Searches for lepton-number-violating B decays at CLEO, BABAR, and Belle Fergus Wilson1 Rutherford Appleton Laboratory, Chilton, Didcot, 3 1 Oxon, OX11 0QX, UNITED KINGDOM 0 2 n Proceedings of CKM 2012, the 7th International Workshop on the CKM Unitarity a J Triangle, University of Cincinnati, USA, 28 September - 2 October 2012 0 1 1 Introduction ] x e CLEO, BABAR, and Belle have searched for lepton-number-violation (LNV) in B - p meson decays at e+e− colliders. In the Standard Model (SM), lepton number L is e h conserved in low-energy collisions and decays, and the lepton flavor numbers for the [ threeleptonfamiliesareconserved ifneutrinosaremassless. Anumber ofmechanisms 1 forLNVhave been proposedincluding multi-Higgs-bosonextensions, leptoquarks and v Majorana neutrinos. 5 7 Acommonthemeamongtheexperiments istheuseoftwo discriminating variables 0 ∗ calculated in the center-of-mass frame (CM) (∆E = E √s/2 and m = m = 2 B ES cand − . M = s/4 p∗2, where E∗ and p∗ are the CM energy and momentum of the 1 bc q B B B − 0 reconstructed B meson candidate), and event-shape discriminants. The signal events 3 peak at zero for ∆E, at approximately the B meson mass for m , and the event- 1 ES : shape is spherical; for the backgrounds, ∆E and m are smoothly varying, while v ES i the event-shape is more jet-like. The major backgrounds are continuum production X of quark pairs e+e− qq (q = u,d,s and c), semileptonic B decays which produce a r → a lepton, and other BB decays where a hadron is mis-identified as a lepton. 2 Recent results from CLEO, BABAR, and Belle CLEO reused the techniques for their search for B Xsℓ+ℓ− to look for B+ h−e+e+, h−e+µ+ and h−µ+µ+ where h− = K−,K∗−→,π− and ρ− [1], based on 9→.6 million BB events. There are three main sources of background: B decays of the type B XJ/ψ and B Xψ(2S); other B decays, with two apparent leptons → → (either real leptons or hadrons misidentified as leptons); and continuum processes 1On behalf of the Belle and BABAR collaborations 1 with two apparent leptons. The backgrounds from J/ψ and ψ(2S) were severe in the searches for B Xℓ+ℓ− as the decays provided two opposite sign leptons but here → they contribute less than 0.1 event per decay mode after careful lepton identification. An unbinned maximum likelihood (ML) method is used to discriminate between signal events and the remaining two background sources. The variables are: the missing event energy, E 2E E , where E denotes the sum of miss beam det det ≡ − P P energiesofallthedetectedparticlesintheevent; aFisher discriminant basedonthe F event-shape; M ; and∆E. Loosecutsof E < 2GeV, 5.2 < M < 5.3GeV/c2 cand miss cand | | and ∆E < 0.25GeVareapplied. Thebranching fractionforthesignal andtheyields | | for the two backgrounds are free parameters in the ML. No evidence for the decays is foundandall modeshaveastatistical significance ofless than1.2standarddeviations. 90% confidence level (CL) upper limits (UL) are placed on the branching fractions as shown in Table 1. The upper limits range from 1.0 to 8.3 10−6. × Mode S UL Mode S UL Mode S UL K− e+ e+ 0.6σ 1.0 K− e+ µ+ 0.0σ 2.0 K− µ+ µ+ 0.0σ 1.8 K∗− e+ e+ 0.0σ 2.8 K∗− e+ µ+ 0.0σ 4.4 K∗− µ+ µ+ 0.5σ 8.3 π− e+ e+ 0.0σ 1.6 π− e+ µ+ 0.0σ 1.3 π− µ+ µ+ 0.0σ 1.4 ρ− e+ e+ 1.1σ 2.6 ρ− e+ µ+ 0.3σ 2.2 ρ− µ+ µ+ 1.0σ 5.0 Table 1: CLEO results with the statistical significance S and 90% confidence level upper limits UL ( 10−6) on the branching fraction (including systematic error) [1]. × BABAR has repeated the CLEO search for the four modes B+ h−µ+µ+ and B+ h−e+e+, whereh− = K− orπ− [2]andtheyalsoreusethetechn→iques developed for t→heir B K(∗)ℓ+ℓ− analyses, using a data sample of 471 3 million BB