Searches for a heavy scalar boson H decaying to a pair of 125 GeV Higgs bosons hh or for a heavy pseudoscalar boson A decaying to Zh, in the final states with h→ττ

Searches for a heavy scalar boson H decaying to a pair of 125 GeV Higgs bosons hh or for a heavy pseudoscalar boson A decaying to Zh, in the final states with h → τ τ

.CMS Collaboration

^

CERN,Switzerland

a r t i c l e i n f o a b s t ra c t

Articlehistory:

Received5October2015

Receivedinrevisedform10January2016 Accepted26January2016

Availableonline9February2016 Editor:M.Doser

Keywords:

CMS Physics MSSM τ Higgs

AsearchforaheavyscalarbosonHdecayingintoapairoflighterstandard-model-like125 GeVHiggs bosons hh andasearchforaheavypseudoscalarboson AdecayingintoaZandanhbosonarepresented.

The searches are performedon adata set corresponding toan integrated luminosity of19.7 fb⁻¹ of pp collisiondataatacentre-of-massenergyof8 TeV,collectedbyCMSin2012.A finalstateconsisting oftwoτ leptonsandtwobjetsisusedtosearchfortheH→^{hh decay.A finalstateconsistingoftwo} τ leptonsfromtheh bosondecay,andtwoadditionalleptonsfromtheZ bosondecay,isusedtosearch forthedecayA→^{Zh.Theresultsare}interpretedinthecontextoftwo-Higgs-doubletmodels.Noexcess isfoundabovethestandardmodelexpectationandupperlimitsaresetontheheavybosonproduction crosssectionsinthemassranges260<mH<350 GeV and220<mA<350 GeV.

©^{2016CERNforthebenefitoftheCMS}Collaboration.PublishedbyElsevierB.V.Thisisanopenaccess articleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP³.

1. Introduction

ThediscoveryofadditionalHiggsbosonsattheLHCwouldpro- vide directevidence ofphysics beyondthe standardmodel (SM).

ThereareseveraltypesofmodelsthatrequiretwoHiggsdoublets [1–3].Forexample theminimal supersymmetricextension ofthe SM(MSSM)requiresthe introductionofanadditionalHiggsdou- blet,whereoneHiggsdoublet couplestoup-typequarksandthe othertodown-typequarks[4–11].Thisleads tothepredictionof fiveHiggsparticles:onelightandoneheavyCP-evenHiggsboson, handH,oneCP-oddHiggsbosonA,andtwochargedHiggsbosons H^±[2,12].Themassesandcouplingsofthesebosonsare interre- latedand,attreelevel,canbedescribedbytwoparameters,which areoftenchosentobethemassofthepseudoscalarbosonm_A and theratioofthevacuumexpectationvaluesoftheneutralcompo- nentsofthe twoHiggsdoublets tanβ.However, radiativecorrec- tions[13–17]introducedependenciesonotherparametersnamely themassofthetopquarkm_t,thescaleofthesoftsupersymmetry breakingmassesM_SUSY,thehiggsinomassparameter

μ

,thewino mass parameter M2, the third-generation trilinear couplings, At, Ab,and Aτ ,themass ofthe gluinom_g˜,andthe third-generation sleptonmassparameterM_˜

3.

 E-mailaddress:[email protected].

Direct searchesfortheneutralMSSM Higgsbosons havebeen performed by the CMS and ATLAS Collaborations [18–20] using thebenchmarkscenarios proposedinRef.[21].Inthesescenarios theparameters involvedintheradiativecorrectionsfortheHiggs bosonmassesandcouplingshavebeenfixed,andonlythetwopa- rametersmA andtanβ remainfree.Thevalue ofMSUSY wasfixed ataround 1 TeV, whichproduces a lightest CP-evenHiggs boson with a mass m_h lower than the observed Higgs boson mass of 125.09±⁰.21(stat)±⁰.11(syst)GeV[22],forvaluesoftanβ^6.

If, however,M_SUSY ismuch largerthan1 TeV, assuggestedby the non-observationsofSUSY partner particles attheLHC sofar, low valuesof tanβ canproduce an h boson withm_h^{125 GeV} [23,24].TheinterpretationoftheHiggsbosonmeasurementsinthe framework oftherecentlydevelopedMSSM benchmarkscenarios [24–27] suggests that the mass of the CP-odd Higgs boson, mA, can be smaller than 2mt. In the mass region below 2mt and at low valuesof tanβ,the decaymode oftheheavy scalar H→^hh andthatofthepseudoscalarA→^{Zh canhavesizeablebranching} fractions.

Thisencouragesaprogrammeofsearchesintheso-called“low tanβ”channels[23,28]:

• ^{for220 GeV}<mA<2mt:A→^Zh;

• ^{for260 GeV}<mA<2mt:H→^hh;

• ^formA>2mt:A/H→^t¯t.

http://dx.doi.org/10.1016/j.physletb.2016.01.056

0370-2693/©^{2016CERNforthebenefitoftheCMS}Collaboration.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense (http://creativecommons.org/licenses/by/4.0/).FundedbySCOAP³.

The decay modes H→^{hh and A}→^{Zh, studied in this pa-} per, are alsopresentin other typesoftwo-Higgs-doublet models (2HDM)[2,3].Therearedifferenttypesof2HDMwiththosemost similar to the MSSM (i.e. whereup-type fermions coupleto one doublet and down-type fermions to the other) being “Type II”

2HDM.Thediscovery ofaHiggsbosonattheLHC[29–31] witha massaround125 GeV pushesthe2HDMparameterspacetowards eitherthealignmentordecouplinglimits[24].Intheselimitsthe propertiesofh areSM-like.

Inthealignment limitof2HDMwhen cos(β−

α

)^{1 (where}

α

is themixinganglebetweenthetwo neutralscalarfields),the Hhh and AZh couplings vanish at Born level [32]. However, in the MSSM, the Hhh and AZh couplings do not vanish, even in thealignmentlimit,becauseofthelargeradiativecorrectionsthat arise in the model. In the decoupling limit of 2HDM the scalar HiggsbosonH hasaverylargemassandthedecayH→t¯t domi- nates[32].

