eventswere generated usingPythia 8withthe A2tune [48] and theMSTW2008LOPDF set [49].Theeventswerereweightedsothatthedistributionofthe averagenumberofinteractionsperbunchcrossingagreedwiththe data.
All generated background samples were passed through the Geant4-based [50] detectorsimulation [51] oftheATLASdetector.
TheATLFAST2 simulation [51] was usedforthesignalsamplesto allowforthegenerationofmanydifferentA andHbosonmasses.
The simulatedeventswerereconstructed inthesamewayasthe data.
4. Objectreconstruction
Electrons are reconstructed from energy clusters in the elec- tromagnetic calorimeter that are matched to tracks in the in- ner detector [52]. Electrons are required to have |η|<2.47 and pT>7 GeV. Todistinguishelectronsfromjets,isolationandqual- ityrequirementsareapplied [53].The isolationrequirements(the
‘LooseTrackOnly’ working point) are defined by the pT of tracks within cones aroundthe electronwith asize that decreasesasa function of the transverse energy. The quality requirements (the
‘Loose’ workingpoint)refer to both theinner detector trackand the calorimeter shower shape. The efficiency for an electron to bereconstructedandmeetthesecriteriaisabout85%forelectron pT>7 GeV andincreasestoabout90%forpT>27 GeV.
Muons are reconstructed by matchingtracks reconstructed in theinner detectortotracksortracksegmentsinthe muonspec- trometer [54].Muonsusedforthissearchmusthave|η|<2.5 and pT>7 GeV,andarerequired tosatisfy ‘LooseTrackOnly’isolation requirements,similartothoseusedforelectrons,aswell asinner detectorandmuonspectrometertrack‘Loose’qualitycriteria,cor- respondingtoanefficiencyofabout97%.
Jetsarereconstructed usingtheanti-kt algorithm [55,56] with radius parameter R=0.4 fromclustersof energydepositsin the calorimetersystem [57].Candidatejetsarerequiredtohave pT>
20 GeV (pT>30 GeV)for|η|<2.5 (2.5<|η|<4.5).Low-pT jets from pile-up are rejected by a multivariate algorithm that uses propertiesofthereconstructedtracksintheevent [58].
Jetscontainingb-hadronsareselectedusingamultivariatetag- ging algorithm (b-tagging) [59,60]. The energy of the tagged jet (b-jet)is correctedforthe averageenergyloss fromsemileptonic decaysof b-hadronsandout-of-jet-cone tracks withlargeimpact parameters [61].Theb-taggingefficiencyforthejet pTrangeused in thisanalysis is between65% and 75%. Applying the b-tagging algorithm reducesthenumberoflight-flavour(c-quark)jetsbya factorof250–550(10–20),dependingonthejetkinematics.
Whenelectrons,muonsandjetsarespatiallyclose,thesealgo- rithmscan leadto ambiguousidentifications.An overlapremoval procedure [61] isthereforeapplied touniquelyidentifytheseob- jects.
The missing transverse momentum, EmissT , is computed using reconstructedandcalibratedleptons,photonsandjets [62].Tracks
struction [63].
5. Eventselection
The decay A → Z H →bb features a pair of oppositely charged, same flavour leptons and two b-jets. Three resonances can be formed by combining the selected objects: the Z boson (), the H boson (bb) andthe A boson (bb). Moreover, addi- tional b-jets may be presentif the A boson is produced via the b-associated production mechanism. These features are used to definetheeventselectionassummarisedinTable1.
Theeventsrecordedbythesingle-muonandthesingle-electron triggers are required to contain exactly two muons ortwo elec- trons, respectively. At least one of the leptons must have pT>
27 GeV.Only events that contain a primary vertex withat least two associatedtrackswith pT>400 MeV [64] areconsidered.In the case of muons, they are required to have opposite electric charges.Nosuch requirementisapplied toelectrons duetotheir non-negligiblechargemisidentificationratesresultingfromconver- sionsofbremsstrahlungphotons. Theinvariantmassofthelepton pair,m,must be inthe rangeof 80–100 GeV tobe compatible withthemassofthe Z boson.
