Study of B -->D^(*)D_s(J)^(*) Decays and Measurement of D_s^- and D_sJ(2 460)^- Branching Fractions

We present branching fraction measurements of twelve B meson decays of the form B -->D^(*)D_s(J)^(*). The results are based on Y(4S) decays in BBbar pairs. One of the B mesons is fully reconstructed and the other decays to two charm mesons, of which one is reconstructed, and the mass and momentum of the other is inferred by kinematics. Combining these results with previous exclusive branching fraction measurements, we determine BR(D_s^- -->phi pi^-) = (4.62 +/- 0.36_stat. +/- 0.51_syst.)%, BR(D_sJ(2460)^- -->D_s^*- pi^0) = (56 +/- 13_stat. +/- 9_syst.)% and BR(D-sJ(2460)^- -->D_s^- gamma) = (16 +/- 4_stat. +/- 3_syst.)%.

PACS numbers: 13.25.Hw, 13.25.Ft In this paper we present the study of charged and neutral B mesons decaying to two charm mesons, i.e. B → D meas D X [1]. D meas represents a fully reconstructed D ( * )+,0 or D ( * )− s meson, and the mass and momentum of the D X are inferred from the kinematics of the twobody B decay. This study allows measurements of B branching fractions without any assumption on the decays of the D X . Measurements of these two-body branching fractions can provide tests of the factorization of the decay amplitudes [2] in the high momentum transfer regime [3]. From two separate classes of events with D meas = D  [4], thus extracting for the first time the absolute branching fractions of this recently observed state [5].
This analysis uses Υ (4S) → BB events in which either a B + or a B 0 meson decays into a fully reconstructed hadronic final state (B reco ). The measurements are based on an integrated luminosity of 210.5 fb −1 recorded at the Υ (4S) resonance with the BABAR detector at the PEP-II asymmetric-energy e + e − collider operating near the Υ (4S) resonance. An additional 21.7 fb −1 recorded 40 MeV below the resonance (off-resonance) are used to evaluate backgrounds. The BABAR detector is described in detail elsewhere [6]. Charged-particle trajectories are measured by a vertex tracker with 5 double-sided layers and a 40-layer drift chamber, both operating in a 1.5-T magnetic field of a superconducting solenoid. Chargedparticle identification is provided by the specific energy loss (dE/dx) in the tracking devices and by an internally reflecting ring-imaging Cherenkov detector. Photons are detected by a CsI(Tl) electromagnetic calorimeter. We use Monte Carlo simulations (MC) of the BABAR detector based on GEANT4 [7] to optimize selection criteria and determine selection efficiencies.
To reconstruct a large sample of B mesons, the hadronic decays B reco → DY + , D * Y + are selected. Here, the system Y + consists of hadrons with a total charge of +1, composed of n 1 π ± n 2 K ± n 3 K 0 S n 4 π 0 , where n 1 + n 2 ≤ 5, n 3 ≤ 2, and n 4 ≤ 2. We re- The kinematic consistency of B reco candidates is checked with two variables, the beam energysubstituted mass m ES = s/4 − p 2 B and the energy dif- Here √ s is the total energy in the Υ (4S) center-of-mass (CM) frame, and p B and E B denote the momentum and energy of the B reco candidate in the same frame. The resolution on ∆E is measured to be σ ∆E = 10 − 35 MeV, depending on the decay mode, and we require |∆E| < 3σ ∆E .
For each reconstructed B decay mode, the purity P is estimated as the ratio of the number of signal events with m ES > 5.27 GeV/c to the total number of events in the same range, and is evaluated on data. We only use modes for which P exceeds a decay-mode dependent threshold in the range of 9% to 24%. In events with more than one B reco we select the decay mode with the highest purity. On average, we reconstruct one signal B reco candidate in 0.3% (0.5%) of the B 0 B 0 (B + B − ) events.
The selected sample of B reco is used as normalization for the determination of the branching fractions. It is contaminated by e + e − → qq (q = u, d, s, c) events and by other Υ (4S) → B 0 B 0 or B + B − decays, in which the B reco is mistakenly reconstructed from particles coming from both B mesons in the event. To significantly reduce the e + e − → qq background we require the angle θ * T B , defined in the CM frame, between the thrust axis [8] of the B reco and the thrust axis of all charged and neutral particles in the event excluding the ones that form the B reco , to satisfy the requirement | cos θ * T B | < 0.7. On this signal-enriched sample ( Fig. 1), the contributions from the background are estimated as the sum of three components: the e + e − → qq, the B 0 B 0 , and the B + B − events. The shapes of these background distributions are taken from MC simulation. The normalization is extrapolated and subtracted from the data to estimate the signal yield. After correcting for the | cos θ * T B | cut efficiency estimated in the MC, the size of the total sample of fully reconstructed B decays is N B 0 reco = (2.90±0.01 stat. )×10 5 and N B + reco = (4.63±0.01 stat. )×10 5 . From the charged tracks and the neutral clusters that do not belong to the B reco we reconstruct the charmed mesons (D meas ) in the modes D 0 → K − π + , K − π + π 0 , . We select φ and K * 0 candidates with a reconstructed mass within 15 MeV/c 2 and 70 MeV/c 2 from their nominal values [9], respectively. The D * candidates are reconstructed in the decay modes D * + → D 0 π + , D + π 0 , D * 0 → D 0 π 0 , D 0 γ, and D * − s → D − s γ. We require the reconstructed masses of the D 0 , D + , and D − s candidates and the differences ∆m between the masses of the D * and D candidates to be within 1.5 − 3 times its measured resolution from their nominal values [9], depending on the background level.
