Prospects for a LowMass Higgs Boson Thomas R
Prospects for a Low-Mass Higgs Boson Thomas R. Junk University of Illinois at Urbana-Champaign C 2 CR 07, February 25, 2007 • • Theoretical Motivation What we Know from LEP and SLC Higgs Searches at the Tevatron Higgs Searches at the LHC Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 1
The Goal: Test the Hypothesis of Spontaneous Symmetry Breaking • SU(2)L U(1)Y is extremely well tested in collider experiments • But it cannot be a symmetry of our vacuum – otherwise quarks, leptons, and gauge bosons would all be massless • Simplest model – one complex doublet with a VEV – W and Z get three DOF, one left over: fundamental scalar HSM • Not the only possibility • SUSY Higgs • General 2 HDM • Higgs Triplets • Technicolor • Little Higgs • Higgs Boson couples preferentially to massive particles – a window to new physics! Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 2
Why is a Low-Mass Higgs Boson Interesting? Define Low-Mass to be m. H<200 Ge. V or so. It’s preferred by the Electroweak Fits! New LEP+SLD+Tevatron EW Fit New MW measurement from CDF (Jan 2007) See W. Trischuk’s talk Jan 2007’s blue-band plot (does not include CDF m. W) New Fit numbers Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 3
LEP Exclusions for Higgsstrahlung with specific Higgs decay modes SM Decays bb Phys. Lett. B 565 (2003) 61 -75 Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 4
Higgs Coupling to the Z with SM couplings Excluded all the way down to m. H=0 Decay-mode Independent Search in e+e-! Z+anything (CP-even Higgs component!) Including invisible decay products. Mass reach extends down to exactly zero. Many “stealthy Higgs” and continuous-spectrum Higgs models ruled out too. Assumes Higgsstrahlung production. k=1: SM Higgs production rate. Eur. Phys. J C 27 (2003) 311 -329 Separate dedicated searches for CP-even Higgses decaying to bb, , jets, , invisible done at LEP 2. Limits all ~110 Ge. V Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 5
Vacuum Stability and Triviality Bounds on SM m. H Higgs Self-Coupling Diverges with Increasing Energy Finding a Higgs boson with m. H~120 Ge. V is powerful evidence for Te. V-scale New Physics Allowed Mexican Hat Brim Turns Over hep-ph/9708416 m. H too light! Hambye and Riesselman Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 6
The Higgs Bosons of the MSSM • Two Complex Higgs Doublets! Needed to avoid anomalies. • Five Degrees of Freedom plus W+, -, Z 0 longitudinal polarization states • Five scalars predicted: h, H, A, H+, H • CP-conserving models: h, H are CP-even, A is CP-odd • Independent Parameters: • m. A • tan = ratio of VEV’s • • M 0, M 1/2 (parameterizes squark, gaugino masses in CMSSM) • Trilinear couplings A (mostly through stop mixing) • And a real prediction: Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 Let’s test it! 7
Precision Measurements of MW and Mt Favor a Light Higgs Boson Has new m. W from CDF in it Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 8
LEP Exclusions for MSSM Higgs bosons ~114. 4 Ge. V mh-max Scenario ( <0) Eur. Phys. J. C 47: 547 -587, 2006 (and many more scenarios) Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 9
LEP Exclusions are Weaker in CP-Violating Scenarios Eur. Phys. J. C 47: 547 -587, 2006 Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 10
CMSSM Favors m. H <120 Ge. V • Bayesian scan over CMSSM parameter space. • Inputs • Direct LEP 2 Higgs searches • Precision EW • Muon g-2 • WMAP assuming CDM=neutralinos: h 2 • Bs Mixing Rate: MBs • Br(B! s ) • Br(Bs! + -) MSSM h is SM-like for these models (production, decay) hep-ph/0611173 v 2 (Feb. 27, 2007) • Sophisticated MCMC guided search for high-posterior-probability parameter values • CMSSM parameters (flat prior) 50 Ge. V < m 0 < 4 Te. V 50 Ge. V < m 1/2 < 4 TEV |A 0| < 7 Te. V 2 < tan < 62 Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 11
Heavier Higgs Bosons are, Well, Heavier hep-ph/0611173 v 2(Feb. 27, 2007) They could be out of reach at the LHC New estimations of SM br(b! s ) lowered these predictions. More on this later. Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 12
An Old SM Posterior Density Plot from 2001 Jens Erler, hep-ph/0010153 Includes ~3 excess at m. H=115 Ge. V from LEP 2 which now has a p-value of 9% (not even 2 now) But: Precision EW has pulled in the long, high tail. Mt has decreased, MW has risen. Errors are smaller. At that time, precision EW gave No vacuum stability or triviality bounds. Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 13