pairs. → ± The key difference with CLEO is the ML function used. The two main backgrounds from continuum events and BB decays are suppressed through the use of a ML ratio constructed from boosted decision tree discriminants (BDTs). Before the ML fit, R a selection on retains 85% of the simulated signal events while rejecting more than R 95% of the background. The selection efficiency for simulated signal is 13%-48%. The signal branching fraction and background yields are extracted from the data with an unbinned ML using m and . The signal m distributions are taken from ES ES R data using a Gaussian shape unique to each final state, with the mean and width determined from fits to the analogous final states in the B+ J/ψ( ℓ+ℓ−)h+ → → events from the data. No significant yields are observed and the results of the ML fits to the data are summarized in Table 2. Belle have performed the first searches for the decays B+ D−e+e+, D−e+µ+ and D−µ+µ+ using a data sample containing 772 million BB p→airs [3]. As the CKM matrixelement Vcb > Vub , thesedecays couldbeenhanced byanorderofmagnitude compared to B|+ | h|−ℓ+|ℓ+. Belle look for an energetic same-sign dilepton and → 2 Mode Events Yield (σ) ( 10−8) UL ( 10−8) B+ π−e+e+ 123 0.6+2.5 S 0.4 0.2B7+1×.1 0.1 B × 2.3 B+ → K−e+e+ 42 0.7−+21..78 0.5 0.49−+11..23 ±0.1 3.0 B+ → π−µ+µ+ 228 0.0−+13..22 0.0 0.03−+05..81 ±0.6 10.7 B+ → K−µ+µ+ 209 0.5−+23..05 0.2 0.45−+33..22 ±0.4 6.7 → −2.5 −2.7 ± Table 2: BABAR results, showing the total events in the sample, signal yield and its statistical uncertainty, significance , branching fraction , and 90% CL branching S B fraction upper limit UL [2]. B combine itwith aD candidate requiring aproper chargecombination forthedilepton. The same-sign lepton pair must have a total energy in the Υ(4S)center-of-mass (CM) frame greater than 1.3GeV. More than 95% of events have only one same-sign lepton pair. Candidate D− mesons are reconstructed in the D− K+π−π− decay. The three − → tracks from the D candidate are fit to a common vertex and are required to have a K+π−π− invariant mass (MKππ)within approximately 10 MeV/c2 fromthenominal − ± D mass. The MKππ distribution is fit to two Gaussian functions with a common mean in a mass window 3 times the width of the narrower Gaussian component. ± − The average multiplicity of D candidates is 1.3 per event. The continuum background is rejected using a Fisher discriminant based on F event-shape variables, and cosθB, the cosine of the polar angle of the B candidate flight direction in the CM frame. Semileptonic decays such as B D−ℓ+νℓX have → missing energy due to the undetected neutrino and the two reconstructed leptons do notcome fromthesame vertex. These events canberejected using the missing energy E and δz, the separation between the impact parameter of the two leptons in the miss beam direction. The four variables, , cosθB, Emiss and δz, are combined together F into a single likelihood ratio, R . The optimal requirement on R is determined by s s maximizing the figure of merit, ǫ /(a/2+√N ), where ǫ is the MC signal efficiency, s b s N is the number of expected background events in the signal region, and a is set to b zero. After applying the R requirements, 5, 23 and 40 events remain in the background s region for the e+e+, e+µ+ and µ+µ+ modes, respectively. The signal efficiencies are evaluated to be 1.2% - 1.9%. The expected numbers of background events in the signal region are 0.18, 0.83 and 1.44 events for the e+e+, e+µ+ and µ+µ+ modes, respectively. Figure 1 shows the M -∆E distributions of the selected events. No bc eventsareobservedinthesignalregion. Table3summarisestheupperlimitsachieved. BABAR has