Rare B and D Decays at BaBar John Walsh INFN, Sezione di Pisa - PowerPoint PPT Presentation

Rare B and D Decays at BaBar John Walsh INFN, Sezione di Pisa Representing the BaBar Collaboration at the 46th Rencontres de Moriond: Electroweak Interactions and Unified Theories La Thuile, Italy March 13-20, 2011 1

Rare Decays: Window into New Physics • New Physics (NP) can be revealed in low-energy flavour processes • Particularly fruitful areas of investigation are processes with small effects predicted by the Standard Model - Rare decays - Flavour-changing neutral currents (FCNC) • Strong constraints on already put on 2HDM-II NP from these kinds of decays (Haisch, arXiv:0805.2141) Moriond EW, La Thuile, 2011 2 John Walsh, INFN Pisa

Outline • Direct CP asymmetry in B ➝ X (s+d) γ • Rare leptonic decays - B + ➝ τ + ν - - B + ➝ K + τ + τ - • Very rare decays - B 0 ➝ γγ and - D 0 ➝ γγ Moriond EW, La Thuile, 2011 3 John Walsh, INFN Pisa

BaBar Experiment Total Dataset: • Asymmetric B- ~ 470 x 10 6 BB pairs factory: E cms = 10.58 GeV • Performed a wide range of flavour physics results in B, charm and τ sectors • General purpose detector: - precision tracking - photon/electron detection - Particle ID - muon/K L identification Moriond EW, La Thuile, 2011 4 John Walsh, INFN Pisa

A CP in B ➝ X (s+d) γ 5

Direct CP asymmetry in inclusive B → X s γ • This mode has been the “workhorse” of radiative B decay physics — the SM branching fraction ( PRL 98, 022002 (2007) ), is known to 7-8% and deviations from NP can be substantial ( arXiv:0805.2141 Semileptonic Signal provides an interesting summary ) tag decay • Inclusive measurement: γ e - or μ - - B B What? no requirements on X S system, identify events by high-energy photon - Why? because smaller theoretical X uncertainties - X (s+d) Consequence : mode studied is actually B → X (s+d) γ • CP asymmetry is also sensitive to CP Asymmetry NP. SM A CP ~ 10 -6 , while it can be ~ 10% in NP scenarios (Nucl. Phys. B704 (2005) 56-74) Moriond EW, La Thuile, 2011 6 John Walsh, INFN Pisa

A CP in inclusive B → X (s+d) γ Photon Spectrum (MC) Experimental issues: HE photon only • High-energy photon: E γ * >1.5 GeV • Lepton tag: log scale - Electron or muon candidate with p* > 1.05 GeV - E *miss > 0.7 GeV • Continuum background - subtracted using off-resonance Full selection data • BB background - mostly π 0 →γγ , η→γγ ⇒ explicitly vetoed, remaining BG determined from control sample - control samples also used for other sizable BB backgrounds Moriond EW, La Thuile, 2011 7 John Walsh, INFN Pisa

A CP inclusive B → X (s+d) γ : 383 M BB pairs BaBar • Control regions: Preliminary - continuum: On-Off Data = -100±138 events - BB: 1252±272±841 ⇒ 1.4 σ (assumes no signal, where expect 100-400 signal events in low-energy tail) • A CP is insensitive to photon energy cut: optimize ⇒ (2.1-2.8) GeV • Yields: BaBar preliminary: - N(l+) = 2623±158 - N(l-) = 2397±151 A CP =0.056±0.060 stat ±0.018 syst • Account for mistag and bias: Babar, prev. - lepton tag Babar, hadronic tag (B 0 mixing) CLEO, lepton tag (A CP of background) Babar, this analysis Moriond EW, La Thuile, 2011 8 John Walsh, INFN Pisa

B + →τ + ν 9

B + →τ + ν • Powerful probe of NP • Sensitive to charged Higgs boson: Latest SM Value: • Helicity suppression leads to higher BF for τ final state in Inputs: SM: Moriond EW, La Thuile, 2011 10 John Walsh, INFN Pisa

B + ➝ τ + ν : Experimental technique Tag efficiency ~ 0.2% • Few kinematic handles to Hadronic tag Signal decay reconstruct B + ➝ τ + ν • Reconstruct “other B” completely B - π + B + τ + D (*) - remove many tracks from further consideration ν X ν - look for signal in remaining tracks and clusters • Reconstruct 4 τ decay modes: - e νν , μ νν , π ν , ρ ν ⇒ about 72% of total BF • Key discriminating variable: Eextra : sum of neutral energy not associated with either reconstructed B meson • B + ➝ τ + ν will show an excess of events at Eextra ~ 0. Total efficiency, including τ BFs: 8 x 10 -4 Moriond EW, La Thuile, 2011 John Walsh, INFN Pisa 11

B + →τ + ν : results 383 M BB pairs • Likelihood fit to Eextra variable to extract B + →τ + ν signal • Signal template of Eextra from MC, corrected using + Data double-tagged data Total • Background Background templates: Preliminary - combinatoric: m ES sideband arXiv:1008.0104 - peaking: from B + B - MC • Fit yield: 69 ± 21 events • Preliminary branching fraction: Combined results (10 Combined results (10 -4 ) BaBar 1.76±0.49 New World Belle 1.54±0.48 Average HFAG 1.64±0.34 Moriond EW, La Thuile, 2011 12 John Walsh, INFN Pisa