Sensitivity Projections from 2000 and 2003 hep-ph/0010338 2003: Realistic detector models Data-based backgrounds @ Run II energies Analysis upgrades assumed Acceptance upgrades Sophisticated s/b separation No systematic uncertainties! No WW channel! 1999: Run I extrapolations Attempt at syst. errors: scale with 1/sqrt(L) Includes WW channel MC models of Run II detector performance Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 Fermilab-Pub-03/320 -E 14
The State of the Individual Channels as of ICHEP 2006 Getting close at m. H=160 Ge. V! Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 15
Just a Short While ago, It looked like THIS La Thuile/Moriond 2005 In just over a year, we analyzed 6 the data, optimized the analyses, and combined them together to close in on the SM. Searches gaining staff and momentum! Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 16
CDF and DØ’s Contributions for ICHEP 06 were “Complementary” DØ: H bb (<= 400 pb-1) H WW (1 fb-1) CDF: H bb (1 fb-1) H WW (360 pb-1) Different analyses ready at different times: A snapshot of the horse-race. Much more data to come! Even though CDF+DØ were combined, factor of sqrt(2) not realized for ICHEP 2006 due to non-overlap in most sensitive searches. Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 17
We’ve Come a Long Way Since Run I Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 18
The Old To-Do List, of October 2005 SM Channels only Doesn’t specify what WW people should do. We thought they were pretty optimal already. We are learning there is much more to be gained in WW Much has been learned in the last year. Lots left to learn, and do. Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 19
Newly Released (Jan 2007) CDF H! WW Search with 1 fb-1 • Cut-based for now. • NN and Matrix Element analyses in the works More sensitivity from the same data sample. • To do: Combination with DØ Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 20
Accomplishments and Ideas – Mass Resolution MET resolution too. . Current Status: Many of these we have tools for, but need to convince ourselves in data that they perform as promised. • Many tools available: • NN Jet tools • Hyperball (kind of like an NN) • Track-Cluster Association • B-specific Corrections • Wider Jet Cones (llbb already uses these) • Selection nets implicitly optimize mjj resolution if they include such things as MET and EM fraction and other things a NN needs (llbb). • Double-tagged events have better mjj resolution than singletagged events (less combinatorics, less semileptonic decay) Watch out for multiple collision effects. . Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 21
Sensitivity gain from B-tag Improvements Example: WH! l bb channel *Find the most sensitive b-tagging option a priori ~20% improvement by separating single and double tagging Significance: *S, B are number of events in 3 window in dijet mass distribution ~10% improvement by NN tagging Focus on =1 SECVTX tag w/ NN tag 2 SECVTX tag Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 Still more to gain. 22
Example – Splitting WH! l bb Channel into single and double-tags • Different s/b in the two sub-channels: • lower fraction of mistags, W+jets in double-tagged sample • better mjj resolution • General Strategy – If you can classify events in two or more ways with different s/b, sensitivity is gained by separating them and combining the results. Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 23
A Continuous Job: NN Selection llbb Channel already has a 2 D NN! WH lvbb tried a NN analysis already. Claim: 1. 75 factor in HSWG note; Run I NN gets 1. 2 factor in s/sqrt(b)). Single top NN’s have shown tremendous improvements in sensitivity over simple variables like Mlvb and HT. We learn as we go: Matrix Element Techniques perform very well and give results not 100% correlated with NN’s. But – some measured quantities are not input to the Matrix Element, like NN b-tags. Many kinds of backgrounds lack convincing matrix elements Non-W Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 24
Tevatron Performance Getting Impressive! Higher peak luminosity – mainly due to higher pbar stacking rates. Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 25
