Search for Invisible Decay of Higgs boson at LHC

Transcription

1 Search for Invisible Decay of Higgs boson at LHC Kajari Mazumdar Tata Institute of Fundamental Research Mumbai DHEP seminar, TIFR, Mumbai

2 Plan of the talk Preliminaries Status of measurements for standard model Higgs boson Search for invisible decay of Higgs boson produced in -- vector boson fusion process -- association with Z Combined results Phenomenological implications: one example DHEP seminar 2

3 It was not that difficult! DHEP seminar 3

4 What was done at LHC Cross section in nano-barn Event rate : dn/dt = σ. L Maximum L = 7.5 *10 33 /cm 2 /s At s = 8 TeV, production cross section for 125 GeV Higgs boson ~ 20 pb ~ 500 Higgs events/hr.! Efficiency of actual measurement << DHEP seminar 4

5 Basic principles of the experiments The detectors for general-purpose experiments, ATLAS and CMS must be able to detect as many particles and signatures as possible: e, µ, τ, ν, γ, jets, b-quarks,. Should also cover largest possible 4π geometry. Momentum / charge of tracks and secondary vertices (e.g. from b-quark decays) are measured in central tracker (Silicon layers). Energy and positions of electrons and photons measured in high resolution electromagnetic calorimeters. Energy and position of hadrons and jets measured mainly in hadronic calorimeters Muons identified and momentum measured in external muon spectrometer (+central tracker) Neutrinos detected and measured through measurement of missing transverse energy (E T miss ) in calorimeters (hermeticity good resolution) DHEP seminar

6 Stairways to Hell/Heaven? DHEP seminar 6

7 Higgs production at LHC gluon-gluon fusion Vector boson fusion Associated productions with W, Z, top DHEP seminar 7

8 Relevant points about the production process The dominant process, ggf (87%), at leading order, produces only longitudinally boosted (if x 1 x 2 ) Higgs boson. Higher order corrections additional activity (jets) in the event Higgs has non-zero transverse momentum (p T ), need not be too high VBF produces 2 jets along with Higgs, with distinctive topology. Associated production: Higgs likely to have non-zero and significant p T demanding transversely boosted Higgs reduces relevant backgrounds Potentially contribute to H+ 2jets category, if W/Z decays hadronically. Typically experiments utilize leptonic decay modes of W, Z Categorize events according to number of jets (0,1,2) boost sensitivity. H+2 jet category can be even subdivided according to the conformity of jets to VBF like criteria or not. Events can also be categorized acc. to Higgs p T in 0,1 jet category DHEP seminar 8

9 Decay of standard model Higgs boson Decay variety depends on mass Fortuitous value of m H! Fermionic decays are possible to observe Chance to study the Yukawa coupling Hff At 125 GeV, total decay width Γ H ~ 4 MeV Experimental resolution for photon and leptonic channels crucial. Branching ratios (in %)for 125 GeV standard model Higgs boson Some of the decay modes have not yet reached the sensitivity at the level of standard model production rate. If observed, in collected data beyond standard model effect H WW* : 23 H ZZ* :2.9 H bb : 56 H cc: 2.8 H ττ : 6.2 H µµ: H gg : 8.5 H γγ : 0.23 H γ Z : DHEP seminar 9

10 After the discovery Remarkable progress in understanding the Higgs sector Observation in boson channels : WW, ZZ, γγ Evidence in fermion channels: Channel Significance (σ) Best-fit mh=125gev Expected Observed μ VH bb ±0.5 H ττ ±0.27 Combined ±0.24 σ obs /σ SM = µ Spin-parity studies : no real measurement. Experiments have tested alternative hypotheses utilizing kinematics of decay products: O + or something different? J P hypothesis of 0 + favoured Looks more and more like a standard model Higgs boson! DHEP seminar

11 Property measurement-signal strength Given the position of resonance Higgs couplings are well defined theoretically. Define σ obs /σ SM = µ measure of signal strength compared to SM expectation. Narrow width approximation simplifies the problems to event yield only. g ggh g HWW g HWW g HZZ g HWW Within errors, couplings are equal to standard model expectations DHEP seminar 11

12 H VV Couplings to bosons and fermions H γγ H ff H Loop induced couplings & higher dimensional operators in other couplings can receive contributions from BSM physics as well couplings can be expressed in terms of separate SM and BSM parts DHEP seminar 12