Thispaperreportsthe resultsofsearches forthedecays H→ hh→^bb

τ τ

and A→^Zh→ 

τ τ

(where  denotes

μμ

or ee).

Thechoiceof

τ

pairfinalstatewasdrivenbyitsquitecleansigna- tureandbythemostrecentresults,whichgavestrongerevidence ofthe125 Higgsbosoncouplingtothefermions[33].Thisanalysis exploitssimilartechniquesasusedforthesearchfortheSMHiggs boson at 125 GeV [34] and several different

τ τ

signatures are studied.Forthechannel H→^hh→^bb

τ τ

,the

μτ

h,e

τ

h,and

τ

h

τ

h final states are used,where

τ

h denotes the visibleproducts ofa hadronicallydecaying

τ

,whereasforthechannelA→^Zh→ 

τ τ

, the

μτ

h,e

τ

h,

τ

h

τ

h,ande

μ

finalstatesareselected.

SearchesforthedecaysH→^hh,andA→^{Zh havealreadybeen} performedbytheATLAS [35–38]andCMSCollaborations[39–41]

indi-photon,multileptonandbb finalstates.

This analysishas the powerto bring important results inthe low tanβ region for the m_A range, which has been previously discussed andwhere theseprocesses havean enhanced sensitiv- ity [23]. This region has not yet been excluded by the direct or indirectsearches fora heavy scalaror pseudoscalarHiggsboson, that have been mentioned above, therefore the described decay modeslooktobequitepromising.

Forsimplicityofthepaper,weareneitherindicatingthecharge oftheleptonsnortheparticle–antiparticlenatureofquarks.

2. TheCMSdetector,simulationanddatasamples

A detailed description of the CMS detector can be found in Ref. [42]. The central feature of the CMS apparatus is a super- conductingsolenoidof6 m internal diameterproviding afield of 3.8 T.Withinthefieldvolumeareasiliconpixelandstriptracker, a crystal electromagnetic calorimeter (ECAL), and a brass/scintil- lator hadron calorimeter. Muons are measured in gas-ionisation detectorsembeddedinthesteelreturnyokeofthemagnet.

TheCMScoordinatesystemhastheorigincentred atthenom- inalcollision pointandis orientedsuch thatthe x-axispoints to thecentre oftheLHCring,they-axispointsverticallyupwardand the z-axisisinthedirectionofthebeam. Theazimuthal angleφ ismeasuredfromthex-axisinthexy planeandtheradialcoordi- nateinthisplaneisdenotedby r.Thepolarangleθ isdefinedin therz planeandthepseudorapidityis

η

= −^ln[^tan(θ/2)]^[42].The momentumcomponenttransversetothebeamdirection,denoted by p_T,iscomputedfromthex- and y-components.

Thefirstlevel(L1)oftheCMStriggersystem,composedofcus- tomhardware processors, usesinformationfromthecalorimeters andmuondetectorstoselectthemostinterestingeventsinafixed time interval of less than 4 μs. The high-level Trigger processor farm decreases the L1 accept rate from around 100 kHz to less than1 kHz beforedatastorage.

Thedatausedforthissearchwere recordedwiththeCMSde- tectorinproton–protoncollisionsattheCERNLHCandcorrespond to an integrated luminosity of19.7 fb⁻¹ at acentre-of-mass en- ergy of √

s=^{8 TeV. The H}→^{hh signals are modelled with the} pythia6.4.26[43] eventgeneratorwhilethe A→^{Zh signalswere} modelled with MadGraph 5.1 [44]. When modelling background processes,the MadGraph 5.1generatorisusedforZ+^jets,W+^jets, t¯t,anddibosonproduction,and powheg 1.0[45–48]forsingle top quark production.The powheg and MadGraph generators arein- terfacedwith pythia forpartonshoweringandfragmentationusing the Z2*tune [49].All generators are interfacedwith tauola[50]

forthe simulationof the

τ

decays.All generatedeventsare pro- cessed through a detailed simulation of the CMS detector based on Geant4 [51] andare reconstructed with the samealgorithms as the data. Parton distribution functions (PDFs) CT10 [52] or CTEQ6L1[53]fortheprotonareused,dependingonthegenerator inquestion, togetherwithMSTW2008[54] accordingto PDF4LHC prescriptions[55].

3. Eventreconstruction

Duringthe2012LHCrun therewere anaverageof21proton–

proton interactions per bunch crossing. The collision vertex that maximises the sumofthe squares ofmomenta componentsper- pendiculartothebeamline(transverse momenta)ofalltracksas- sociated withit, 

p²_T, is taken to be the vertex of the primary hard interaction.The otherverticesare categorisedaspileupver- tices.

A particle-flow algorithm [56,57] is used to reconstruct indi- vidual particles, i.e. muons, electrons, photons, charged hadrons andneutralhadrons,usinginformationfromallCMSsubdetectors.

Composite objects such as jets, hadronically decaying

τ

leptons, andmissingtransverseenergyarethenconstructedusingthelists ofindividualparticles.

Muonsare reconstructed byperforming a simultaneousglobal trackfit to hitsinthe silicontracker andthemuon system [58].

Electronsare reconstructedfromclustersofECALenergydeposits matched to hits in the silicontracker [59].Muons andelectrons assumed to originate from W or Z bosondecays are required to be spatially isolatedfromother particles [59,60].The presenceof charged and neutral particles from pileup vertices is taken into account in the isolation requirement of both muons and elec- trons. Muon andelectron identification and isolation efficiencies aremeasuredviathetag-and-probetechnique[61]usinginclusive samples of Z→  ^{events from data and} simulation. Correction factors are applied to account for differences between data and simulation.

Jets are reconstructed fromall particles using the anti-kT jet clustering algorithm implemented in fastjet [62,63] with a dis- tance parameter of 0.5. The contribution to the jet energy from particles originatingfrom pileup vertices is removed following a procedure based on the effective jet area described in Ref. [64].

