Search for anomalous Wtb couplings in single top quark production in ppbar collisions at sqrt(s) = 1.96 TeV

We present new direct constraints on a general $Wtb$ interaction using data corresponding to an integrated luminosity of 5.4 fb$^{-1}$ collected by the D0 detector at the Tevatron $p\bar{p}$ collider. The standard model provides a purely left-handed vector coupling at the $Wtb$ vertex, while the most general, lowest dimension Lagrangian allows right-handed vector and left- or right-handed tensor couplings as well. We obtain precise limits on these anomalous couplings by comparing the data to the expectations from different assumptions on the $Wtb$ coupling.

S. Söldner-Rembold, 43 L. Sonnenschein, 20 K. Soustruznik, 8  We present new direct constraints on a general W tb interaction using data corresponding to an integrated luminosity of 5.4 fb −1 collected by the D0 detector at the Tevatron pp collider. The standard model provides a purely left-handed vector coupling at the W tb vertex, while the most general, lowest dimension Lagrangian allows right-handed vector and left-or right-handed tensor couplings as well. We obtain precise limits on these anomalous couplings by comparing the data to the expectations from different assumptions on the W tb coupling. The top quark was discovered in 1995 at the Tevatron [1,2] via the pair production mode involving strong interactions. In 2009, the electroweak production of the top quark was observed by the D0 and CDF collaborations [3,4]. At the Tevatron, the dominant production modes for single top quark are the s-channel ("tb") [5] and t-channel ("tqb") [6,7]   tron compared to the tb and tqb processes [8]. Recently, we presented improved measurements of the single top quark production cross sections [9] and the observation of t-channel single top quark production [10].
The large mass of the top quark implies that it has large couplings to the electroweak symmetry breaking sector of the standard model (SM) and may have non-standard interactions with the weak gauge bosons. Single top quark production provides a unique probe to study the interactions of the top quark with the W boson.
The most general, lowest dimension, CP -conserving Wtb vertex is given by [11]: where M W is the mass of the W boson, q ν is the W boson four-momentum, P L = (1 − γ 5 )/2 is the left-handed projection operator, P R = (1+γ 5 )/2 is the right-handed projection operator, L V,T = V tb ·f LV,T and R V,T = V tb ·f RV,T .
The form factor f LV (f LT ) represents the left-handed vector (tensor) coupling, f RV (f RT ) represents the righthanded vector (tensor) coupling, and V tb is the Cabibbo-Kobayashi-Maskawa matrix element. In the SM, the W tb coupling is left-handed with L V ≡ |V tb | ≃ 1 and R V = L T = R T = 0. The magnitudes of the righthanded vector coupling and the tensor couplings can be indirectly constrained by the measured branching ratio of the b → sγ process [12]. Measurements of top quark decays in tt production, e.g. the W boson helicity [13,14], can directly constrain the Lorentz structure of the Wtb vertex [15]. Assuming single top quarks are produced only via W boson exchange, the single top quark cross section is directly proportional to the square of the effective Wtb coupling. Moreover, the event kinematics and angular distributions are also sensitive to the existence of anomalous top quark couplings [16,17]. Therefore, direct constraints on anomalous couplings can be obtained by measuring single top quark production [18]. This analysis uses the same data, event selection, and background modeling as the recent single top quark cross section measurements [9,10]. We perform a study of anomalous W tb couplings and obtain substantial improvements on the limits of these couplings following the general framework given in Ref. [18]. Out of the four couplings (L V , L T , R V , R T ), we consider three cases pairing the left-handed vector coupling with each of the other three couplings: (L V , R V ), (L V , L T ) and (L V , R T ), and for each case we assume the other two non-SM couplings are negligible. This pairing allows us to limit the complexity of the analysis, while increasing the statistical power and sensitivity for the anomalous coupling under study. We assume that single top quarks are produced exclusively through W boson exchange. Therefore other single top quark production mechanisms, such as flavorchanging neutral current interactions [19,20], the decay of new scalar boson [21], or the exchange of new vector boson [22,23] are not considered here. We also assume that the W tb vertex dominates top quark production and decay, i.e., |V td | 2 + |V ts | 2 ≪ |V tb | 2 . The results presented here supersede those contained in Ref. [18]. In addition to the analysis of additional data and to the improvements in the event selection and background modeling used for the results presented in Ref. [9,10], we use an updated calculation of the cross section as a function of the anomalous couplings which addresses a mistake present in the original analysis [18] (in that analysis the R T and L T couplings were accidentally swapped).