The H→bb decayis reconstructed by requiring atleast two b-jets withthe highest-pT one having pT>45 GeV. When more thantwob-jetsarepresent, thetwohighest-pT b-jets areconsid- eredtobefromtheHdecay.Requiringb-jetsincreasesthefraction oftop-quarkbackgroundinthesignal region,includingtop-quark pairandsingle-top-quarkproduction.Thisisreducedby requiring EmissT /√
HT<3.5 GeV1/2,where HT isthescalarsumofthe pT of all jetsand leptons inthe event. In addition, a requirementthat reducesthe Z+jetsbackgroundisalsoapplied:
p2T/mbb>0.4, where mbb is the four-body invariant mass of the two-lepton, two-b-jet systemassignedto the A bosonandthe summationis performedoverthe pT oftheseobjects.
Subsequently,two categoriesare defined:the nb=2 category, whichcontainseventswithexactlytwob-jets,andthenb≥3 cat- egory, which contains events with three or more b-jets. Forthe gluon–gluonfusionproduction,94%–97%oftheeventspassingthe above selection fall into the nb=2 category, depending on the assumed mA andmH. However for the b-associated production, 27%–36%fallintothenb≥3 category.Theremaining b-associated produced signal events are categorised as nb =2 events, even though more than two b-jets are expected, dueto the relatively
2 Theprimaryvertexistakentobethe reconstructedvertexwiththehighest p2Toftheassociatedtracks.
Finally,theinvariantmassofthetwoleadingb-jets,mbb,must be compatiblewiththeassumed H boson massby satisfyingthe requirementof0.85·mH−20 GeV<mbb<mH+20 GeV forthe nb=2 category,and0.85·mH−25 GeV<mbb<mH+50 GeV for thenb≥3 category.Thewiderwindowfornb≥3 ismotivatedbya slightlydegradedresolutionduetopotentialb-jetmis-assignments (see later). Theoverallsignal efficiencyofthenb=2 category af- ter this requirement is 5%–11% (3%–7%) for gluon–gluon fusion (b-associated production), depending on the mA and mH values.
Similarly, the efficiency of the nb≥3 category is 2%–4% for the b-associatedproduction.Thesignalregionselectionissummarised inTable1.
Thembb distributionafterthembb requirementisusedtodis- criminate between signal and background. Toimprove the mbb resolution,thebbsystem’sfour-momentumcomponentsarescaled to matchthe assumed H boson mass and the system’s four- momentum components are scaled to matchthe Z boson mass.
Thisprocedure,performedaftertheeventselection,improvesthe mbb resolutionbyafactoroftwowithoutsignificantlydistorting thebackgrounddistributions,resultinginan A bosonmassresolu- tionof0.3%–4%.
The dominant backgrounds after these selections are from Z+jetsandtop-quarkproduction.Fortop-quark-pairproduction,a verypure(>99% ofpredictedevents)controlregionisusedtode- terminethenormalisationofthebackground,whereasitsshapein thesignalregionistakenfromthesimulation.Thiscontrolregion is definedbykeepingthe sameselection asdiscussedpreviously, apart from an opposite-flavour lepton criterion, i.e., an opposite- charge eμpair isrequiredinstead ofan ee or μμ pair(see also Table 1). The shape ofthe Z+jets backgrounddistribution is ob- tained from simulation and the normalisation is extracted from datatogether withthesignal (seealso Section7). Thisprocedure is possiblebecause ofthevery differentshapesofthe mbb dis- tributions fromsignaland Z+jetsevents.Thenormalisationofthe Z+jets production is further constrained by a control region de- fined by inverting the mbb window criterion for each H boson masshypothesis(seealsoTable1).Thecontrolregionsaredistinct forthenb=2 andthenb≥3 categories,sincetheaccuracyofthe background simulation dependson the numberof b-jets present intheevent.Backgroundsfromdiboson,single top,andHiggsbo- sonproduction,aswellastop-quark-pairproductioninassociation withavectorboson,giveatypicalcontributionof∼5%tothetotal background.Theirshapesaretakenfromsimulation,whereasthey are normalised using precise inclusive cross-sections calculated from theory.The dibosonsamples are normalised using next-to- next-to-leading-order (NNLO) cross-sections [65–68]. Single-top- quark production and top-quark-pair production in association withvectorbosons arenormalisedtonext-to-leading-order (NLO)