Furthermore, jet energy corrections are applied as a function of jet pT and

η

correcting jet energies to the generator level re- sponse ofthe jet, on average. Jets originatingfrom pileup inter- actions areremoved by amultivariate pileupjetidentificational- gorithm[65].

The missingtransverse momentum vector ^pmiss_T ^{is definedas} the negative vector sumofthe transversemomenta ofall recon- structedparticlesinthevolumeofthedetector(electrons,muons, photons, andhadrons). Its magnitudeis referred toas E^miss_T . The E^miss_T reconstructionisimprovedbytakingintoaccountthejeten- ergyscalecorrectionsandtheφmodulation,duetocollisions not beingatthenominalcentreofCMS[66].A multivariateregression correctionof E^miss_T ,wherethecontributingparticlesareseparated

hadron-plus-stripsalgorithm[68],whichconsiderscandidateswith one chargedpion andup to two neutralpions, or threecharged pions. The neutral pions are reconstructed as “strips” of elec- tromagnetic particles taking into account possible broadening of calorimeterenergydepositionsintheφdirectionfromphotoncon- versions.The

τ

h candidatesthat are alsocompatiblewithmuons orelectronsarerejected.Jetsoriginatingfromthehadronisation of quarksandgluonsaresuppressedbyrequiringthe

τ

hcandidateto beisolated.Thecontributionofchargedandneutralparticlesfrom pileupinteractionsisremovedwhencomputingtheisolation.

4. Eventselection

Theeventsare selectedwitha combinationof electron,muon and

τ

trigger objects [34,59,60,69]. The identification criteria of these objects were progressively tightened and their transverse momentum thresholds raised asthe LHC instantaneous luminos- ityincreasedoverthedatatakingperiod.A tag-and-probemethod wasusedtomeasuretheefficienciesofthesetriggersindataand simulation,andcorrectionfactorsareappliedtothesimulation.

Electrons,muons,and

τ

hareselectedusingthecriteriadefined intheCMSsearch fortheSM Higgsbosonat125 GeV[34]. Spe- cificrequirementsfortheselectionoftheH→^hh→^bb

τ τ

andthe A→^Zh→ 

τ τ

channelsaredescribedbelow.

4.1.EventselectionofH→^hh→^bb

τ τ

IntheH→^hh→^bb

τ τ

channel,the threemostsensitive final statesare analysed, distinguished by the decaymode of the two

τ

leptonsoriginatingfromtheh boson(

μτ

h,e

τ

hand

τ

h

τ

h).

Inthe

μτ

hande

τ

hfinalstates,eventsareselectedwithamuon with p_T>20 GeV and |

η

|<2.1 or an electron of p_T>24 GeV and|

η

_|<2.1,and an oppositely charged

τ

h of pT>20 GeV and

|

η

_|<2.3. To reduce the Z→

μμ

,ee contamination, events with twomuonsorelectronsof p_T>15 GeV,ofopposite charges,and passinglooseisolationcriteriaarerejected.

In the

μτ

h and e

τ

h final states, the transverse mass of the muonorelectronandp^miss_T

m_T

= 

2p_TE^miss_T

(

1

−

^cos

φ),

(1)

where pT isthelepton transversemomentumandφ isthe dif- ferenceintheazimuthalanglebetweentheleptonmomentumand



p^miss_T ,isrequiredto be lessthan 30 GeV to rejecteventscoming fromW+^{jets andt}¯t backgrounds.Them_T distributionforthe

μτ

h finalstateisshowninFig. 1.

In the

τ

h

τ

h final state, events with two oppositely charged hadronically decaying

τ

leptons with p_T>45 GeV and |

η

|<2.1 areselected.

Inadditiontothe

τ τ

selection,each selectedeventmustcon- tain atleast two jetswith pT>20 GeV and |

η

_|<2.4.These pT and

η

requirementsare necessary to selectjetsthat have a well definedvalueoftheCSVdiscriminator(Section3),whichisimpor- tant forcategorisingsignal-like events withtwo b jet candidates comingfromthe125 GeV Higgsbosondecayingtobb.

Fig. 1. DistributionofmTforeventsintheμτh finalstate,containingatleasttwo additionaljets.TheW+jets backgroundisincludedinthe“electroweak”category.

MultijeteventsareindicatedasQCD.TheH→hh→bbτ τselectionrequiresmT<

30 GeV fortheμτhandeτhfinalstates.

Simulationstudiesshowthatthemajorityofsignaleventswill have at least one jet passing the medium working point of the CSVdiscriminator.ThejetsareorderedbyCSVdiscriminatorvalue, suchthattheleadingandsubleadingjetsaredefinedasthosewith the two highest CSV values. Then the events are separated into categories,definedas:

• ^{2jet–0tagwhen neithertheleading norsubleadingjet passes} themediumCSVworkingpoint.Onlyasmallamountofsignal iscollectedinthiscategory,whichisbackground-dominated.

• ^{2jet–1tag when only the leading but not the subleading jet} passesthemediumCSVworkingpoint.

• ^{2jet–2tagwhenboththeleadingandsubleadingjetspassthe} mediumCSVworkingpoint.

Thesignalextractionisperformedusingthedistributionofthe reconstructedmassoftheHbosoncandidate.

4.2. EventselectionofA→^Zh→ 

τ τ

IntheA→^Zh→ 

τ τ

channeleight finalstatesareanalysed.

ThesearecategorisedaccordingtothedecaymodeoftheZboson andthe decaymode ofthe

τ

leptons originatingfromthe h bo- son.

The Z bosonis reconstructedfrom twosame-flavour, isolated, andoppositelychargedelectronsormuons.IntheZ→

μμ

(ee)fi- nalstatethemuons(electrons)arerequiredtohave|

η

_|<2.4 (2.5) with p_T>20 GeV for the leading lepton and p_T>10 GeV for the subleading lepton. The invariant mass of the two leptons is required to be between60 GeV and 120 GeV. When more than onepairofleptonssatisfythesecriteria,thepairwithaninvariant massclosesttotheZbosonmassisselected.