We select single top quark events which are expected to contain exactly one isolated large transverse momentum (p T ) electron or muon and large missing transverse energy (/ E T ). Events with 2, 3 or 4 jets are selected, and one or two of the jets are required to originate from the hadronization of long-lived b hadrons (b-jets) as determined by a multivariate b-tagging algorithm [24]. To increase the search sensitivity, we divide our data into six independent analysis channels, each with a different background composition and signal-to-background ratio. The channels are based on the number of identified b jets (1 or 2) and jet multiplicity (2, 3 or 4 jets). The signal selection efficiencies with different W tb couplings, including branching fraction, trigger efficiencies and the b-tagging requirements, vary between 2.7% and 3.0% for tb and 1.9% and 2.2% for tqb production, estimated using Monte Carlo (MC) simulations. The "associated production" process gives a negligible contribution to this analysis.
Single top quark signal events with the SM and anomalous W tb couplings are modeled using the comphepbased MC event generator singletop [25] for a top quark mass m t = 172.5 GeV using the CTEQ6M [26] parton distribution functions. The anomalous W tb couplings are taken into account in both production and decay in the generated samples. The event kinematics for both schannel and t-channel processes reproduce distributions from next-to-leading-order calculations [27,28]. The decay of the top quark and the resulting W boson are carried out in the singletop [25] generator in order to preserve information about the spin of the particles. The theoretical cross sections for anomalous single top quark production (s+t-channel) with |V tb | ≃ 1 are 3. [16]. All other couplings are set to zero when calculating these cross sections. The SM single top quark production cross section is 3.3 ± 0.1 pb [8].
The main background contributions are those from W bosons produced in association with jets (W +jets), tt production, and multijet production in which a jet with high electromagnetic content mimics an electron, or a muon contained within a jet originating from the decay of a heavy-flavor quark (b or c quark) appears to be isolated. Diboson (W W , W Z, ZZ) and Z+jets processes add small additional contributions to the background. The tt, W +jets, and Z+jets events are simulated with the alpgen leading-log MC generator [29]. The effect of anomalous W tb couplings on kinematic distributions of the tt background has been found to be negligible, thus only SM tt samples are considered in the analysis. Diboson processes are modeled using pythia [30]. For all of the signal and background MC samples, pythia is used to simulate parton showers and to model hadronization of all generated partons. The presence of additional pp interactions is modeled by events selected from random beam crossings matching the instantaneous luminosity profile in the data. All MC events are processed through a geant-based simulation [31] of the D0 detector and reconstructed using the same algorithm as data. Differences between simulation and data in lepton and jet reconstruction efficiencies and resolutions, jet energy scale, and b-tagging efficiencies are corrected in the simulation by applying correction functions measured from separate data samples. The tt, Z+jets and diboson MC samples are scaled to their theoretical cross sections [32,33]. We use data containing non-isolated leptons to model the multijet background. W +jets and multijet backgrounds are normalized by comparing the prediction for background to data before b-tagging. Details of the selection criteria and background modeling are given in Ref. [9].
