EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH (CERN) CERN-PH-EP-2014-301 LHCb-PAPER-2014-066 3 March 2015 5 Angular analysis of the 1 0 2 0 ∗0 + − 2 B → K e e decay in the low-q r p A region 1 2 ] x e - p The LHCb collaboration† e h [ Abstract 2 v An angular analysis of the B0 K∗0e+e− decay is performed using a data sample, 8 corresponding to an integrate→d luminosity of 3.0 fb−1, collected by the LHCb 3 experiment in pp collisions at centre-of-mass energies of 7 and 8 TeV during 2011 0 3 and 2012. For the first time several observables are measured in the dielectron mass 0 squared (q2) interval between 0.002 and 1.120GeV2/c4. The angular observables F . L 1 andARe whicharerelatedtotheK∗0 polarisationandtotheleptonforward-backward 0 T asymmetry, are measured to be F = 0.16 0.06 0.03 and ARe = 0.10 0.18 0.05, 5 L ± ± T ± ± 1 where the first uncertainty is statistical and the second systematic. The angular v: observables A(2) and AIm which are sensitive to the photon polarisation in this q2 T T Xi range, are found to be A(2) = 0.23 0.23 0.05 and AIm = 0.14 0.22 0.05. The T − ± ± T ± ± r results are consistent with Standard Model predictions. a JHEP 04 (2015) 064. c CERN on behalf of the LHCb collaboration, license CC-BY-4.0. (cid:13) †Authors are listed at the end of this paper. ii 1 Introduction The B0 K∗0e+e− decay is a flavour changing neutral current process that is mediated → by electroweak box and loop diagrams in the Standard Model (SM). Charge conjugation is implied throughout this paper unless stated otherwise and the K∗0 represents the K∗0(892), reconstructed as K∗0 K+π−. The angular distribution of the K+π−e+e− system is → particularlysensitivetocontributionsfromnon-SMphysics(NP).TheleadingSMdiagrams are shown in Fig. 1; the relative contribution of each of the diagrams varies with the dilepton invariant mass. In the region where the dilepton invariant mass squared (q2) is less than 6GeV2/c4, some theoretical uncertainties from long distance contributions are greatly reduced, thereby allowing more control over the SM prediction and increasing sensitivity to any NP effect [1,2]. Furthermore, the contribution from a virtual photon coupling to the lepton pair dominates in the very low q2 region, allowing measurement of the helicity of the photon in b sγ transitions [3,4]. In the SM, this photon is predominantly → left-handed, with a small right-handed component arising from the mass of the s quark and long distance effects. In contrast, in many extension of the SM, NP may manifest as a large right handed current, see for example Refs. [5–8]. The q2 region below 1GeV2/c4 has previously been studied through the analysis of the B K∗(cid:96)+(cid:96)− ((cid:96) = e,µ) [9–11]. Experimentally, an analysis with muons rather than → d d B0 K 0 B0 K 0 ¯b u¯/c¯/t¯ s¯ ∗ ¯b W+ s¯ ∗ e+ u¯/c¯/t¯ e+ W+ γ,Z0 γ,Z0 e − e − d B0 ¯b u¯/c¯/t¯ s¯ K∗0 W+ νe W− e − e+ Figure 1: Dominant Standard Model Feynman graphs for the electroweak loop and box diagrams involved in the B0 K∗0e+e− decay. → 1 electrons in the final state produces a much higher yield at LHCb. This is primarily due to the distinctive signature that muons provide, which is efficiently exploited in the online selection, together with the better mass and energy resolutions and higher reconstruction efficiency of dimuon decays. However, as outlined in Ref. [12], dielectron decays at low q2 provide greater sensitivity to the photon polarisation and therefore to the and (cid:48) Wilson coefficients, which are associated with the left-handed and right-handed C7 C7 electromagnetic operators, respectively [3]. Due to the muon mass, the virtual photon contribution in dimuon decays is suppressed relative to dielectron decays. Additionally, the formalism for the B0 K∗0e+e− decay is greatly simplified as the electron mass can be → neglected. Indeed, the decay with electrons allows for an angular analysis down to a q2 of 0.0004GeV2/c4. However, above a q2 of 1GeV2/c4, the muon mass terms become negligible and the electron and muon modes have essentially the same functional dependence on the Wilson coefficients (within the lepton flavour universality assumption). This work is based on a previous analysis performed by the LHCb collaboration to measuretheB0 K∗0e+e− branchingfractionwithanintegratedluminosityof1.0fb−1 [13], → with the selection re-optimised for the angular analysis. The partial decay width of the B0 K∗0e+e− decay can be described in terms of q2 → and three angles, θ ,θ and φ. The angle θ is defined as the angle between the direction (cid:96) K (cid:96) of the e+ (e−) and the direction opposite to that of the B0 (B0) meson in the dielectron rest frame. The angle θ is defined as the angle between