AftertheZ candidatehasbeenchosen,theh→

τ τ

decayisse- lectedbycombiningthedecayproductsofthetwo

τ

leptonsinthe four final states

μτ

h,e

τ

h,

τ

h

τ

h,e

μ

.The combination of thelarge contributionfromtheirreducibleZZ backgroundandofthesmall branching fractions ofleptonic taudecays makes the

μμ

andee final stateslesssensitivetothe signal,andthereforethey arenot used in the analysis. Depending on the final state, a muonwith pT>10 GeV and |

η

_|<2.4, or an electron of pT>10 GeV and

|

η

|<2.5,or a

τ

h of p_T>21 GeV and |

η

|<2.3 arecombined to

Fig. 2. DistributionofthevariableL^h_Tforeventsintheτhτh finalstate.There- duciblebackgroundisestimatedfromdata,insteadtheZZirreduciblebackground fromsimulation.

formanoppositely chargedpair.Eventswithadditionallightlep- tonssatisfyingtheserequirementsarerejected.

A requirement on L^h_T, which is the scalar sum of the visible transverse momenta of the two

τ

candidates originating from the h boson, is applied to lower the reducible background from misidentifiedleptonsaswellastheirreduciblebackgroundfrom ZZ production.The thresholdsofthisrequirementdepend onthe fi- nalstate andhavebeen choseninsuch awayastooptimisethe sensitivityoftheanalysistothepresenceofan A→^{Zh signal for} A masses between 220 and350 GeV. The distribution of L^h_T for eventsinthe

τ

h

τ

hfinalstatecanbeseeninFig. 2.

In order to reduce the t¯t background, eventscontaining a jet with pT>20 GeV, |

η

|<2.4 and passing the medium working pointoftheCSVbtaggingdiscriminatorareremoved.

Thefourfinalobjectsarefurtherrequiredtobeseparatedfrom eachother byR=

(

η

)²+ (

ϕ

)² largerthan 0.5(where

ϕ

is inradians),andtocomefromthesameprimaryvertex.

Inthischannelthesignalextractionisperformedusingthedis- tributionofthereconstructedmassoftheAbosoncandidate.

5. Backgroundestimation

5.1. BackgroundestimationforH→^hh→^bb

τ τ

ThebackgroundstotheH→^hh→^bb

τ τ

finalstateconsistpre- dominantlyoft¯t events,followedbyZ→

τ τ

+jets events,W+^jets events, and QCD multijet events, with other small contributions from Z→ ^{, diboson, and single top quark} production.The es- timation of the shapes of the reconstructed H mass and of the yields of the major backgrounds isobtained fromdata wherever possible.

TheZ→

τ τ

process constitutesanirreduciblebackgrounddue to its final state involving two

τ

leptons, whichonly differ from theh→

τ τ

signalbyhavinganinvariantmassclosertothemass ofthe Z bosoninstead ofthe Higgsboson. Requiringtwo jetsin the eventgreatly reduces this backgroundand the btagging re- quirements reduce it even further. Nevertheless, it still remains an important source of background events, in particular in the 2jet–1tag and2jet–0tag categories. This background isestimated usingasampleofZ→

μμ

eventsfromdata,obtainedbyrequiring two oppositely charged isolated muons, where the reconstructed

Fig. 3. Distributionsofthereconstructedfour-bodymasswiththekinematicfitafter applyingmassselectionsonm_{τ τ}andmbbintheμτhchannel.Theplotsareshown foreventsinthe2jet–0tag(top),2jet–1tag(middle),and2jet–2tag(bottom)cate- gories.Theexpectedsignalscaledbyafactor10isshownsuperimposedasanopen dashedhistogramfortanβ=^{2 andmH}=^{300 GeV inthelowtan}βscenarioofthe MSSM.Expectedbackgroundcontributionsareshownforthevaluesofnuisancepa- rameters(systematicuncertainties)obtainedafterfittingthesignalplusbackground hypothesistothedata.

Fig. 4. Distributionsofthereconstructedfour-bodymasswiththekinematicfitafter applyingmassselectionsonmτ τandmbbintheeτhchannel.Theplotsareshown foreventsinthe2jet–0tag(top),2jet–1tag(middle),and2jet–2tag(bottom)cate- gories.Theexpectedsignalscaledbyafactor10isshownsuperimposedasanopen dashedhistogramfortanβ=2 andmH=300 GeV inthelowtanβscenarioofthe MSSM.Expectedbackgroundcontributionsareshownforthevaluesofnuisancepa- rameters(systematicuncertainties)obtainedafterfittingthesignalplusbackground hypothesistothedata.

Fig. 5. Distributionsofthereconstructedfour-bodymasswiththekinematicfitafter applyingmassselectionsonmτ τandmbbintheτhτhchannel.Theplotsareshown foreventsinthe2jet–0tag(top),2jet–1tag(middle),and2jet–2tag(bottom)cate- gories.Theexpectedsignalscaledbyafactor10isshownsuperimposedasanopen dashedhistogramfortanβ=^{2 andm}H=^{300 GeV inthelowtan}βscenarioofthe MSSM.Expectedbackgroundcontributionsareshownforthevaluesofnuisancepa- rameters(systematicuncertainties)obtainedafterfittingthesignalplusbackground hypothesistothedata.

Fig. 6. InvariantmassdistributionsfordifferentfinalstatesoftheA→^{Zh processwhereZ decaystoee.Theexpectedsignalscaledbyafactor5isshown}superimposed asanopendashedhistogramfortanβ=^{2 andmA}=^{300 GeV inthelowtan}βscenarioofMSSM.Expectedbackgroundcontributionsareshownforthevaluesofnuisance parameters(systematicuncertainties)obtainedafterfittingthesignalplusbackgroundhypothesistothedata.(Forinterpretationofthereferencestocolourinthisfigure, thereaderisreferredtothewebversionofthisarticle.)

muons are replaced by the reconstructed particles from simu- lated

τ

decays. A correction for a contamination from t¯t events is applied to the Z→

μμ

selection. Thistechnique substantially reduces the systematic uncertainties dueto the jet energy scale andthe missingtransverse energy, asthese quantities are mod- elledwithdata.