We apply further selection criteria to enhance the signal contributions in the sample. For D 0 X we require positive charged B reco candidates. We suppress background from B → D ( * ) lν, while keeping events with a semileptonic D X decay, by rejecting any event with a remaining identified lepton with the appropriate charge and a momentum in the B rest frame (p * ) greater than 1 GeV/c. In order to minimize the contamination of the modes with a D * to the modes with a D meson, we assign the events consistent with both the hypotheses (B → DD X and B → D * D X ) to the D * sample.
The invariant mass of D X (m X ) is derived from the missing four-momentum p X = p Υ (4S) − p Breco − p Dmeas , where all momenta are measured in the laboratory frame. The m X resolution is improved by a global Υ (4S) kinematic fit [10] that includes beam position and energy information and constrains the masses and decay vertices of the D meas . The χ 2 of this fit is used to reduce the combinatorial background. We remove reconstructed D mesons with χ 2 probability smaller than 0.1%.
Of the selected events, 3 − 6% (9 − 30%) contain multiple D (s) (D * (s) ) candidates. We retain those in the D meas decay mode with the lowest combinatorial background. If there are multiple candidates with the same decay mode, we select the one with the lowest value of |m D − m P DG | and (m Dmeas −m P DG ) 2 /σ 2 mD meas +(∆m−∆m P DG ) 2 /σ 2 ∆m for D (s) and D * (s) respectively, where m is the reconstructed mass of the D meas candidate and the subscript P DG indicates nominal values [9].
Finally, we consider only candidates in the range 1.65 GeV/c 2 < m X < 2.71 GeV/c 2 for the D ( * )+/0 D X modes and 1.68 GeV/c 2 < m X < 2.31 GeV/c 2 for D ( * )− s D X . These ranges were chosen to minimize the total uncertainty introduced by the background shape and normalization.
The yield of each decay mode is extracted from the m X distribution by a binned χ 2 fit of a sum of n sig signal contributions (N sig ) and the total background contribution (N bkg ), which is a sum of the combinatorial background, other B → D where N meas i is the number of observed events in bin i, µ i corresponds to µ i = j=1,nsig C j N sig ij + C bkg N bkg i , the index j denotes the signal component, and δN meas i and δN MC i are the statistical uncertainties for data and MC samples, respectively. The relative normalizations of each component (C j and C bkg ) are allowed to vary in the fit. The measured m X distributions and the results of the fits are shown in Fig. 2.
The branching fractions are extracted as B(f ) = N fit /(εN Breco ), where N fit is the number of signal events obtained from the fit to the m X distribution for a given mode and ε, which includes the intermediate branching fractions of D meas and its decay products, is the selection efficiency estimated using MC simulation.
The dominant systematic uncertainties originate from the lack of knowledge of the correct shapes used in the m X fit, and from the determination of efficiencies (because of the limited MC statistics). These uncertainties range from 5.6% to 25%, depending on the mode. The systematic uncertainties due to the determination of N Breco and to the differences between data and MC in the composition of the reconstructed B reco modes range between 3.7% and 6.7% for B 0 , and between 3.5% and 9.0% for B + depending on the mode under study. Other uncertainties come from track reconstruction efficiency (1.4% per track and 2.2% per soft pion), γ and π 0 efficiencies (3.0% per π 0 and 1.8% per γ), and kaon identification (2% per kaon). The uncertainties due to branching fraction measurements for exclusive D ( * ) (s) decays [9] contribute between 3.0% and 7.4%, depending on the mode. We check the uncertainties introduced by the χ 2 cut of the kinematic fit by comparing data and MC control samples for B → D ( * ) lν obtained with all previously mentioned cuts except for the p * > 1 GeV/c criterion applied. The statistical uncertainty of this comparison is used as the systematic uncertainty (between 0.5% and 2.3%).
We combine the sixteen measurements of B → D ( * ) D  Table I.
We further combine the results of this analysis with B → D ( * )+/0 D ( * )− s exclusive branching fractions from [11,12,13,14] and the BABAR results for B(B → D sJ (2460) − D ( * ) ) [4], obtaining the following branching fractions: In conclusion, we have measured the branching fractions for the decays B → D ( * )+,0 D ( * )− s . These are consistent with the existing measurements [9] and, in several cases, have a significantly smaller uncertainty. The combination of these results with the existing measurements provide the branching fraction for D − s → φπ − , which is also consistent with the most recent measurement  We are grateful for the excellent luminosity and machine conditions provided by our PEP-II colleagues, and for the substantial dedicated effort from the computing organizations that support BABAR. The collaborating institutions wish to thank SLAC for its support and kind hospitality. This work is supported by DOE