Electroweak Measurements at NuTeV: A Departure from Prediction

Transcription

1 Electroweak Measurements at NuTeV: A Departure from Prediction Mike Shaevitz Fermilab and Columbia University for the NuTeV Collaboration WIN00 Conference Christchurch, New Zealand January, 00 Introduction to Electroweak Measurements NuTeV Experiment and Technique Experimental and Theoretical Simulation Data Sample and Checks Electroweak Fits Interpretations and Summary 1

2 Electroweak Theory Standard Model SU() U(1) gauge theory unifying weak/em weak Neutral Current interaction Measured physical parameters related to mixing parameter for the couplings, g =g tanθ W g MW e = gsin θw, GF =, = cosθ W 8M M Z Couplings g L g R e, µ, τ 1/ 0 e, µ, τ 1/ + sin θ W sin θ W u, c, t 1/ /3 sin θ W /3 sin θ W d, s, b 1/ + 1/3 sin θ W 1/3 sin θ W W Z Neutrinos are special in SM Only have left-handed weak interactions W ± and Z boson exchange Charged-Current Neutral-Current

3 History of EW Measurements Gargamelle Discovery of the Weak Neutral Current Summer 1973 (Gargamelle, CERN) SM predicted: µ N µ X Gargamelle HPWF CIT-F First Generation EW Experiments Experiments in the late 1970 s Precision at the 10% level Tested basic structure of SM M W,M Z CCFR, CDHS CHARM, CHARM II UA1, UA Petra, Tristan APV, SLAC ed Second Generation EW Experiments Experiments in the late 1980 s Discovery of W,Z boson in Precision at the 1-5% level Radiative corrections become important First limits on the M top NuTeV D0 CDF LEP1 SLD LEPII APV SLAC-E158 Third Generations Experiments Precision below 1% level Test consistency of SM Search for new physics and Constrain M Higgs Predict light Higgs boson (and possibly SUSY) 3

4 Current Era of Precision EW Measurements Precision parameters define the SM: α 1 EM = (40) 45ppb (00ppm@M Z ) G µ = (1) 10 5 GeV - 10ppm M Z = (1) 3ppm Comparisons test the SM and probe for new physics LEP/SLD CDF/D0 N, APV Radiative corrections are large and sensitive to m top and m Higgs M Higgs constrained in SM to be less than 196 GeV at 95%CL 4

5 Are There Cracks? All data suggest a light Higgs except A FB b Global fit has large χ χ =3/15 (9%) A FB b is off about 3σ Γ inv also off by σ N =.9841 ± Higgs Mass Constraint Leptons Quarks g R b too low χ =1.8/5 5

6 NuTeV Adds Another Arena Precision comparable to collider measurements of M W Sensitive to different new physics Different radiative corrections Measurement off the Z pole Exchange is not guaranteed to be a Z Measures neutrino neutral current coupling LEP 1 invisible line width is only other precise measure Sensitive to light quark (u,d) couplings Overlap with APV, Tevatron Z production Tests universality of EW theory over large range of momentum scales 6

7 Neutrino EW Measurement Technique (3) Coupling J weak Coupling ( ( 3) sin θ ) J weak Q em W For an isoscalar target composed of u,d quarks: Llewellyn Smith Relation : R ( ) = σ σ ( ) NC ( ) CC = ρ 1 sin θ W sin 4 θ W (1 + σ σ ( ) CC ( ) CC ) NC/CC ratio easiest to measure experimentally but... Need to correct for non-isoscalar target, radiative corrections, heavy quark effects, higher twists Many SF dependencies and systematic uncertainties cancel Major theoretical uncertainty m c Suppress CC wrt NC 7

8 Charm Mass Effects Charged-Current Production of Charm Neutral-Current CC is suppressed due to final state c-quark Need to know s-quark sea and m c Modeled with leading-order slow-rescaling ( Q m ) Q c x = M ξ = + M Measured by NuTeV/CCFR using dimuon events (N µ cx µµx) (M. Goncharov et al., Phys. Rev. D64: 11006,001 and A.O. Bazarko et al., Z.Phys.C65: ,1995) 8

9 Before NuTEV N experiments had hit a brick wall in precision Due to systematic uncertainties (i.e. m c...) on shell MW sin θw = 1 M M W Z = 0.77 ± = ± 0.19 GeV (All experiments corrected to NuTeV/CCFR m c and to large M top > M W ) 9

