Transcript Document
Physics with radiative B decays at Babar CERN EP Seminar, June 19th 2006 Wouter Hulsbergen (CERN) (* ) o B physics in a nutshell o radiative B decays o PEP-II and BaBar o b->sγ branching fraction and CP asymmetry o probing the photon polarization with B0->Ksπ0 γ Physics with B decays o the aim of the heavy flavour physics program is to understand the flavour structure of the quark sector of the standard model u c t d s b ~ 0.01 1.2 173 0.01 0.1 4.5 GeV why are there 6 quarks? why are there masses so different? is the standard model ‘CKM’ picture of quark mixing correct? is it the only source of CP violation at low energies? o decays of B hadrons are our richest source of flavour phenomenology B’s are heavy: many different decay modes; decays to both other families long lifetime (cτ=0.5mm), large mixing probability (for B0: τΔm=0.8) large CP violation effects mB>>ΛQCD perturbative QCD works, at least sometimes o B decays also improve our understanding of hadronization (long distance effects), important to extract the short distance physics 2 The CKM matrix o for quarks weak eigenstates are different from mass eigenstates νe leptons e+ d quarks u g g Vud W+ W+ o V is unitary and called the Cabibbo-Kobayashi-Maskawa (CKM) matrix o it has 4 physical parameters, one of which is a complex phase this phase is the “origin of CP violation in the SM” o VCKM is almost diagonal: 1 λ λ λ3 1 λ3 λ2 λ2 1 λ ≈ 0.22 3 The unitarity triangle o the condition that Vckm is unitary can be visualized with triangles in the complex plane, for example o there are 6 such triangles, but this one is most relevant for B0 decays o the angles of this triangle are the famous ‘CKM angles’ *) since the ‘phase’ of a quark field itself is arbitrary, Vckm depends on a phase convention. however, as you can see, these angles do not. 4 The unitarity triangle o branching fractions and CP asymmetries provide experimental constraints on different sides and angles of this triangle, for example o we can test the SM quark flavour mixing picture by measuring all sides and angles: overconstraining the triangle measuring a single side or angle in more than one way 5 Where do we stand? this graph shows the constraints used for a ‘global fit’ of all 4 CKM parameters one way to express the consistency is to compare the measured WA for sin(2β) from CP-violation in B0->ψ Ks sin(2β)exp = 0.69 ± 0.03 with the value of a fit that does not use this measurement sin(2β)fit = 0.74 (+0.07 -0.03) o the success of the SM shows that new physics effects in B decays are small ! o therefore, we now concentrate on searches for NP in processes dominated by loop diagrams in the SM, in particular b->s transitions 6 bs and bd transitions o b -> s and b -> d transitions are a Flavour Changing Neutral Current absent in the standard model at tree-level exist only at loop level, for example via a W-top loop Vtb* Vts o SM contribution receives even additional suppression CKM suppression: t- and c-quark loop ~λ2, u-quark loop ~λ4 for radiative decays there is also a helicity suppression o new physics enters ‘at leading order’, for example SUSY o physics at the virtual high-energy frontier! 7 sin(2β) in penguins: a smoking gun for NP? The decay B0->ψKs is dominated by a tree diagram Vcb* b c c ψ B0 Vcs d s d K0 The rare decays to the left all proceed through a b->s gluonic penguin: Vtb* Vts For all these decays, time-dependent CP-violation probes sin(2β) in the SM … Is there new physics in b->s penguins? ‘naïve’ penguin average is ~2.5σ away from SM value Let’s look at radiative b->s decays … 8 Radiative B decays o radiative B decays are b ->s,d transitions with a high energy photon or lepton pair in the final state `radiative penguin’ `WW box’ o in contrast with other b → s decays QCD plays a relatively minor role one can actually calculate something (even formfactors) one can measure inclusive BFs and asymmetries o last point is important because predictions for inclusive decays are more accurate than those for exclusive decays 9 How do we calculate things? The theoretical framework is the operator product expansion (OPE) VtbVtd* x C7γt(mB) x Wilson coefficient: physics above mB calculated pertubatively in SM and beyond this where new physics enters Local operator: physics below mB use Heavy Quark Expansion: systematic expansion in Λ /mB exclusive decays need formfactors