Forthet¯t background,bothshapeandnormalisationaretaken from Monte Carlo simulation (MC), and the results are checked against data ina control region wherethe presence of t¯t events isenhanced by requiringe

μ

in thefinal state insteadof aditau, andatleastonebtaggedjet.

Another significant source of background is from QCD multi- jetevents,whichcanmimicthesignalinvariousways, e.g.where oneormorejetsaremisidentifiedas

τ

h.Inthe

μτ

hande

τ

hchan- nels,the shapeoftheQCD backgroundisestimatedusingan ob- servedsample ofsame-sign (SS)

τ τ

events.Theyield isobtained by scaling theobserved numberof SSevents by theratio ofthe opposite-sign(OS)to SSeventyields obtainedina QCD-enriched regionwithrelaxedleptonisolation.Inthe

τ

h

τ

hchannel,theshape isobtainedfromOS eventswithrelaxed

τ

isolation. The yieldis obtained by scaling these events by the ratio of SS events with tighterandrelaxed

τ

isolation.

In the

μτ

h and e

τ

_h channels, W+jets events in which there isa jetmisidentified asa

τ

h are anothersizeablesourceofback- ground. The W+^{jets shape is modelledusing MC simulationand} the yieldisestimatedusinga controlregion ofeventswithlarge mTclosetotheW mass.Inthe

τ

h

τ

hchannelthisbackgroundhas been foundto be lessrelevantandits shape andyieldare taken fromMCsimulation.

The contribution of Drell–Yan production of muon and elec- tron pairs is estimatedfrom simulationafter rescaling the simu- latedyieldtothatmeasuredfromobservedZ→

μμ

events.Inthe e

τ

hchannel, theZ→ee simulationisfurthercorrectedusingthe e→

τ

_h misidentification ratemeasured indata using a tag-and- probetechnique[61]onZ→^{ee events.}

Finally the contributions of other minor backgrounds such as diboson and single top quark events are estimatedfrom simula- tion. Possible contributions fromSM Higgs boson production are estimatedandfoundtohaveanegligibleeffectonthefinalresult.

5.2. BackgroundestimationforA→Zh→ 

τ τ

The backgrounds to the A→^{Zh channel can be divided into} a reduciblecomponentandanirreduciblecomponentwhichcon- tributeinequalparts.

Fig. 7. InvariantmassdistributionsfordifferentfinalstatesoftheA→^{Zh processwhereZ decaysto}μμ.Theexpectedsignalscaledbyafactor5isshownsuperimposed asanopendashedhistogramfortanβ=^{2 andmA}=^{300 GeV inthelowtan}βscenarioofMSSM.Expectedbackgroundcontributionsareshownforthevaluesofnuisance parameters(systematicuncertainties)obtainedafterfittingthesignalplusbackgroundhypothesistothedata.(Forinterpretationofthereferencestocolourinthisfigure, thereaderisreferredtothewebversionofthisarticle.)

Thepredominant source ofirreducible backgroundis from ZZ production that yields exactly the same final states as the ex- pectedsignal. Other “rare” sources ofirreducible background are SMHiggsbosonassociatedproductionwithaZ boson,t¯tZ produc- tionwheretheZ bosondecaysintoamuonoranelectronpairand both top quarksdecay leptonically (to e,

μ

,or

τ

h), andtriboson events(WWZ,WZZ,ZZZ). The contributions ofall theirreducible backgrounds after the final selection are estimated fromsimula- tion.

Thereduciblebackgroundshaveatleastoneleptoninthefinal statethatisduetoamisidentifiedjetthatpassestheleptoniden- tification.In

τ

h

τ

hfinalstates,thereduciblebackgroundisessen- tiallycomposed ofZ+^{jetsevents withatleast twojets, whereas} in

μτ

h ande

τ

hfinal states,the maincontributionto there- duciblebackgroundcomesfromWZ+^{jetswiththreelightleptons.}

Thecontributionfromtheseprocessestothefinalselectedevents isestimatedusingcontrolsamplesindata.

Theprobabilities fora jet that passesrelaxedlepton selection criteriatopassthefinalidentificationandisolationcriteriaofelec- trons,muons, and

τ

leptonsare measuredin a signal-freeregion

asafunctionofthetransversemomentumoftheobjectclosestto thecandidate, f(p^fake_T ).Inthisregion,eventsarerequiredtopass all thefinal stateselections, exceptthatthe reconstructed

τ

can- didates are required to have the same sign and to pass relaxed identificationandisolationcriteria. Thiseffectivelyeliminates any possiblesignal,whilemaintainingroughly thesameproportionof reduciblebackgroundevents.

Inordertousethemisidentificationprobabilities f(p^fake_T ),side- bands are defined for each channel, where, unlike the relaxed criterion, the final identification or isolation criterion is not sat- isfied for one or more of the final state lepton candidates. The number of reducible background events due to a lepton being misidentified in the final selection is estimated by applying the weight f(p^fake_T )/(1− ^f(p^fake_T )) to the observed events with lep- toncandidatesinthesidebandthatsatisfytherelaxedbutnotthe finalidentificationorisolationcriterion.Finally,thereducibleback- groundshape ofthe reconstructedA mass isobtainedfroma SS signal-free region wherethe

τ

candidates havethe same charge andrelaxedisolationcriteria.PossiblecontributionsfromSMHiggs bosonproductionareestimatedandfoundtohaveanegligibleef- fectonthefinalresult.

Fig. 8. Upperlimitsat95%CLontheH→hh→bbτ τ crosssectiontimesbranchingfractionfortheμτh(topleft),eτh(topright),τhτh(bottomleft),andforfinalstates combined(bottomright).