10 NuTeV s Technique Cross section differences remove sea quark contributions Reduce uncertainties from charm production and sea Paschos - Wolfenstein Relation R = σ σ NC CC σ σ NC CC = ρ 1 ( + sin θ ) W = R rr 1 r ( µ dsea ) σ ( µ d sea ) = 0 Only dvalence contribute ( µ u sea ) σ ( µ usea ) = 0 Only uvalence contribute ( s ) σ ( µ s sea ) = 0 No strange sea contribution σ σ σ µ sea R manifestly insensitive to sea quarks Charm and strange sea error negligible ( Need to ha ve xs( x) = xs( x) ) Charm production small since only enters from d V quarks only which is Cabbibo suppressed and at high-x Note : NuTeVmeasuresR usedsimultaneously, is equivalent and R to which, when R. But R requires separate and beams NuTeV SSQT (Sign-selected Quad Train) 10

11 NuTeV Sign-Selected Beamline Dipoles make sign selection - Set / type - Remove e from K long (Bkgnd in previous exps.) Beam is almost pure or ( in mode , in mode ) Beam only has 1.6% electron neutrinos Important background for isolating true NC event 11

12 NuTeV Lab E Neutrino Detector 690 ton target Target / Calorimeter Toroid Spectrometer 168 Fe plates (3m 3m 5.1cm) 84 liquid scintillation counters Trigger the detector Measure: Visible energy interaction point Event length 4 drift chambers Localize transverse vertex Solid Fe magnet Measures µ momentum/charge P T =.4 GeV for δp/p 10% Continuous Test Beam simultaneous with runs - Hadron, muon, electron beams - Map toroid and calorimeter response 1

13 NuTeV Detector Picture from Detector is now dismantled 13

14 NuTeV Collaboration Cincinnati 1, Columbia, Fermilab 3, Kansas State 4, Northwestern 5, Oregon 6, Pittsburgh 7, Rochester 8 (Co-spokepersons: B.Bernstein, M.Shaevitz) 14

15 Neutral Current / Charged Current Event Separation Separate NC and CC events statistically based on the event length defined in terms of # counters traversed R exp = SHORT events LONG events (measure this ratio in both = L L > L L cut cut = and NC Candidates CC Candidates modes) 15

16 NuTeV Data Sample Events selections: Require Hadronic Energy, E Had > 0 GeV Require Event Vertex with fiducial volume target - calorimeter toroid spectrometer Data with these cuts: 1.6 million events 351 thousand events Data/MC Data/MC Events Events Ehad (GeV) Ehad (GeV) 16

17 Determine R exp : The Short to Long Ratio: Use E had dependent L cut to minimize short CC correction L cut Lcut = 16 L cut = 17 Events Events L cut Region L cut = 18 L cut Region Short (NC) Events Long (CC) Events R exp =Short/Long Neutrino 457K 1167K Antineutrino 101K 50K

18 From R exp to R Need detailed Monte Carlo to relate R exp to R n and sin q W Cross Section Model LO pdfs (CCFR) Radiative corrections Isoscalar corrections Heavy quark corrections R Long Higher twist corrections Detector Response CC NC cross-talk Beam contamination Muon simulation Calibrations Event vertex effects Neutrino Flux µ and e flux Analysis goal is use data directly to set and check the Monte Carlo simulation 18

19 Background Corrections Short µ CC s (0%, 10% ) muon exits, range out at high y Short e CC s (5%) e N e X Cosmic Rays (0.9%/4.7%) Long µ NC s (0.7%) punch-through effects 19

20 Key Elements of Monte Carlo Parton Distribution Model Needed to correct for details of the PDF model Needed to model cross over from short µ CC events Neutrino fluxes µ, e, µ, e in the two running modes Electron neutrino CC events always look short Shower Length Modeling Needed to correct for short events that look long Detector response vs energy, position, and time Test beam running throughout experiment crucial Source Top Five Largest Corrections δr exp δr exp Comments Short CC Background Check medium length events Electron Neutrinos Direct check from data EM Radiative Correction Well understood Heavy m c R technique Cosmic-ray Background Direct from data Compare to statistical error ± ±

21 Use Enhanced LO Cross-Section NCand CC quark modelfor / cross - sections needs: q( x, Q ) and q( x, Q PDFs extracted from CCFR data exploiting symmetries: Isospin symmetry: u p =d n, d p =u u, and strange = anti-strange Data-driven: uncertainties come from measurements ) Neutrino xsec vs y at 190 GeV Antineutrino xsec vs y at 190 GeV CCFR Data LO quark-parton model tuned to agree with data: Heavy quark production suppression and strange sea (CCFR/NuTeV N µ + µ X data) R L, F higher twist (from fits to SLAC, BCDMS) d/u constraints from NMC, NUSEA(E866) data Charm sea from EMC F cc This tuning of model is crucial for the analysis 1

22 NuTeV Neutrino Flux Use beam Monte Carlo simulation tuned to match the observed µ spectrum Tuning needed to correct for uncertainties in SSQT alignment and particle production at primary target Data vs Monte Carlo E Spectrum Events Events Simulation is very good but needs small tweaks at the 0.3 3% level for E π, E K, K/π