performed the first measurement of the branching fraction upper lim- its for the baryon- and lepton-number violating decays B0 Λ+ℓ−, B− Λℓ− and c B− Λℓ− [4] with 471 3 million BB pairs. The Λ+ and→Λ candidates→are recon- c struc→ted through the dec±ay modes Λ+ pK−π+ and Λ pπ−, respectively. The c → → 3 0.3 0.3 0.3 0.2 0.2 0.2 V) 0.1 V) 0.1 V) 0.1 e e e G 0 G 0 G 0 E ( E ( E ( ∆ -0.1 ∆ -0.1 ∆ -0.1 -0.2 -0.2 -0.2 -0.3 -0.3 -0.3 5.2 5.22 5.24 5.26 5.28 5.2 5.22 5.24 5.26 5.28 5.2 5.22 5.24 5.26 5.28 M (GeV/c2) M (GeV/c2) M (GeV/c2) bc bc bc Figure 1: The M -∆E distributions of D−e+e+ (left), D−e+µ+ (middle) and bc D−µ+µ+ (right) final states in data. The (red) boxes indicate the signal regions. Mode ǫ [%] N Nbkg UL [10−6] obs exp B+ D−e+e+ 1.2 0 0.18 0.13 < 2.6 B+ →D−e+µ+ 1.3 0 0.83±0.29 < 1.8 B+ →D−µ+µ+ 1.9 0 1.10±0.33 < 1.1 → ± Table 3: Belle results for the B+ D−ℓ+ℓ+ search; ǫ is the signal reconstruction effi- → ciency; N is the number of events in the signal region; Nbkg is the expected number obs exp of background events, and UL is the branching fraction 90% CL upper limit [3]. final state tracks for both the Λ+ and Λ decays are constrained to a common spatial c vertex and their invariant mass is constrained to the Λ+ or Λ mass. The candidates c are required to be within 15 MeV/c2 of the nominal Λ+ mass and 4 MeV/c2 of c ± ± the nominal Λ mass. B meson candidates are formed by combining baryon candidate − − with a µ or e and constraining them to a common point. Bremsstrahlung energy recovery is performed for electrons. Background from e+e− e+e−γ events in the → Λℓ channel are eleminated by requiring more than four charged tracks in the event. Candidate selection is optimized using a figure of merit (see above) with a = 5. A neural net (NN) is used to provide further discrimination between signal and back- groundsuchthatabout90%ofthesignalisretainedandabout50%ofthebackground is rejected. The remaining background after selection for the Λ+ℓ− modes is com- c posed of roughly equal amounts of BB and continuum events, while the background for the Λℓ modes is almost entirely continuum. The variables ∆E and m are used in the ML fit to the two Λℓ modes; the Λ+ℓ− ES c decay has more background and the NN is used as a third discriminating variable. No significant signal is observed and an upper limit is calculated for the branching fraction for each decay mode, as shown in Table 4. 4 Mode N ( 10−8) ǫ (%) ( 10−8) cand 90% B0 Λ+µ− 814 B ×4+71 26.3 0.9 B 18×0 B0 → Λ+c e− 651 1−90+−15360 25.7±0.7 520 B−→ Λcµ− 320 2.3−+930.5 28.7±0.9 6.2 B− → Λe− 194 −1.2+−3.27.5 27.2±0.6 8.1 B− → Λµ− 192 1.5−+22..66 31.3±1.0 6.1 B− → Λe− 74 0.9−+1.07.7 30.0±0.6 3.2 → − −0.0 ± Table 4: Summary of number of candidates (N ), branching fraction central value cand ( ), signal efficiency (ǫ), and branching fraction 90% CL UL ( ) [4]. 90% B B b) C H L r 10-6 o % f 5 9 s (10-7 mit Li r 10-8 e p p U L. 10-9 CLEO C. BaBar % Belle 0 LHCb 910-10 +µΛ-c+Λ-ec-µΛ -Λ eµΛ- Λ- eπ++-eeµµπ-++ µπ+-+e-++eeK-µµ++K-µ++eKρ-++eeµµρ-++ µρ-++e*-++eeK*-µµ++K*-µ++eK-++eeD-µµ++D-µ++eD*+µµ--D+µµ--Ds0µµπ--+D Figure 2: Branching fraction upper limits for B meson LNV decays from CLEO [1], BABAR [2, 4], Belle [3], and LHCb [5]. References [1] K. W. Edwards et al. (CLEO Collaboration), Phys. Rev. D 65, 111102 (2002). [2] J.P. Lees et al. (BABAR Collaboration), Phys. Rev. D 85, 071103 (2012). [3] O. Seon et al. (Belle Collaboration), Phys. Rev. D 84, 071106 (2011). [4] P. del Amo Sanchez et al. (BABAR Collaboration), Phys. Rev. D 83, 091101 (2011). [5] R.Aaijetal.(LHCbCollaboration),Phys.Rev.Lett.108,101601(2012);R.Aaij et al. (LHCb Collaboration), Phys. Rev. D 85, 112004 (2012). 5