o calculations become a ‘double expansion’ in αs and ΛQCD /mB o precise predictions for (a.o.) inclusive b->sγ and b->sl+l- branching fractions CP asymmetries, both inclusive and exclusive polarization of the photon mass and angular distribution of the lepton pair in b->sll 10 Inclusive b->sγ branching fraction o in the Heavy Quark Expansion: BF(B->Xs γ) ≈ BF(b->s γ) + O[ ( ΛQCD / mB )2 ] o NLO SM prediction BF(B→Xs) = (3.57 ± 0.30) x 10-4 (Buras et al 2002) (for recent update, see eg Hurth, Lunghi, Porod 2005) o expected theoretical accuracy 5% within a few years most importantly: scale dependence reduced by going to NNLO o there is a subtlety here connected to the minimum photon energy: convention: ‘inclusive’ means Eγ > 1.6 GeV experiments cannot measure that low: needs extrapolation 11 Direct CP Violating Asymmetries o direct CP violation or ‘charge asymmetry’: o occurs if >= 2 amplitudes with different weak and strong phases PRL93:131801,2004 o well established in B0->K+π- decays WA (HFAG): Acp(K+ π-) =-0.108 +/- 0.017 12 Predictions for direct CP Violation in b->Xγ o b->sγ loops always dominated by top-quark contribution, VtbVts* small asymmetry: Acp = 0.004 ± ~0.002 (Nucl.Phys.B704,2005) o b->dγ ‘up’ contributions VubVud* about as important as ‘top’ VtbVtd* large asymmetry: Acp = -0.10 ± ~0.04 (Nucl.Phys.B704,2005) o in the limit ms=md (‘U-spin’), CP violation vanishes in the SM inclusive asymmetry Acp(B->Xdγ+B->Xsγ) ≈ 0 corrections are of order (ms/mb)2 times small CKM factors o for exclusive decays results are more model dependent, but uncertainties not much larger 13 The photon polarization o SM: W couples to left-handed quarks: bL tL sL o photon has spin 1: to conserve helicity in a two body decay, one of the quarks needs to ‘flip helicity’ bRsL γL: spin-flip on the b-quark bLsR γR: spin-flip on the s-quark o the probability for spin-flip is proportional to the quark mass two important consequences: 1. SM: b -> γL and anti-b -> γR opposite helicity suppressed by ms/mb. measurement of polarization is excellent probe for NP 2. anomalous WbRtR coupling would strongly affect the b->sγ branching fraction, because it is enhanced by mt/mb 14 B factories: e+e- Y(4S) BB o B factories operate at the Y(4S) resonance (10.58 GeV) o hadronic cross-sections: uds/cc/ bb = 2.1 / 1.3 / 1.1 nb o in the Y(4S) frame the B mesons are practically at rest need boost to measure decay lengths with high accuracy PEP-II is an asymmetric collider 9.0 GeV electrons vs 3.1 GeV positrons 15 PEP-II and BaBar at SLAC linac PEP-II storage ring SLD BaBar 16 Integrated luminosity when performing well, off-resonance data on-resonance data PEP-II produces about 10 BB event per second since 2000 BaBar has recorded over 300M BB events about 8% of data is taken below the Y(4S) resonance results presented here are based on 80 fb-1 and 210 fb-1 on-resonance data 17 The BaBar detector Electromagnetic Calorimeter 6580 CsI crystals e+ ID, π0 and γ reco Instrumented Flux Return 19 layers of RPCs μ and KL ID Cherenkov Detector (DIRC) 144 quartz bars K, π, p separation 3 GeV positrons Drift Chamber 9 GeV electrons 1.5 T magnet 40 layers, tracking + dE/dx Silicon Vertex Tracker 5 layers of double-sided silicon strips 18 BaBar (artist’s perspective) 19 Examples of samples of exclusive B decays very many: B-> D(*) + N π qqbar continuum very clean: B->charmonium Ks about 0.4% of all produced B very rare: B->Kll BF≈0.4x10-6 bump from other B decays: ‘peaking background’ Distributions show ‘beam-energy substituted mass’: One of main ingredients in multi-dimensional ML fits used for the analysis of exclusive decays 20 B->Xγ at the B-factories Three types of probes o various exclusive final states, such as K*γ, Kππγ, Kφγ, Kη’γ branching fractions, CP asymmetries, isospin asymmetries small systematic uncertainties theory accuracy limited by FF calculations, but improving o semi-inclusive: as a sum-of-exclusive-modes reconstruct B->Kγ + up to 3 or 4 pions, at most 1 pi0 cross-feed between difference B->Xγ modes not negligible larger uncertainty from backgrounds than exclusive analyses extrapolation to fully inclusive BF not trivial o fully inclusive use only the photon as a tag measures really b-> (s+d) γ dominating systematic: background from other B decays 21 What do BXsγ events look like? o one high energy photon and an s-quark ‘jet’ Eγ≈M/2 B M = 5.28 