And a Concern: Trigger Rates Rise Faster than Luminosity Example: Muon triggers at inermediate : multiply cross-section by instantaneous luminosity to get trigger rate. At least a cubic dependence on luminosity! • Hit deadtime wall at a L 2 trigger rate of ~1 KHz. Fake track rate grows with occupancy. • CDF has a stereo trigger upgrade (L 1 already done, L 2 in test) to reject fake tracks. Works very well! • MET trigger upgrade in progress – sharpen the MET trigger turnon. • L 2 Cluster upgrade Still compromises to be made at Lumi = 300 E 30 cm 2 s-1. • CDF Has a working group dedicated to the Higgs triggers. More high-lumi concerns: • B-tagging degrades with increasing # primary vertices/BX • MET resolution degrades Some good news: • Only a fraction of data collected at the highest luminosities though (beginnings of stores) Would like to make the best use of it, though. Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 26
Expected Signal Significance CDF+DØ vs Luminosity It’s possible to be lucky or unlucky! per experiment m. H=115 Ge. V assumed Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 Date 27
Parton Distribution Functions of the Proton “The LHC is a gluon-gluon collider” (approximately). Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 28
SM Higgs Boson Production Rates at the LHC Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 29
Backgrounds and Signals at the LHC Backgrounds dominated by strong processes ATLAS TDR q q p Signals much smaller l q q W l Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 W H p q q p 30
SM Higgs Discovery Potential with ATLAS discovery channels excluded by LEP v GGF: H gg v GGF: H ZZ 4 l+v GGF: H WW 2(ln) v tt. H: H bb v VBF: H tt v VBF: H WW since 2000 For MH>300 Ge. V also: VBF: H ZZ llnn VBF: WW lnqq ? • Some Channels allow very precise m. H measurement (ZZ, ) • Calibration of backgrounds with data • k-factors increase signal rates (and backgrounds) Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 31
CMS Higgs Search Similar to ATLAS Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 32
Low-mass SM Higgs Boson Searches at the LHC H : Low s/b, but good mass resolution allows for a peak. Large irreducible background: pp X Measure background with sidebands Optimizing signal resolution – conversions, dead material tt. H, H bb: Only channel where H bb is possible (WH, ZH, H bb are hopeless) • Jet Combinatorics • Energy Resolution • Background Rate and Shape (get from data as much as possible) Can reduce Jets faking photons with cuts ATLAS TDR, 100 fb-1 Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 33
A Very Clean Mode: gg H ZZ(*) l+l- at ATLAS, CMS Simulated H ZZ e+e- + Event in ATLAS TDR Expect to measure m. H to 1% with this mode if it’s there. NLO k-factor may improve expected significance by ~25% I Isolation, anti-B veto (lepton impact parameters) Small and flat background: estimate with data use m 4 l to discriminate Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 34
Luminosity Required to Discovery a SM Higgs boson at LHC • VBF H WW strongest channel for 130 < m. H< 200 Ge. V • Sensitivity all the way up to 1 Te. V with H ZZ • m. H<130 Ge. V – several difficult channels in combination, >5 fb-1 needed (N)NLO k-factors increase signal rates Data needed to calibrate backgrounds Analyses are cut-based: Advanced separation techniques to be applied ATLAS Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 35
Higgs Boson Production and Decay in MSSM at High tan • Interesting feature of many MSSM scenarios (but not all!): [mh , m. H] m. A at high tan (most benchmark scenarios. . ) Two for the price of one! • At leading order, (A 0 bb) and (A 0 + -) are both proportional to tan 2. • Decays to W, Z are not enhanced and so Br. falls with increasing tan (even at high m. A) • Br(A 0 bb) ~ 90% and Br(A 0 + -) ~ 10% almost independent of tan (some gg too). Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 36
Higgs Boson Production Mechanisms + 0 t 0 b Amplitude 1/tan suppressed! b 0 Amplitude tan enhanced! And many other diagrams At high tan , (h, A+X) tan 2 b Amplitude tan (low tan and SM case: cross-sections too small to test with current data. ) enhanced! Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 37
CDF’S h! Search ~ 2 excess around m. H=160 New! Released Feb. 2, 2007 Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 38