6. Systematicuncertainties

The shape of the reconstructed mass of the A and H boson candidates, used for signal extraction, and the normalisation are sensitivetovarioussystematicuncertainties.

The main contributions to the normalisation uncertainty that affectthe signal andthe simulated backgrounds include the un- certaintyinthetotalintegratedluminosity,whichamountsto2.6%

[70], andthe identificationandtriggerefficiencies ofmuons(2%) andelectrons (2%). The

τ

h identificationefficiency has a 6% un- certainty(8%inthe

τ

h

τ

hchannel),whichismeasuredinZ/

γ

^∗→

τ τ

_→

μτ

_h eventsusinga tag-and-probetechnique. Thereis a3%

uncertaintyintheefficiencyonthe hadronicpartofthe

μτ

h and e

τ

h triggers,anda4.5%uncertaintyoneachofthetwo

τ

h candi- datesrequiredbythe

τ

h

τ

htrigger.Thebtaggingefficiencyhasan uncertaintyof2–7%,andthemistag rateforlight-flavourpartons is accurate to 10–20% depending on

η

and pT [67]. The back- groundnormalisation uncertainties fromthe estimation methods discussedinSection5arealsoconsidered.IntheH→^hh→^bb

τ τ

channel this uncertainties amount to 2–40% depending on the eventcategoryandonthefinalstate.Theuncertaintiesofreducible backgroundstotheA→^{Zh channelareestimatedbyevaluatingan}

individual uncertainty for each lepton misidentification rate and applyingittothebackgroundcalculation.Thisamountsto15–50%

depending on the final 

τ τ

state considered. The main uncer- tainty in the estimation of the ZZ background arises from the theoreticaluncertaintyintheZZ productioncrosssection.

Uncertaintiesthat contribute to variations inthe shape ofthe mass spectrum include the jet energy scale, which varies with jet p_T andjet

η

[71],andthe

τ

lepton(3%)energyscale[34].

Theoretical uncertainties onthe crosssection forsignal derive fromPDFandQCDscaleuncertaintiesanddependonthechoiceof signalhypothesis.Formodelindependentresultsnochoiceofcross sectionismadeandhencenotheoreticaluncertaintiesareconsid- ered.FortheMSSMinterpretationtheuncertaintiesdependonm_A andtanβ andamountto2–3%forPDFuncertaintiesand5–9%for scale uncertainties, evaluated as described in [27] andusing the PDF4LHC recommendations [55]. No theoretical uncertainties are consideredinthe2HDMinterpretation.

7. Resultsandinterpretation

The ditau(mττ ) massisreconstructedusinga dedicatedalgo- rithmcalled SVFit[72],whichcombinesthevisiblefour-vectorsof

Fig. 9. Upperlimitsat95%CLoncrosssectiontimesbranchingfractiononA→Zh→LLτ τ foreμ(topleft),μτh (topright),eτh(bottomleft),andτhτh(bottom right)finalstates.

the

τ

leptoncandidatesaswellasthe E^miss_T andits experimental resolutioninamaximumlikelihoodestimator.

Forthe H→^hh→^bb

τ τ

process, the chosen distribution for signalextractionisthefour-bodymass.Thedecayproducts ofthe twoh bosonsneedtofulfillstringentkinematicconstraints,dueto thesmallnaturalwidthoftheh.Theseconstraintscanbeusedin akinematic fit inorder toimprove theevent reconstruction and tobetterseparatesignaleventsfrombackground.Thecollinearap- proximationfor the decay products of the

τ

leptons isassumed inthefit,sincethe

τ

leptons arehighlyboostedasthey originate froman objectthat isheavy when comparedto their ownmass.

Furthermore,it is assumed that the reconstruction of the direc- tionsofallfinalstateobjectsisaccurateandtheuncertaintiescan beneglected comparedto theuncertainties ontheenergyrecon- struction.Inthedecayofthetwo

τ

leptons,atleasttwoneutrinos are involved and there is no precise measurement of the origi- nal

τ

lepton energies. Forthisreason, the

τ

lepton energies are constrainedfromthebalanceofthefittedH bosontransversemo- mentumandthereconstructedtransversalrecoildeterminedfrom E^miss_T reconstructionalgorithms,asdescribedinSec.3.Therecon- structedmassobtainedwiththekinematicfitisdenotedbym^kinfit_H (seetheSupplementarymaterialforadetaileddescription).

The signal-to-background ratio is greatly improved by select- ing events that are consistent with a mass of 125 GeV for both thedijet(mbb)massandtheditaumass(mττ )reconstructedwith SVFit.The mass windows of theselections are optimised to col- lect as much signal as possible while rejecting a large part of the background. They correspond to 70<m_bb<150 GeV and 90<_mττ <150 GeV.TheinvariantmassdistributionsoftheH bo- sonindifferentfinalstatesareshowninFigs. 3,4 and 5.

For the A→^Zh→ 

τ τ

process, the A boson mass is recon- structed from the four-vector information of the Z boson can- didate and the four-vector information of the h boson candi- date as obtainedfrom SVFit. The invariant mass distributions of the A boson in the different final states are shown in Figs. 6 and 7. The 

τ

h

τ

h final states have a comparable contribution from reducible and irreducible backgrounds, while the e

μ

fi- nal states are dominated by the irreducible ZZ production. The backgroundlabelledas“rare”collects togetherthesmallercontri- butions fromthe triboson processesasdiscussed inthe previous section.

In neithersearch do the invariant massspectra show anyev- idence ofa signal.Model independentupper limitsat95% confi- dencelevel(CL)onthecrosssectiontimesbranching fractionare

Fig. 10. Upperlimitsat95%CLoncrosssectiontimesbranchingfractiononA→ Zh→LLτ τforallτ τ finalstatescombined(top)andcomparisonofthedifferent finalstates(bottom).

setusingabinned maximumlikelihoodfitforthesignalplusback- ground andbackground-only hypotheses.Thelimitsaredetermined using the CL_s method [73,74] and the procedure is described in Refs.[75,76].