GeV Xs K+ hadrons o not exactly a 2-body decay, because the b quark is bound in a meson and because of higher order corrections (e.g. bsgγ) K*γ o the simplest final state is B->K*(890)γ (about 12%) o the photon spectrum itself is important physics as well: it can be used to extract HQET parameters photon spectrum measured by Babar (PRD 72, 052004) 22 Radiative penguin portrait B+→ K*+γ (K*+->Ksπ+) candidate Muon from other B decay Detached vertex from Ks → ππ High energy photon in EMC π+ from K*+ 23 Measuring BF(B->Xsγ) Simplest technique: count events with one high energy photon Very large backgrounds from o continuum qq-bar (q=u,d,s,c) and ττ-bar qq + ττ BB o other B decays most photons come from π0 and η decays Backgrounds are suppressed by B->Xsγ o explicit veto of π0 and η candidates o ‘tagging’ the other B o using event-shape o analysis does not separate b->dγ (about 4% of total rate) subtracted using theory prediction o B not exactly at rest in CMS: measure ‘Eγ*’ rather than ‘Eγ’ 24 Continuum background suppression Tagging the second B o leptons common in B decays: B->Xlν ~ 20% o leptons are not common in qqbar background require high pT muon or electron Exploiting event shape: o qqbar events are ‘jet-like’ o B events are spherical combine ‘event-shape’ variables in multivariate discriminant 25 Background subtraction Remaining background is subtracted o use ‘off-resonance’ data to subtract continuum. reliability tested in high Eγ sideband o use Monte Carlo to subtract BB background. mostly π0 and η, but also electrons, anti-neutrons, ω, … MC carefully tuned on control samples (for example by reversing π0/η veto) tested in low Eγ sideband BB continuum 26 Result Remaining events in 81/fb, for Eγ = 1.9-2.7 GeV Nsig = 1042 ± 84 (stat) ± 62(syst) Preliminary result (hep-ex/0507001) BF(b->sγ, Eγ>1.9 GeV) = stat o main systematic: BB background subtraction improves with larger control sample size syst model-dependence of efficiency o efficiency depends on Eγ correction model-dependent improves with better measurement of photon energy spectrum o result still requires extrapolation to Eγ = 1.6 GeV leads to additional uncertainty also improves with better measurement of photon energy spectrum 27 Other techniques for measuring BF(b->sγ) 1. as a sum-of-exclusive-modes (BaBar, PRD 72,052004 (2005) ) reconstruct as many final states as possible, about 55% of total kinematic constraint on Eγ: detailed information on photon spectrum dominant systematic: fraction of missing final states (fragmentation) not competitive with fully inclusive measurement in the long run 2. in the recoil of fully reconstructed B decays (no results yet) exploit large sample of about 200,000 fully reconstructed B decays may provide interesting information on fragmentation 28 Measurements of the bsγ branching fraction 10% measurements agree with 10% NLO SM predictions improvements of both to 5% seems feasible: theory: NNLO experiments: towards 10x more statistics agreement already highly constrains new physics! 29 An example: bsγ and the charged Higgs b->sγ receives contribution from charged Higgs in many SM extensions (eg supersymmetry) assuming that only 2 doublet Higgs sector contributes at low energy (2HDM), we rule out a considerable part of the parameter space! m(H+,type-II 2HDM) > ~450 GeV (using technique from Gambino and Misiak) limit is much better than that from direct searches at LEP and Tevatron! of course, exclusion power model dependent … 30 Another example: b→ sγ and top quark couplings new Wtb couplings will also affect BF(bsγ) from hep-ph/9906329, (uses slightly smaller value for SM contribution) new right-handed coupling, enhanced by mt/mb new left-handed coupling 10% error on BF constrains right-handed coupling to be few % of SM coupling 31 Direct CP violation in the inclusive sample o event selection uses lepton tag events are already tagged! B0 contribution ‘mixes’: asymmetry diluted by factor 0.816+/-0.004 cannot subtract b->dγ contribution: measure really Acp[ b -> (s+d)γ ] o statistical precision optimized by reducing Eγ* window to [2.2,2.7] GeV ‘extrapolation’ error for theory predictions better under control than for the branching fraction o events observed: N(l+) = 349 +/- 48 versus N(l-) = 409 +/- 45 o stat. uncertainty much larger than for ‘exclusive’ analyses (next slide) o systematic uncertainty again dominated