the direction of the kaon and the K direction opposite to that of the B0 (B0) meson in the K∗0 (K∗0) rest frame. The angle φ is the angle between the plane containing the e+ and e− and the plane containing the kaon and pion from the K∗0 (K∗0) in the B0 (B0) rest frame. The basis is designed such that the angular definition for the B0 decay is a CP transformation of that for the B0 decay. These definitions are identical to those used for the B0 K∗0µ+µ− analysis [11]. ˜ → As in Ref. [11], the angle φ is transformed such that φ = φ+π if φ < 0, to compensate for the limited signal yield. This transformation cancels out the terms that have a sinφ or cosφ dependence and simplifies the angular expression without any loss of sensitivity to the remaining observables. In the limit of massless leptons and neglecting the K+π− S-wave contribution, which is expected to be negligible1 at low q2 with the current sample 1Using Refs [1,14] it can be shown that the ratio of the S-wave fraction to the fraction of longitudinal polarisation of the K∗0 is constant as function of q2 in the 0-6 GeV2/c4 range. 2 size [14], the B0 K∗0e+e− angular distribution reads as → 1 d4(Γ+Γ¯) 9 (cid:20) 3 = (1 F )sin2θ +F cos2θ + d(Γ+Γ¯)/dq2dq2dcosθ dcosθ dφ˜ 16π 4 − L K L K (cid:96) K (cid:18) (cid:19) 1 (1 F )sin2θ F cos2θ cos2θ + L K L K (cid:96) 4 − − 1 (1 F )A(2)sin2θ sin2θ cos2φ˜ + 2 − L T K (cid:96) (1 F )AResin2θ cosθ + − L T K (cid:96) (cid:21) 1 (1 F )AImsin2θ sin2θ sin2φ˜ . 2 − L T K (cid:96) (1) The four angular observables F , A(2), ARe and AIm are related to the transversity L T T T amplitudes through [2] A 2 0 F = | | L A 2 + A 2 + A 2 0 || ⊥ | | | | | | A 2 A 2 A(2) = | ⊥| −| ||| T A 2 + A 2 ⊥ || | | | | (2) 2 e(A A∗ +A A∗ ) ARe = R ||L ⊥L ||R ⊥R T A 2 + A 2 || ⊥ | | | | 2 m(A A∗ +A A∗ ) AIm = I ||L ⊥L ||R ⊥R , T A 2 + A 2 || ⊥ | | | | where A 2 = A 2 + A 2, A 2 = A 2 + A 2 and A 2 = A 2 + A 2. The 0 0L 0R ⊥ ⊥L ⊥R || ||L ||R | | | | | | | | | | | | | | | | | | amplitudes A , A and A correspond to different polarisation states of the K∗0 in the 0 || ⊥ decay. The labels L and R refer to the left and right chirality of the dielectron system. Given the definition of φ˜, the observable A(2) is averaged between B0 and B0 decays, T while AIm corresponds to a CP asymmetry [15]. The observable F is the longitudinal T L polarisation of the K∗0 and is expected to be small at low q2, since the virtual photon is then quasi-real and therefore transversely polarised. The observable ARe is related to T the forward-backward asymmetry A by ARe = 4A /(1 F ) [2]. The observables FB T 3 FB − L A(2) and AIm, in the limit q2 0, can be expressed as simple functions of the and (cid:48) T T → C7 C7 coefficients [2] 2 e( (cid:48)∗) 2 m( (cid:48)∗) A(2)(q2 0) = R C7C7 and AIm(q2 0) = I C7C7 . (3) T → 2 + (cid:48) 2 T → 2 + (cid:48) 2 7 7 7 7 |C | |C | |C | |C | These measurements therefore provide information on photon polarisation amplitudes, similar to that obtained by the CP asymmetry measured through time-dependent analyses in B0 K∗0( K0π0)γ decays [16,17]. → → S This paper presents measurements of F , A(2), AIm and ARe of the B0 K∗0e+e− L T T T → decay in the bin corresponding to a reconstructed q2 from 0.0004 to 1GeV2/c4. 3 2 The LHCb detector and data set The study reported here is based on pp collision data, corresponding to an integrated luminosity of 3.0fb−1, collected at the Large Hadron Collider (LHC) with the LHCb detector [18,19] at centre-of-mass energies of 7 and 8 TeV during 2011 and 2012. The LHCb detector is a single-arm forward spectrometer covering the pseudorapidity range 2 < η < 5,designedforthestudyofparticlescontainingborcquarks. Thedetectorincludes a high-precision tracking system consisting of a silicon-strip vertex detector surrounding the pp interaction region [20], a large-area silicon-strip detector located upstream of a dipole magnet with a bending power of about 4Tm, and three stations of silicon-strip detectors and straw drift tubes [21] placed downstream of the magnet. The tracking system provides a measurement of momentum, p, with a relative uncertainty that varies from 0.5% at low momentum to 1.0% at 200GeV/c. The minimum distance of a track to a primary vertex, the impact parameter (IP), is measured with a resolution of (15+29/p )µm, where T p is the component of the momentum transverse to the beam, in GeV/c. Different types T of charged hadrons are distinguished using information from two ring-imaging Cherenkov detectors [22]. Photons, electrons and hadrons are identified by a calorimeter system consisting of scintillating-pad and preshower detectors, an electromagnetic calorimeter (ECAL) and a hadronic calorimeter. Muons are identified by a system composed of