13 Property measurement: mass and width Combined mass value using accurate measurements in H ZZ and γγ channels ATLAS: ± 0.2 ± 0.6 GeV CMS : ± 0.23 ± 0.3 GeV Given the mass, now use global fit to measurements to estimate total decay width while fixing unmeasured modes to SM predictions. Allow additional new physics contribution to γγ and gg loops. Higgs coupling to SM particles could be different than prescribed in SM rescale the branching ratios of all decay modes. Γ SM H ~4 MeV Non-Standard Model Higgs decay modes will modify the total Higgs decay width Γ H total = Σ ι Γ i H + Γ H BSM Γ i H = partial width in SM for ith decay mode. CMS measurements: H ZZ 4l: Γ H < 3.4 GeV H γγ : Γ H < 6.9 GeV Recently using off-shell production of H ZZ 4l and 2l 2ν Γ H < 5.3 * Γ SM H ~ 22 MeV! CMS-HIG DHEP seminar 13

14 Invisible decay of Higgs boson Is invisible decay of Higgs boson possible? Certainly! Higgs can decay invisibly in SM via process H ZZ* νννν (~0.1%) Current LHC results do not exclude the possibility of a sizeable decay branching ratio to invisible particles of the discovered Higgs boson candidate. eg, H to stable or long-lived particle(s) which can t be detected in experiments H decays to a pair of dark matter (DM) particles which have very low interaction probability with SM particles. Even H could be a mediator between SM particles and DM: Higgs portal model H plays a role in early evolution of the universe. Since we have not yet found a candidate for DM, search for Invisible Higgs decay is a highly relevant pursuit! 95% confidence level upper limit on invisible branching ratio from global fits. ATLAS: Br(H inv.) < 41% (expected < 55%) CMS: Br(H inv.) < 52% (expected <67%) DHEP seminar 14

15 How to directly search for invisible Higgs boson Issues: H inv. will produce large missing energy in the event. For triggering any event a visible object in the detector is needed For invisible decay of H, dominant ggf process can t be utilized! UNLESS, experimentally we can tag a mono-jet or mono-photon event due to initial state QCD or, QED radiation. Way out: use production mode of type H+X and event can be triggered with X. 1. VBF process: has higher rate than associated Higgs (VH, tth) production Events are tagged with distinctive topology of 2 jets to combat large backgrounds. 2. ZH process where H recoils against Z whose visible decay can be reconstructed. Estimate main backgrounds using data-driven methods. Perform simple counting experiment or shape analysis Search for excess over expected SM background in a suitable distribution We discuss today in a bit detail the VBF analysis first DHEP seminar 15

16 Searches for Invisible Higgs boson Invisible decay of Higgs boson has been a favourite topic for a long time Several works done at TIFR and other parts of India since long phenomenological as well as experimental search strategy for LHC, including estimate of discovery potential and now in analysis of collision data. Experimental bounds: Direct searches assume Higgs decay predominantly through Invisible mode. 1. LEP (e + e - Z* ZH) : Invisible Higgs mass > GeV 2. LHC: utilize different production modes Search in mass range between 115 to 300 GeV. ATLAS: ZH (Z ll, H inv.) arxiv: % CL Upper limit on branching fraction for H invisible : 0.75 CMS i) ZH (Z ll, H inv.) and ZH (Z bb, H inv.) PAS-HIG , PAS-HIG ii) VBF (qq qqh, H inv.) PAS-HIG Sanjeev Kumar, KM + others CMS combined NEW PAS-HIG DHEP seminar 16

17 VBF H(inv) Candidate event display Final state of VBF H inv. process: Only 2 jets and missing transverse energy (E T mis ) Major backgrounds: Z ( νν)+jets, W+jets due to QCD and electroweak processes and QCD production of multijets Minor backgrounds tt, WW, ZZ, WZ, DHEP seminar 17