The main contributions to the systematic uncertainty on the predicted number of events arise from the signal modeling, the jet energy scale (JES), jet energy resolution (JER), corrections to b-tagging efficiency and the correction for jet-flavor composition in W +jets events. These uncertainties affect the normalization of the distributions, and in some cases (JES, JER, and b-tagging) also change the differential distributions. There are smaller contributions due to limited statistics of the MC samples, uncertainties on the measured luminosity, and the trigger modeling. In addition, we also consider a signal cross section uncertainty (3.8% for tb and 5.3% for tqb) given by the NLO calculation. Details of systematic uncertainties are given in Ref. [9]. Table I lists the numbers of events expected and observed for each process as a function of jet multiplicity.
We use a multivariate analysis technique called Bayesian neural networks (BNN) [34] to separate the signal from the backgrounds. The BNN discriminant is trained using the lepton and jets four-vectors, a twovector for / E T , and variables that include lepton charge and b-tagging information. In addition, four angular variables are added based on top quark spin and W boson helicity information to provide additional discriminating power. The total number of variables used in training for events with 2, 3, and 4 jets is 18, 22, and 26, respectively [9]. Three example distributions from some of the most sensitive variables are shown in Figure 2: the p T spectrum of the lepton from the decay of the top quark and the cosine of the angles between the lepton, the leading b-tagged jet and the reconstructed top quark.
For each of the three scenarios, we consider the anomalous coupling sample as the signal when training BNN discriminants: for (L V , R V ), the signal is the single top quark sample generated with f RV = 1; for (L V , R T ), the signal is the sample generated with f RT = 1; for (L V , L T ), the signal is the sample generated with f LT = 1. The background includes the SM single top quark sample with f LV = 1 and all the backgrounds described above. Each background component is represented in proportion to its expected fraction given by the background model. Figure 3 shows representative BNN discriminant output distributions for the three different scenarios with all six analysis channels combined. We follow a Bayesian statistical approach [3,35,36] to compare data to the signal predictions given by different anomalous couplings using BNN discriminant output distributions. We compute a two-dimensional (2D) posterior probability as a function of |V tb · f LV | 2 and |V tb · f X | 2 , where V tb · f X is one of the two non-SM couplings X = {R V , L T }. For these two cases the single top quark contribution is represented by a superposition of two samples: where s LV (s X ) are the mean expected count of single top quarks for the assumptions f LV = 1 (f X = 1) and the other couplings are set to zero. In the (L V , R T ) scenario, the two couplings interfere, and to account for the effect of the interference, the single top quark contribution is represented by the superposition of three samples: where s RT is the mean count assuming a left-handed tensor coupling only f RT = 1, and s LV RT is the one where both couplings f LV = 1 and f RT = 1. The last sample is indicated as "L V + R T " in Fig. 3f. We assume a Poisson distribution for data counts and uniform prior probability for nonnegative values of the SM and non-SM couplings. The output discriminants for the signal, backgrounds, and data are used to form a binned likelihood as a product over all six analysis channels and all bins, taking into account all systematic uncertainties and their correlations. The expected posterior probabilities are obtained by setting the number of data counts to be equal to the predicted sum of the signal and backgrounds. Figure 4 shows the 2D posterior probability density distributions for the three scenarios. We do not observe significant deviations from the SM expectations and therefore compute 95% C.L. upper limits on the anomalous couplings by integrating out the left-handed vector coupling to get a one-dimensional posterior probability density. The measured values are given in Table II. With the SM constraint on the left-handed vector coupling, i.e. |V tb · f LV | = 1, the 95% C.L. limits on left-handed tensor, right-handed vector and tensor couplings are |V tb · f LT | 2 < 0.05, |V tb · f RV | 2 < 0.50 and |V tb · f RT | 2 < 0.11, respectively.
In summary, we have presented a search for anomalous W tb couplings using 5.4 fb −1 of D0 data in the single top quark final state. We find no evidence for anomalous couplings and set 95% C.L. limits on these couplings. These represent improvements in the limits by factors of 2.6 to 5.0 in terms of couplings squared compared to the previous results [18] while a factor of approximately 2.5 is expected from the increase in integrated luminosity. This result represents the most stringent direct constraints on