Systematicuncertaintiesaretakenintoaccountasnuisancepa- rameters in the fit procedure: normalisation uncertainties affect the signal and backgroundyields. Uncertaintieson the

τ

energy scaleandjetenergyscalearepropagatedasshapeuncertainties.

The model independent expected and observed cross section times branching fraction limits for the H→^hh→^bb

τ τ

process areshowninFig. 8andfortheA→^Zh→^LL

τ τ

processinFigs. 9 and10whereL=^e,

μ

or

τ

inordertoreflectthe smallZ→

τ τ

contributiontothesignalacceptance.

We interpret the observed limits on the cross section times branchingfractionintheMSSMand2HDMframeworks,discussed inSection1.

IntheMSSMweinterprettheminthe“lowtanβ”scenario[27, 78]inwhichthevalueofMSUSY isincreaseduntilthemassofthe lightestHiggsbosonisconsistentwith125 GeV overarangeoflow tanβ andmA values.The exclusionregionin themA–tanβ plane forthecombinationofthe H→^hh→^bb

τ τ

andA→^Zh→ 

τ τ

Fig. 11. The95%CLexclusionregioninthemA–tanβplaneforthelowtanβsce- narioasdiscussedintheintroduction,combiningtheresultsoftheH→^hh→^bbτ τ and theA→Zh→ τ τ analysis.Theareahighlighted inblue belowthe black curvemarkstheobservedexclusion.Thedashedcurveandthegreybandsshow theexpectedexclusionlimitwiththerelativeuncertainty.Theredareawiththe back-slashlinesatthelower-leftcorneroftheplotindicatestheregionexcludedby themassoftheSM-likescalarbosonbeing125 GeV.Thelimitfallsoffrapidlyas mAapproaches350 GeV becausedecaysoftheA totwotopquarksarebecoming kinematicallyallowed.(Forinterpretationofthereferencestocolourinthisfigure legend,thereaderisreferredtothewebversionofthisarticle.)

analyses,insuchascenario,isshowninFig. 11.Thelimitfallsoff rapidlyasmAapproaches350 GeV becausedecaysoftheA totwo topquarksarebecomingkinematicallyallowed.

The interpretationoftheobserved limitsinaType II2HDMis performedinthe“physicsbasis”.The inputstothisinterpretation are the physicalHiggs boson masses(mh,mH, mA,m_H±),the ra- tioofthevacuumexpectationenergies(tanβ),theCP-evenHiggs mixingangle(

α

) andm²₁₂=^m2_A[^tanβ/(1+^tanβ²)]^{.Forsimplicity} weassumethatmH=^mA=^mH^±.

The cross sections andbranching fractionsin the2HDMwere calculated as described by the LHC Higgs Cross Section Working Group [77,78].The exclusionregions,calculatedusingthe combi- nation ofthe H→^hh→^bb

τ τ

andA→^Zh→ 

τ τ

analyses, in the cos(β−

α

)vs. tanβ plane forsuch a Type II2HDMscenario withaheavy Higgsbosonmassof300 GeV areshowninFig. 12.

ThiscanbecomparedtoFig. 5inRef.[41].

8. Summary

A search for a heavy scalar Higgs boson (H) decaying into a pairofSM-likeHiggsbosons(hh)andasearchforaheavyneutral pseudoscalarHiggsboson (A)decayingintoa Z bosonandaSM- like Higgsboson(h), havebeenperformedusingeventsrecorded by theCMS experiment atthe LHC. The data set corresponds to anintegratedluminosityof19.7 fb⁻¹,recordedat8 TeV centre-of- massenergyin2012.Noevidenceforasignalhasbeenfoundand exclusion limitson the production cross section times branching fraction forthe processes H→^hh→^bb

τ τ

andA→^Zh→^LL

τ τ

are presented. The results are also interpreted in the context of theMSSMand2HDMmodels.

Fig. 12. The95%CLexclusionregionsinthecos(β−α)vs.tanβplaneof2HDM Type IImodelformA=^mH=^{300 GeV,combiningtheresultsofthe H}→^hh→ bbτ τ andA→Zh→ τ τ analysis.Theareashighlightedinblueboundedbythe blackcurvesmarktheobservedexclusion.Thedashedcurvesandthegreybands showtheexpectedexclusionlimitwiththerelativeuncertainty.(Forinterpretation ofthereferencestocolourinthisfigurelegend,thereaderisreferredtotheweb versionofthisarticle.)

Acknowledgements

WecongratulateourcolleaguesintheCERNacceleratordepart- ments for the excellent performance of the LHC and thank the technicalandadministrativestaffs atCERN andatother CMSin- stitutes for their contributions to the success of the CMS effort.

Inaddition,wegratefullyacknowledgethecomputingcentresand personneloftheWorldwideLHCComputingGridfordeliveringso effectivelythecomputinginfrastructure essential toour analyses.

Finally, we acknowledge the enduring support for the construc- tionandoperationofthe LHCandtheCMSdetectorprovided by thefollowingfundingagencies:BMWFWandFWF(Austria);FNRS and FWO (Belgium); CNPq, CAPES, FAPERJ, and FAPESP (Brazil);

MES(Bulgaria);CERN;CAS,MOST,andNSFC(China);COLCIENCIAS (Colombia);MSESandCSF(Croatia);RPF(Cyprus);MoER,ERCIUT andERDF(Estonia); AcademyofFinland,MEC, andHIP (Finland);

CEA and CNRS/IN2P3 (France); BMBF, DFG, and HGF (Germany);

GSRT (Greece); OTKA and NIH (Hungary); DAE and DST (India);

IPM (Iran); SFI (Ireland); INFN (Italy); MSIP and NRF (Republic ofKorea); LAS (Lithuania);MOE andUM (Malaysia); CINVESTAV, CONACYT,SEP,andUASLP-FAI(Mexico);MBIE(NewZealand);PAEC (Pakistan);MSHEandNSC(Poland);FCT(Portugal);JINR(Dubna);

MON,RosAtom,RASandRFBR(Russia);MESTD(Serbia);SEIDIand CPAN(Spain);SwissFundingAgencies(Switzerland);MST(Taipei);

ThEPCenter,IPST,STARandNSTDA(Thailand);TUBITAKandTAEK (Turkey);NASUandSFFR(Ukraine);STFC (UnitedKingdom);DOE andNSF(USA).