18 VBF process: selection of events To start with, look for inclusive dijet + E T miss final state daunting QCD multijet process, ~ 10 8 times higher rate Apply lenient (loose) VBF-like criteria at trigger + more stringent conditions at analysis level. Use event characteristics to discriminate against backgrounds 2 jets with p T >40 (50) GeV, η <4.7 η1*η2<0 Signal like topology η jj >3.5 (4.2) M jj > 800 (1000) GeV Reject W/Z+jets Veto events with leptons, if moderately isolated with p T > 10 GeV, η <2.1 E mis T > 65 (130) GeV Reject QCD multi-jets Central jet Veto: reject events having a 3 rd jet with p T >30 GeV, η 1 <η 3 <η 2 Final event count in signal region φ jj < 1.0 rad. To define control region for background estimation from data: identify leptons from W,Z Accept events with 2 strictly isolated muons, Recalculate E mis t after excluding muons DHEP seminar 18

19 Estimation of W ( lν)+2jets background Estimate the number in signal region from the number in background region. Keep only 1 isolated lepton in control region, veto additional leptons. Number of W lν (l= e, µ) events in the signal region, where the lepton is not identified. N s e = 62.7 ± 8.7 (stat) ± 18.1(syst.), N s µ = 66.8 ± 5.2 (stat.) ± 15.7(syst.) W τ ν, where τ decays through hadrons has to be treated differently, since no veto has been applied for hadronic decays of τ. But central jet veto is applied for selecting signal sample! Define τ control region by accepting one hadronic decay of τ and no central jet veto ( efficiency for W τ had selection =16%) For signal region apply central jet veto efficiency (42%) N s τ = 53 ± 18 (stat.) ± 18 (syst.) DHEP seminar 19

20 Estimation of Z ( νν) + 2jets background Use Z ( µµ /ee )+ 2jets background control sample to estimate rate of Z νν + 2jets background Select dimuon events (no 3 rd hard lepton) with invariant mass between GeV. Define missing transverse energy without counting the muons Apply VBF criteria on jets E mis T > 130 GeV Take into account different cross-sections and efficiencies for QCD and EW productions of Z+2 jets. N C µµ obs : Observed dimuon + dijet events satisfying above criteria = 12 N C bkg : backgrounds in dimuon + dijet events satisfying above criteria < 1 Other µµ events, e.g., tt, WW, WZ, ZZ ( estimated from monte carlo ) From MC estimate efficiency factors ε s ZMC :dimuon selection efficiency in signal region = (1.65 ± 0.27) * 10-6 ε c ZMC :dimuon selection efficiency in control region = (1.11 ± 0.17) * 10-6 N s νν = 99 ± 29 (stat) ± 25 (syst) DHEP seminar 20

21 Uncertainties in V+jets estimates The number of background events in signal region are derived using a control region Use efficiency factors derived from monte carlo (which are at most at NLO) Efficiency values from Madgraph generator compared with that of MCFM package There are significant differences! Ratio of VBF efficiencies Z νν jj / Z µµjj = 1.28 (MCFM LO), 1.14 (MCFM NLO) = 1.2 (Madgraph) Madgraph events reweighted according to MCFM a la method developed by CMS. Add 20% theoretical uncertainty on the estimates of all V+jets backgrounds. Consistency check of data-driven estimates Eg. Lepton η distribution agrees very well. Important since extrapolation is using leptons (e, µ) within acceptance to leptons outside (Z νν). Consistency check for Central jet veto DHEP seminar 21

22 Estimation of QCD multijet background from data ABCD method based on E mis T and Central Jet Veto selections Assume 2 variables are uncorrelated. N_D = N_B*N_C /N_A A: fail MET, fail CJV B: pass MET, fail CJV C: fail MET pass CJV D: pass MET, pass CJV (signal region) E mis T > 130 GeV, no 3 rd jet above p T >30 GeV First, closure test in control region φ > 2.6 rad.. Prediction in region D: 2959, Observed: 2551 assign a systematic uncertainty of 40%, to take into account possible correlation between 2 variables. signal region φ< 1 rad. : estimated no. of QCD events : 31 Confirmation using E t mis and φ variables DHEP seminar 22