alternating layers of iron and multiwire proportional chambers [23]. The trigger [24] consists of a hardware stage, based on information from the calorimeter and muon systems, followed by a software stage, which applies a full event reconstruction. For signal candidates to be considered in this analysis, all tracks from the B0 K∗0e+e− → decay must have hits in the vertex detector and at least one of the tracks from the B0 K∗0e+e− decay must meet the requirements of the hardware electron or hadron → triggers, or the hardware trigger must be fulfilled independently of any of the decay products of the signal B0 candidate (usually triggering on the other b hadron in the event). The hardware electron trigger requires the presence of an ECAL cluster with a minimum transverse energy between 2.5GeV and 2.96GeV depending on the data taking period. The hardware hadron trigger requires the presence of a cluster in the hadron calorimeter with a transverse energy greater than 3.5GeV. The software trigger requires a two-, three- or four-track secondary vertex with a significant displacement from the primary pp interaction vertices (PVs). At least one charged particle must have a transverse momentum p > 1.7GeV/c and be inconsistent with originating from the PV. T A multivariate algorithm [25] is used for the identification of secondary vertices consistent with the decay of a b hadron. Samples of simulated B0 K∗0e+e− events are used to determine the efficiency to → trigger, reconstruct and select signal events. In addition, specific samples of simulated events are utilised to estimate the contribution from exclusive backgrounds and to model their mass and angular distributions. The pp collisions are generated using Pythia [26] with a specific LHCb configuration [27]. Decays of hadronic particles are described by EvtGen [28], in which final-state radiation is generated using Photos [29]. The interaction of the generated particles with the detector, and its response, are implemented 4 using the Geant4 toolkit [30] as described in Ref. [31]. The simulated samples are corrected for known differences between data and simulation in particle identification [22], detector occupancy and hardware trigger efficiency. 3 Selection of signal candidates Bremsstrahlung radiation, if not accounted for, would worsen the B0 mass resolution. If the radiation occurs downstream of the dipole magnet, the momentum of the electron is correctly measured and the photon energy is deposited in the same calorimeter cell as the electron. If photons are emitted upstream of the magnet, the electron momentum is evaluated after photon emission, and the measured B0 mass is shifted. In general, these bremsstrahlung photons deposit their energy in different calorimeter cells than those hit by the electron. In both cases, the ratio of the energy detected in the ECAL to the momentum measured by the tracking system, an important variable in identifying electrons, remains unbiased. To improve the momentum reconstruction, a dedicated bremsstrahlung recovery is used. Contributions from photon candidates, neutral clusters with transverse energy greater than 75MeV, found within a region of the ECAL defined by the extrapolation of the electron track upstream of the magnet, are added to the measured electron momentum. Oppositely charged electron pairs formed from tracks with p exceeding 350MeV/c and T with a good-quality vertex are used to form signal candidates. If the same bremsstrahlung photon is associated with both the e+ and the e−, its energy is added randomly to one of the tracks. The reconstructed e+e− invariant mass is required to be in the range 20–1000 MeV/c2 (0.0004 < q2 < 1GeV2/c4). Thechoiceofthelowerboundisacompromisebetween the gain in sensitivity to the photon polarisation from measuring as low as possible in q2 and a degradation of the resolution in φ˜ as q2 decreases, due to multiple scattering, as shown in Fig. 2. The lower bound requirement at 20MeV/c2 on the e+e− invariant mass also serves to reduce the background from B0 K∗0γ decays followed by a photon → conversion in the material, noted below as B0 K∗0γ . e+e− → Candidate K∗0 mesons are reconstructed in the K∗0 K+π− mode where the p of T → the K+ (π−) meson is required to be larger than 400 (300)MeV/c and charged pions and kaons are identified using information from the RICH detectors. Candidate K∗0 mesons and e+e− pairs are required to have a common good-quality vertex to form B0 candidates. When more than one PV is reconstructed, the one giving the smallest IP χ2 for the B0 candidate is chosen. The reconstructed decay vertex of the B0 candidate is required to be significantly separated from the PV and the candidate momentum direction to be consistent with its direction of flight from the PV. The B0 mass resolution, the angular acceptance and the rates of physics and combinatorial backgrounds depend on how the event was triggered. The data sample is therefore divided into three mutually exclusive categories: events for which one of the electrons from the B0 decay satisfies the hardware electron trigger, events for which one of the hadrons from the B0 decay satisfies the hardware hadron trigger and events triggered by activity in the event not due to any of the signal decay particles. 