Individuals have received support from the Marie-Curie pro- gramme and the European Research Council and EPLANET (Eu- ropean Union); the Leventis Foundation;the A. P. Sloan Founda- tion; the Alexander von Humboldt Foundation; the Belgian Fed- eral Science Policy Office; the Fonds pour la Formation à la Recherchedansl’Industrieetdansl’Agriculture(FRIA-Belgium);the

doctoral Fellowship,Chulalongkorn University(Thailand); andthe WelchFoundation,contractC-1845.

Appendix A. Supplementarymaterial

Supplementarymaterialrelatedtothisarticlecanbefoundon- lineathttp://dx.doi.org/10.1016/j.physletb.2016.01.056.

References

[1] S.L.Glashow,S.Weinberg,Naturalconservationlawsforneutralcurrents,Phys.

Rev.D15(1977)1958,http://dx.doi.org/10.1103/PhysRevD.15.1958.

[2]J.F.Gunion,H.E.Haber,G.Kane,S.Dawson,TheHiggsHunter’sGuide,Frontiers inPhysics,vol. 80,Addison–Wesley,2000.

[3] G.C.Branco,P.M.Ferreira,L.Lavoura,M.N.Rebelo,M.Sher,J.P.Silva,Theory and phenomenology oftwo-Higgs-doublet models,Phys. Rep.516(2012) 1, http://dx.doi.org/10.1016/j.physrep.2012.02.002,arXiv:1106.0034.

[4] P. Fayet, Supergauge invariant extension of the Higgs mechanism and a modelfortheelectronanditsneutrino,Nucl.Phys.B90(1975)104,http://

dx.doi.org/10.1016/0550-3213(75)90636-7.

[5] P.Fayet,Supersymmetry andweak,electromagneticandstronginteractions, Phys.Lett.B64(1976)159,http://dx.doi.org/10.1016/0370-2693(76)90319-1.

[6] P.Fayet,Spontaneouslybrokensupersymmetrictheoriesofweak,electromag- neticand stronginteractions, Phys.Lett.B69 (1977)489,http://dx.doi.org/

10.1016/0370-2693(77)90852-8.

[7] S.Dimopoulos,H.Georgi,SoftlybrokensupersymmetryandSU(5),Nucl.Phys.

B193(1981)150,http://dx.doi.org/10.1016/0550-3213(81)90522-8.

[8] N. Sakai, Naturalness in supersymmetric GUTS, Z.Phys. C 11 (1981) 153, http://dx.doi.org/10.1007/BF01573998.

[9] K.Inoue,A.Kakuto,H.Komatsu,S.Takeshita,Low-energyparametersandpar- ticlemassesinasupersymmetricgrandunifiedmodel,Prog.Theor.Phys.67 (1982)1889,http://dx.doi.org/10.1143/PTP.67.1889.

[10] K.Inoue,A.Kakuto,H.Komatsu,S.Takeshita,Aspectsofgrandunifiedmodels with softlybrokensupersymmetry,Prog.Theor.Phys.68(1982)927,http://

dx.doi.org/10.1143/PTP.68.927.

[11] K.Inoue,A.Kakuto,H.Komatsu,S.Takeshita,Renormalizationofsupersym- metrybreakingparametersrevisited,Prog.Theor.Phys.71(1984)413,http://

dx.doi.org/10.1143/PTP.71.413.

[12] A. Djouadi, The anatomy ofelectroweak symmetry breaking. II: the Higgs bosonsintheMinimalSupersymmetricModel,Phys.Rep.459(2008)1,http://

dx.doi.org/10.1016/j.physrep.2007.10.005,arXiv:hep-ph/0503173.

[13] Y.Okada,M.Yamaguchi,T.Yanagida,UpperboundofthelightestHiggsboson massintheMinimalSupersymmetricStandardModel,Prog.Theor.Phys. 85 (1991)1,http://dx.doi.org/10.1143/PTP/85.1.1.

[14] J.R. Ellis, G. Ridolfi, F.Zwirner,Radiative corrections to the massesof su- persymmetric Higgsbosons, Phys. Lett. B 257(1991) 83, http://dx.doi.org/

10.1016/0370-2693(91)90863-L.

[15] H.E.Haber,R.Hempfling,CanthemassofthelightestHiggsbosonoftheMini- malSupersymmetricModelbelargerthanmZ?,Phys.Rev.Lett.66(1991)1815, http://dx.doi.org/10.1103/PhysRevLett.66.1815.

[16] M. Carena, J.R. Espinosa, M. Quiros, C.E.M. Wagner, Analytical expressions for radiatively corrected Higgs masses and couplings in the MSSM, Phys.

Lett.B355(1995)209,http://dx.doi.org/10.1016/0370-2693(95)00694-G,arXiv:

hep-ph/9504316.

[17] M.Carena,D.Garcia,U.Nierste,C.E.M.Wagner,EffectiveLagrangianfor the

¯

tbH⁺interactionintheMSSMandchargedHiggsphenomenology,Nucl.Phys.

B577(2000)88,http://dx.doi.org/10.1016/S0550-3213(00)00146-2,arXiv:hep- ph/9912516.

[18] CMSCollaboration,SearchforneutralMSSMHiggsbosonsdecayingtoapair oftau leptonsinpp collisions, J. High EnergyPhys. 10(2014) 160,http://

dx.doi.org/10.1007/JHEP10(2014)160,arXiv:1408.3316.

[19]CMSCollaboration,SearchforneutralMSSMHiggsbosonsdecayingintoapair ofbottomquarks,arXiv:1506.08329,2015.