23 N-1 plots Estimate of signal & background events in signal region DHEP seminar 23

24 Control and signal regions Count events here DHEP seminar 24

25 Systematic uncertainties in VBF analysis DHEP seminar 25

26 Limits from VBF channel CMS-Hig Observed(expected) 95% CL upper limit on Br. (H inv): 0.65 (0.49) DHEP seminar 26

27 ZH process with Z l+l-, H invisible Signature: 2 energetic isolated, same flavour, opposite sign leptons within Z-mass window + E T mis l l Main backgrounds:zz 2l 2ν, WZ lν ll, WW lν lν, tt, W+jets, Z/γ* ll Signal characteristics: Recoil of H against Z Projection of E T mis in direction parallel to the direction of dilepton is significantly large (reduced E t mis ) Opening between leptons is small ( φ ll ) Opening between E T mis and dilepton system large Dilepton p T balances E T mis Shape based discrimination fit background and expected signal Define Transverse mass of dilepton + E T mis system: 2D shape analysis for 8 TeV data : m T vs. φ l in various categories (0,1 jets and e, µ) 1d analysis for 7 TeV (less statistics) DHEP seminar 27

28 CMS ZH analysis Observed(expected) 95% CL DHEP upper seminar limit on Br. (H inv): 75% (91%) 28

29 ATLAS analysis of ZH, Z ll, H invisible arxiv: Search for excess in distribution of E T mis Observed 95% CL upper limit on Br. (H inv): 75% DHEP seminar 29

30 CMS analysis of Z bb, H inv. Analyse events with 2 identified b-jets + E T mis Backgrounds: W/Z + jets, tt, WW, ZZ, WZ, QCD multijets Analysis similar to Z νν, H bb, but M bb excess at m H Fit shape of output of boosted decision tree analysis in mass range GeV Inputs to BDT: dijet mass, transverse momentum, b-jet id, separation with E t mis Cross check with transverse mass distribn. Obtained with stringently selected sample Observed(expected) 95% CL upper limit on Br. (H inv): 1.82 (1.99) DHEP seminar 30

31 Combined Limits CMS: 95% confidence level upper limit on branching ratio to invisible decay H invisible, by combining ZH (H ll or bb) and VBF production modes: 58(44)% observed (expected) Indirect limit from total width:< 52(56)% ATLAS: 75(62)% observed (expected) in Z( ll)h Combining direct and indirect analyses Br(H inv.) < 37 (39)% DHEP seminar 31

32 Dark matter interpretation of invisible width Higgs portal model Higgs is the mediator between SM particles and DM particle Dark matter candidate couple mainly to standard model Higgs boson Complementary search for DM through invisible decay of Higgs Look for H χχ using invisible branching ratio There is no sensitivity to these models once the mass of the DM candidate exceeds m H /2. Partial width G inv depends on spin (single DM candidate is considered and is either a scalar, a vector or a Majorana fermion, eg.. In direct detection expt. DM-nucleon scattering takes place via Higgs exchange For a give mass of dark matter an upper bound on invisible decay of Higgs puts a limit on the upper bound on the DM-nucleon cross section DHEP seminar m N : nucleon mass, f N : Higgs-nucleon coupling parameter = from lattice calculation of MILC collabroation 32

33 Upper limit on spin-independent DM-nucleon cross section Note: 90% CL UL is used for all experimental measurements. CMS limit: < 51% DHEP seminar 33

34 Conclusion Run1 of LHC was a success! Since the discovery Higgs boson, its properties are being established with increasing precision. No significant deviation observed so far from SM predictions. Invisible decay of the Higgs bosons predicted in beyond standard model scenarios have been searched. Results are negative. Upper limits on σxbr (H invisible) /σ SM estimated at 95% CL In ZH channel observed sensitivity < 0.75 by ATLAS & CMS experiments, using Z µµ CMS limits using Z bb is 1.82 In VBF channel CMS observed (expected) sensitivity < 0.58 (0.49) CMS observed (expected) combined limit at 95% CL: σxbr (H invisible) /σ SM < 0.58 (0.44) LHC Run2 starts in 2015 at CM energy 13 TeV and higher luminosity. It will be an even more interesting time! Stay tuned! DHEP seminar 34

35 Backup DHEP seminar 35

36 Variables for BDT training for Z(bb) H(inv.) analysis in CMS DHEP seminar 36

37 Pre-fit observed data compared with signal (assuming 100% Branching for H inv. and background in Z( ll) H( invisible) analysis in CMS DHEP seminar 37

38 Z( bb) H ( invisible) Selection criteria DHEP seminar 38

39 Z( bb) H ( invisible) Inputs to Boosted Decision Tree analysis DHEP seminar 39