CDF’s h! Search: Limits and Interpretations ~300 pb-1 results This analysis is one to watch closely: Backgrounds are small for m. H>150 Ge. V. Probing tan of 40! (interesting value) Still a lot of mileage left in this analysis – benefits greatly from luminosity and combination Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 39
DØ Doesn’t See Anything Though Sensitivity is similar to CDF’s analysis, even with the NN Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 40
We Would Like to Observe More than One Higgs Boson ATLAS, 300 fb-1 (VBF 30 fb-1 only in plot) • At least one Higgs boson visible over entire plane • Many models for which only one Higgs boson is detectable. • Decays of H ? Typical CPX scenario: Other CPC Scenarios Similar Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 41
Cascade Decays and a Low-Mass Higgs h! AA! bbbb, bb , , photons Example: Dermisek, Gunion hep-ph/0611197 h! AA! bbbb LEP Interprets h! bb searches in h! bbbb with extra signal MC. No analysis re-optimization. Some NMSSM models with low fine-tuning just sneaking under the LEP exclusion. 3 channel: DØ 4gamma channel: CDF just getting started Still pretty hard at the Tevatron. Maybe taus? But low m. A taus together, not isolated. Look like jets! Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 42
Interesting A 0 decay modes: taus, charm, gluons. Could extend this if we could get someone interested in reading old LEP data. Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 43
A New Exotic Channel pp! H+hf! W+hfhf! W+ DØ Sought this in 3 +X (hf is fermiophobic and decays to photons) DØ Note 5067 -CONF Signal proposed by Akeroyd, Diaz Phys. Rev. D 67: 095007, 2003. hep-ph/0301203 CDF wants to look at the 4 final state. Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 44
Summary and Prospects • SM Higgs Searches at the Tevatron are going full speed ahead • We are getting smarter in many ways (foreseen and unforeseen) • Chance of Evidence/Discovery of a light Higgs boson at the Tevatron Difficult range of m. H for LHC. Light Higgs preferred by EW fits, Preferred by MSSM. • MSSM Higgs Searches at Tevatron • Complementary to LEP Searches • Tau channels benefit greatly from additional data. • LHC: Full coverage of entire allowed SM Higgs mass range • Discovery guaranteed if it’s there! • Measurement of m. H • May only see one of five MSSM Higgses though • MSSM covered for many scenarios with the LHC. Tevatron is testing interesting parts of model space. May need ILC to disentangle ambiguities in LHC discoveries. By 2009 we will know quite a lot about Gauge Symmetry Breaking Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 45
Backup Slides Follow Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 46
Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 47
Vector-Boson Fusion with H WW v signature: Forward tagging jets -2 forward jets with large rapidity gap - only Higgs decay products Higgs Decay f in central part of detector h Jet Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 48
VBF: Forward Jet Reconstruction, Backgrounds, Pileup Signal and Background Discrimination Does it still hold up @NLO? Jet Reconstruction Efficiency Background due to Pileup High Luminosity Low Luminosity Jet Need to see how first data look Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 49
Couplings of MSSM Higgs Bosons Relative to SM W and Z couplings to H, h are suppressed relative to SM (but the sum of squares of h 0, H 0 couplings are the SM coupling). Yukawa couplings (scalar-fermion) can be enhanced Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 50
LHC Sensitivity to MSSM Higgses h 0 (light), H 0, A 0, H± (heavy) Parameter phase: m. A, tan Mh<135 Ge. V SUSY particles heavy: • SM like h 0 , bb, , WW; H 0 4 l • MSSM specific A/H µµ, , tt H hh A Zh H± ± Most of plane explored with 10 fb-1 VBF enhances sensitivity to light MSSM Higgs Heavy Higgs channels dominated with final state Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 will need Linear Collider 51
WH Wbb Signal added atop background a typical expected data outcome for 30 fb-1 at LHC WZ Wbb and backgrounds lepton, 2 b-tagged jets third jet veto. WH Wbb and backgrounds m. H=100 Ge. V Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 52
MSSM Higgs Masses vs. Tan Thomas R. Junk, C 2 CR 07 Feb. 25, 2007 53
- Slides: 53