23 Charged-Current Control Sample Medium length events (L>30 cntrs) check modeling and simulation of Short charged-currents sample Similar kinematics and hadronic energy distribution E Had (GeV) Good agreement between data and MC for the medium length events. 3

24 NuTeV Electron Neutrino Flux Approximately 5% of short events are e CC events Main e source is K ± decay (93% / 70%) Others include K L,S (4%/18%) reduced by SSQT and Charm (%/9%) Main uncertainty is K ± e3 branching ratio (known to 1.4%)!! 1.7% 98% 1.6% 98% But also have direct e measurement techniques. 4

25 Direct Measurments of e Flux 1. µ CC (wrong-sign) events in antineutrino running constrain charm and K L production. Shower shape analysis can statistically pick out events (80 < E < 180 GeV) N meas / N MC :1.05 ± 0.03 ( 1.01± 0.04 ( e e ) ) 3. e from very short events (E > 180 GeV) Precise measurement of e in tail region of flux Observe 35% more e than predicted above 180 GeV, and a smaller excess in beam Conclude that we should require E had < 180 GeV Data MC E Had (GeV) NuTeV preliminary result did not have this cut shifts sin θ W by

26 R exp Stability Tests vs. Experimental Parameters Verify systematic uncertainties with data to Monte Carlo comparisons a function of exp. variables. Longitudinal Vertex: checks detector uniformity χ = 50 / 6 dof χ = 49 / 6 dof Note: Shift from zero is because NuTeV result differs from Standard Model 6

27 Stability Tests (cont d) R exp vs. length cut: Check NC CC separation syst. 16,17,18 L cut is default: tighten loosen selection Yellow band is stat error R exp vs. radial bin: Check corrections for e and short CC which change with radius. 7

28 Distributions vs. E had Short Events (NC Cand.) vs E had Long Events (CC Cand.) vs E had 8

29 Stability Test: R exp vs E Had Short/Long Ratio vs E Had checks stability of final measurement over full kinematic region Checks almost everything: backgrounds, flux, detector modeling, cross section model,... exp exp E Had (GeV) E Had (GeV) 9

30 Fit for sin θ W R ( ) = σ σ ( ) NC ( ) CC = ρ 0 1 sin θ W sin 4 θ W (1 + σ σ ( ) CC ( ) CC ) d sin R dr exp exp θ W sin large θ W R d sin exp dr exp θ W small systematics (i.e. m c ) Simultaneous fit of Also input m c sin R θ exp W and R and m exp c to two parameters: = 1.38 ± 0.14 from dimuon measurements This fit is equivalent to using R - in reducing systematic uncertainty Result : sin m c θ ( on shell ) W = 0.77 ± ( stat.) = 1.3 ± 0.09( stat.) ± 0.06( syst.) ± ( syst.) Can also do a two parameter fit to ρ and sin ( on shell ) sin θw = 0.65 ± ρ = ± ( Correlation Coef. 0 θ W : = 085). 30

31 Uncertainties in Measurement sin θ W error statistically dominated R technique R uncertainty dominated by theory model 31

32 NuTeV Technique Gives Reduced Uncertainties 4 better 6 better 3

33 The Result sin ( on shell ) θ W NuTeV result: = 0.77 ± ± ( stat.) ± ( syst.) = 0.77 ± Error is statistics dominated - Is.3 more precise than previous N experiments where sin θ W = 0.77± and syst. dominated Standard model fit (LEPEWWG): 0.7 ± A 3σ discrepancy... R R exp exp = = ± ± ( SM ( SM : ) : ) 3σ difference Good agreement 33

34 Comparison to M W Measurements sin θ ( on shell ) W 1 M M W Z Extract M W from NuTeV sin θ W value M W = ± GeV QCD and electroweak radiative corrections are small Precision comparable to collider measurements but value is smaller 34

35 SM Global Fit with NuTeV sin θ W (Courtesy M. Grunewald, LEPEWWG) Without NuTeV: χ /dof = 1.5/14, probability of 9.0% With NuTeV: χ /dof = 30.5/15, probability of 1.0% Upper m Higgs limit weakens slightly GeV 35

36 Possible Interpretations Changes in Standard Model Fits Change PDF sets Change M Higgs Old Physics Interpretations: QCD Violations of isospin symmetry Strange vs anti-strange quark asymmetry Are s Different? Special couplings to new particles Majorana neutrino effects New Physics Interpretations New Z or lepto-quark exchanges New particle loop corrections 36