5 d] a LHCb r [ 2 simulation ~f s 1.5 1 0.5 0 0 50 100 m(e+e- ) [MeV/c2] Figure 2: Resolution on the φ˜ angle as a function of the e+e− invariant mass as obtained from LHCb simulated events. In order to maximise the signal efficiency while reducing the high level of combina- torial background, a multivariate classifier based on a boosted decision tree algorithm (BDT) [32,33] is used. The signal training sample is composed of simulated B0 K∗0e+e− → events and the background training sample is taken from the upper invariant mass sideband (m(K+π−e+e−) > 5600MeV/c2) of B0 K∗0e+e− decays reconstructed in half of the data → sample. Two separate BDTs are used, one each for half of the data sample. They are optimised separately and applied to the complementary half of the data in order to avoid any potential bias due to the use of the data upper sideband for the background sample. The BDT uses information about the event kinematic properties, vertex and track quality, IP and p of the tracks, flight distance from the PV as well as information about isola- T tion of the final state particles.2 The selection is optimised to maximise N /√N +N S S B separately for the three trigger categories and the two BDTs through a grid search of the set of criteria for the particle identification of the four final state particles and the BDT response. The background yield (N ) is extrapolated into the signal range using B the m(K+π−e+e−) distribution outside a 300MeV/c2 window around the known B0 ± mass. The expected signal yield (N ) is obtained using the B0 K∗0e+e− simulation and S → the known B0 K∗0e+e− branching fraction [13], and correcting for data-to-simulation → differences in the selection efficiency obtained using the well known B0 J/ψ(e+e−)K∗0 → decay. The efficiency of this requirement on the selected signal is 93% while the background is reduced by two orders of magnitude. The expected values for N /√N +N range S S B from 3.9 to 7.5 depending on the trigger category. 2The isolation is defined as the number of good two-track vertices that one of the candidate signal tracks can make with any other track in the event [34]. 6 4 Exclusive and partially reconstructed backgrounds Several sources of background are studied using samples of simulated events, corrected to reflect the difference in particle identification performances between data and simulation. A large non-peaking background comes from the B0 D−e+ν decay, with → D− e−νK∗0 which has a combined branching fraction about four orders of magni- → tude larger than that of the signal. In the rare case where both neutrinos have low energies, the signal selection is ineffective at rejecting this background which tends to peak towards cosθ 1. In order to avoid any potential bias in the measurement of the ARe parameter, (cid:96) ≈ T a symmetric requirement of cosθ < 0.8 is applied to suppress this background, resulting (cid:96) | | in a loss of signal of the order of 10%. To suppress background from B0 φe+e− decays, with φ K+K−, where one of s → → the kaons is misidentified as a pion, the two-hadron invariant mass computed under the K+K− hypothesis is required to be larger than 1040MeV/c2. Background from the decay Λ0 pK−e+e− is suppressed by rejecting events where b→ the pion is consistent with being a proton, according to the information from the RICH detectors. The probability for a decay B0 K∗0e+e− to be misidentified as B0 K∗0e+e− is → → estimated to be 1.1 % using simulated events and this background is therefore neglected. Another important source of background comes from the B0 K∗0γ decay, where the → photon converts into an e+e− pair. In LHCb, approximately 40% of the photons convert before reaching the calorimeter, and although only about 10% are reconstructed as an e+e− pair with hits in the vertex detector, the resulting mass of the B0 candidate peaks in the signal region. Two very effective criteria for suppressing this background are the minimum requirement on the e+e− invariant mass, m(e+e−) > 20MeV/c2, and a requirement that the uncertainty of the reconstructed z coordinate of the e+e− pair, σ (e+e−), is less than z 30mm. These requirements reject more