by B background 32 Summary of Acp measurements inclusive semi-inclusive exclusive o current B-factories cannot reach inclusive Acp better than ~0.03 o semi-inclusive and exclusive below 0.01 seems feasible 33 Intermezzo: time-dependent CP-violation in a nutshell o suppose ‘f’ is a final state that is accessible to both B0 and anti-B0 o the total amplitude contains at least two contributions, one appearing through B0-anti-B0 mixing decay B0 In simpl(istic) words: f anti-B0 o the contribution from ‘mixing’ depends on how long it takes the B to decay relative contribution is ‘time-dependent’ o furthermore, if the contributions have a relative phase, we get CP-violation o the sum of these 2 effects is “time-dependent CP violation” (TDCPV) coefficients between -1 and 1 Mixing frequency B factory specialty: BB decays in ‘entangled state’ flavour of B Δm=0.51/ps o att t=0 Δt Lifetime: τ=1.5ps o flavour at (t=0) anti-flavour of other B at time of decay *) expression becomes more complicate if f is not a CP eigenstate. 34 Example: B->ψKs (the ‘golden’ mode) In the Standard Model: S(ψKs) = sin 2β ≈ 0.7 C(ψKs) = 0 35 Probing the photon polarization through TDCPV o now, consider the decay BXγ with X a CP-eigenstate o just like before, but interference is suppressed (Atwood,Gronou,Soni) XsγR B0 anti-B0 Xs γL o the suppression is proportional to fraction of opposite helicity photons S ≈ - sin 2β x 2ms/mb ≈ -0.04 o NP contributions with different photon polarization enhance interference TDCPV becomes a probe of the size/chirality of NP in bsγ o this works for several final states, but B0->K*0 γ with K*0->Ksπ0 is currently the only experimentally accessible mode o main theoretical uncertainty: contributions from b->sγg (Grinstein,Grossman,Ligetti,Pirjol 2004) 36 Mixing induced CP violation in B→Ks0 o main experimental complications: measure a B vertex with one trajectory (well established by now!) large continuum background: exploit event shape, 0/η vetos background from other B decays (such as other B->Xsγ and B->Ksπ0π0) Ksπ0 mass distribution o in principle, all B-> Ksπ0 γ final states can be used (Atwood, Gershon,Hazumi,Soni) o however above the K*(890) poorly known background from other B decays becomes much larger theoretical uncertainties are larger only small number of events (background subtracted) K*(890) K*(1430) (Belle seems to gain more from including the ‘high-mass’ region than Babar does) 37 Mixing induced CP violation in B→Ks0 signal+background background data (beamenergy-constrained) mass for B->K*(890)γ candidates Δt distributions and asymmetry Result from fit in 210/fb: Nsig = 156 16 S = −0.21 0.40 0.05 C = −0.40 0.23 0.04 (PRD72,051103,2005) Errors still very large: consistent with both SM and ‘no polarization’ 38 Future prospects for this technique non-SM physics is not going to be very large: this method only becomes interesting with errors of about σ(S)≈0.05 super-B-factory statistics also accessible at LHC-b 39 Not today: other important radiative B decays o the b->dγ transition estimated branching fraction few parts in 10-6 experimentally hard because of large background from b->sγ searched for in exclusive decays B0->ρ0γ, B+->ρ+γ and B->ωγ. together with B->K*γ these modes provide a constraint on |Vtd/Vts| o the b->sl+l- transition branching fraction also few parts in 10-6 studied both in exclusive B->K(*) l+l- and ‘semi-inclusive’ B->(Kll + (1-4)π) 3 competing SM amplitudes at leading order: very rich phenomenology: SM provides clean predictions for l+l- mass and angular distribution, the ratio of e+e- to μ+μ-, etc 40 Summary o b->sγ decays are an excellent prove for new physics beyond-the-standard-model physics enters at leading order serious predictions exists for several measurable observables this is the ‘virtual high energy frontier’ o most important measurement: b->sγ branching fraction agreement between data and theory highly constrains new FCNC either NP is very heavy (not very likely) or it is flavour blind (the SUSY flavour problem) or there are accidental cancellations (bad luck) expect considerable improvements in both experiment and theory within next few years o polarization of the photon experimental uncertainties from TDCPV still large other methods exist (angular distribution in Kππγ), but do not look very promising yet o B factories have still much more data coming … stay tuned! 41