B ➝ K νν 13

Search for B ➝ K νν • b → s FCNC, analogous to B ➝ Kl + l - , and therefore sensitive to NP: unparticles, MSSM, universal extra dimensions, etc. • Standard Model BF small ~ 3.8 x 10 -6 • NP models predict BF up to a factor of 10 larger Previous measurements Upper Limit Mode Experiment Dataset (fb -1 ) Reference (90% CL) B + ➝ K + νν 1.4 x 10 -5 Belle 492 PRL 99, 221802 (2007) B + ➝ K + νν 5.2 x 10 -5 BaBar 82 PRL 94, 101801 (2005) B 0 ➝ K 0 νν 1.6 x 10 -4 Belle 492 PRL 99, 221802 (2007) Moriond EW, La Thuile, 2011 14 John Walsh, INFN Pisa

B ➝ K νν : Experimental Techniques • As for B + ➝ τ + ν , few kinematical constraints on final state ⇒ reconstruct “other” B in event in semileptonic B decay • Look for signal decay among remaining particles of event Semileptonic tag Signal decay e or μ B B ν Tag efficiency ~ 1% D (*) ν ν Total efficiency ~ 0.1% Moriond EW, La Thuile, 2011 John Walsh, INFN Pisa 15

B ➝ K νν : Experimental Techniques II • Suppress high remaining background using multivariate classifier: Bagged Decision Trees (BDT) • Ensemble of BDTs trained on simulated signal and background events • Trees use 26 (K + ) or 38 (K 0 ) variables relating to i) missing energy , ii) event shape , iii) signal kinematics and iv) quality of reconstructed tag • Selection optimized for signal significance: s/(s+b) 1/2 Expected events Some NP models predict q 2 -dependence Moriond EW, La Thuile, 2011 John Walsh, INFN Pisa 16

B ➝ K νν : Results 459 M BB pairs Observed events N obs - N BG Upper limits Phys. Rev. D 82 , 112002 (2010) Best limits achieved to date Moriond EW, La Thuile, 2011 John Walsh, INFN Pisa 17

B + ➝ K + τ + τ - 18

Search for B + ➝ K + τ + τ - • Relative of B ➝ Kl + l - , where l=e or μ SM • Standard Model BF smaller due to τ mass ~ 4.3 x 10 -7 for • In some models (NMSSM), BF can depend on lepton mass: e NMSSM τ + τ - τ + τ - τ SM BF vs q 2 =m ττ 2 Moriond EW, La Thuile, 2011 19 John Walsh, INFN Pisa

B + ➝ K + τ + τ - : Experimental Techniques • Hadronic tag on “other B” • Signal side selections: - exactly 3 tracks Lepton (charge opposite to Btag) - particle ID momentum - q 2 > 14.23 GeV 2 - thrust angle of signal/tag side decays - missing energy Signal - extra neutral energy cut - hard track momentum - reduce B - ➝ l - X BG Extra neutral energy BG Signal MC events BG Total efficiency: 4.45 x 10 -4 cut Moriond EW, La Thuile, 2011 John Walsh, INFN Pisa 20

B + ➝ K + τ + τ - : Results 460 M BB pairs Expected events: 65±7 Observed events: 47 Preliminary Primary systematics: signal efficiency: 15% background estimation: 17% Upper limit (90% CL): first experimental limit Moriond EW, La Thuile, 2011 John Walsh, INFN Pisa 21

B 0 ➝ γγ and D 0 ➝ γγ 22

Search for B 0 ➝ γγ • Involves b → d γγ transition - small SM BF ~ JHEP 08 (2002) 054 • Can be enhanced by NP scenarios - extended Higgs sector: Aliev and Iltan, PRD 58, 095014 (1998) - SUSY with broken R-parity: Gemintern, Bar-Shalom and Eilam, PRD 70, 035008 (2004) Best limit to date: (90% CL) Belle, PRD 73, 051107 (2006) Moriond EW, La Thuile, 2011 23 John Walsh, INFN Pisa

B 0 ➝ γγ : Background suppression π 0 likelihood Background sources π 0 • Decays of π 0 and η decays Signal cut - reject with Likelihood ratio based on m( γγ ’) and E γ ’ • out-of-time Bhabha event overlap: - rejected with total energy and timing cuts Neural net response • Generic continuum events Continuum cut Signal - background Multivariate classifier (neural network) based on 19 input variables Moriond EW, La Thuile, 2011 24 John Walsh, INFN Pisa

B 0 ➝ γγ : Results + On-resonance data Total qq background 467 M BB pairs Signal BB Background • 2-D likelihood fit to m ES and Δ E variables • Signal yield: events Measured branching fraction: (1.9 σ significance) Likelihood function Upper limit: 90% CL improves best limit by factor ~ 2 N sig Moriond EW, La Thuile, 2011 25 John Walsh, INFN Pisa