than 99% of simulated B0 K∗0γ events. The → remaining contamination is estimated by normalising the simulated B0 K∗0γ to the e+e− → observed yield without the σ (e+e−) criterion and requiring the e+e− invariant mass to be z lower than 5 MeV/c2. The residual contamination from B0 K∗0γ decays is (3.8 1.9)% → ± of the signal yield. Part of this background comes from low-mass e+e− pairs that are reconstructed at larger masses due to multiple scattering. The remainder comes from direct Bethe-Heitler pair-production at masses larger than 20MeV/c2. To obtain an accurate estimate of this component, the Geant4 simulation is reweighted as a function of the true e+e− mass to match the distribution of Ref. [35] since Geant4 does not model correctly the high-mass e+e− pair production. Another possible source of contamination is the decay B0 K∗0V( e+e−) where → → V is a ρ, ω or φ meson. Expected rates for these backgrounds have been evaluated in Refs. [4,36]. The effects of direct decays or interference with the signal decay are found to be negligible after integrating over the q2 range. Partially reconstructed (PR) backgrounds arising from B0 K∗0e+e−X decays, where → one or more of the decay products (X) from the B0 decay is not reconstructed, are also taken into account. These incomplete events are mostly due to decays involving higher 7 K∗ resonances, hereafter referred to as K∗∗. The decays B0 K∗0η and B0 K∗0π0 → → are also studied and several cases are considered: the case when the e+e− pair comes from a converted photon in the material, the case when the e+ and e− originate from the conversions of the two photons and finally the case of the Dalitz decay of the η or the π0. They contribute about 25% of the PR background in the angular fitting domain. 5 Fit to the K+π e+e invariant mass distribution − − In a first step, a mass fit over a wide mass range, from 4300 to 6300MeV/c2, is performed to estimate the size of the B0 K∗0e+e− signal, the combinatorial background and the PR → background. The fractions of each component are determined from unbinned maximum likelihood fits to the mass distributions separately for each trigger category. The mass distribution of each category is fitted to a sum of probability density functions (PDFs), modelling the different components. Following the strategy of Ref. [37], the signal PDF depends on the number of neutral clusters that are added to the dielectron candidate to correct for the effects of bremsstrahlung. The signal is described by the sum of a Crystal Ball function [38] (CB) and a wide Gaussian function accounting for the cases where background photons have been associated; the CB function accounts for over 90% of the total signal PDF. The shape of the combinatorial background is parameterised by an exponential function. Finally, the shape of the PR background is described by non-parametric PDFs [39] determined from fully simulated events passing the selection. The signal shape parameters are fixed to the values obtained from fits to simulation but the widths and mean values are corrected for data simulation differences using B0 J/ψ(e+e−)K∗0 as a control channel. Since the photon pole contribution dominates → in the low-q2 region, the PR background is expected to be similar for B0 K∗0e+e− and → B0 K∗0γ. The large branching fraction of the decay B0 K∗0γ allows the fractions → → of PR background relative to the signal yield to be determined from the data. These fractions are extracted from a fit to a larger sample of events obtained by removing the requirements on the lower bound of the e+e− invariant mass and on σ (e+e−) and therefore z dominated by B0 K∗0γ events. The invariant mass distribution, together with the e+e− → PDFs resulting from this fit, is shown in Fig. 3(a) for the three trigger categories grouped together. The corresponding distribution for the B0 K∗0e+e− fit is shown in Fig. 3(b). → There are 150 17 B0 K∗0e+e− signal events, 106 16 PR background events and ± → ± 681 32 combinatorial background events in the 4300 6300MeV/c2 window. ± − In this wide mass window, the sample is dominated by combinatorial background, whose angular shape is difficult to model. Furthermore the angular distributions depend on the kinematic properties of the background and may thus vary as functions of mass. Hence, the angular fit is performed in a narrower mass window from 4800MeV/c2 to 5400MeV/c2. In this restricted window there are 124 B0 K∗0e+e− signal events, 38 PR → and 83 combinatorial background events, corresponding to a signal-to-background ratio of the order of one. About half of these events belong to the electron hardware trigger category and the rest are equally distributed between the other two categories. 8