37 g g L R u = u Standard Model Fits to Quark Couplings L R d d L R eff eff ( g L ) and ( g ) For an isoscalar target, then couplings = + + = ρ = ρ Two parameter fit to R g L = ± g = ± R 1 4 ( sin θw + 9 θw ) 5 sin 5 4 ( sin θ ) 9 exp W R and Rexp : ( SM : 0.304) ( SM : ) are :.6σ difference agreement Example variations with LO/NLO PDF Sets (no NLO m c effects) (S.Davidson et al. hep-ph/01130) Difficult to explain discrepancy with SM using: Parton distributions or LO vs NLO or Electroweak radiative corrections: heavy m Higgs 37

38 Old Physics Interpretations: QCD R R techniquecould besensitive to q / q differences : = g L g R xdx + p n p n {( uval dval) ( dval uval) + ( c c) ( s s) } p p xdx ( uval + dval) { 3( g g ) + ( g g )} + Lu Ru Ld p p xdx ( uval + dval) p {( u n d p ) ( d n u ) + ( c c) ( s s) } Valencequark momentumfraction xdx val val could explain the NuTeVvs SM difference val val Rd Isospin symmetry assumption: u p =d n and d p =u n Expect violations around (m u -m d )/Λ QCD 1% δsin θ W = Model dependent: Bag Models, Meson Cloud Models,... give small δsin θ W of this order. (Thomas et al., PL A9 1799, Cao et al., PhysRev C ) Strange vs anti-strange quark asymmetry s = xdx s s The number of strange vs anti-strange needs to be the same but the momentum distributions could differ. An asymmetry of s = 0.00 gives δsin θ W = CCFR/NuTeV -dimuons limit the size of s << 0.00 (M.. Goncharov et al., Phys. Rev. D64: 11006,001) ( ) 38

39 Are s Different? NuTeVresult fit asa change in the / coupling ρ0 = ± 0.006( stat.) ± 0.003( syst.) LEP 1 measures Z lineshape and partial decay widths to infer the number of neutrinos N = Γ ( Z ) 3 exp = 3 ( ± 0.008) 1.9σ Γ ( Z ) SM low ( vvγ ) If neutrinos are Majorana, they may have different fundamental couplings from other particles to an extra U(1) type Z - Majorana neutrinos could have zero charge wrt to extra U(1) - Can this explain why charged leptons are different from s? 39

40 New Physics Interpretations Z, LQ,... exchange NuTeV needs LL enhanced relative to LR coupling Oblique (propagator) corrections Constrained by SM fits Gauge boson interactions Allow generic couplings Example: Extra Z boson Mixing with E(6) Z Z =Z χ cosβ + Z ψ cosβ LEP/SLC mix<10-3 Hard to accommodate entire NuTeV discrepancy. Global fits somewhat better with E(6) Z included Example: Erler and Langacker: SM χ 7.5 m Z =600 GeV, mixing ~10-3, β 1. Almost sequential Z with opposite coupling NuTeV would want m Z ~1. TeV CDF/D0 Limits: m Z > 700 GeV (Cho et al., Nucl.Phys.B531, 65.; Zeppenfeld and Cheung, hep-ph/981077; Langacker et al., Rev.Mod.Phys.64,87; Davidson et al., hep-ph/01130.) 40

41 Recent Summary of Possible Interpretations S.Davidson, S.Forte,P.Gambino,N.Rius,A.Strumia (hep-ph/01130) QCD effects: Small asymmetry in momentum carried by strange vs antistrange quarks CCFR/NuTeV dimuons limits Small isospin violation in PDFs expected to be small Propagator and coupling corrections to SM gauge bosons: Small compared to effect Hard to change only Z MSSM: Loop corrections wrong sign and small compared to NuTeV Contact Interactions: Left-handed quark-quark-lepton-lepton vertices, ε LL qq, with strength 0.01 of the weak interaction Look Tevatron Run II Leptoquarks: SU() L triplet with non-degenerate masses can fit NuTeV and evade π decay constraints Extra U(1) vector bosons: An unmixed Z with B-3L µ symmetry can explain NuTeV Mass: 600 < M Z < 5000 GeV or 1 < M Z < 10 GeV Light Z may relate to: GZK cutoff UHE cosmic-rays ( qq) 41 Source of heavy neutral leptons: NuTeV anomalous dimuon signal.

42 Summary NuTeV measurement has the precision to be important for SM electroweak test For NuTeV the SM predicts 0.7 ± but we measure sin θ W (on-shell) = 0.77 ± (stat.) ± (syst.) (Previous neutrino measurements gave 0.77 ± ) In comparison to the Standard Model The NuTeV data prefers a lower effective left-handed quark coupling The discrepancy with the Standard Model could be related to: Quark model uncertainties but looks like only partially and / or Possibly new physics that is associated with neutrinos and interactions with left-handed quarks 4