Search for pair production of first-generation leptoquarks in p pbar collisions at sqrt(s)=1.96 TeV

A search for pair production of first-generation leptoquarks LQ is performed with data collected by the D0 experiment in p pbar collisions at sqrt(s)=1.96 TeV at the Fermilab Tevatron Collider. In a sample of data corresponding to ~1 fb-1, the search has been performed on the final states with two electrons and two jets or one electron, two jets and missing transverse energy. We find our data consistent with standard model expectations. The results are combined with those found in a previous analysis of events with two jets and missing transverse energy to obtain scalar LQ mass limits. We set 95% C.L. lower limits on a scalar LQ mass of 299 GeV, 284 GeV and 216 GeV for beta=1, beta=0.5 and beta=0.02 respectively, where beta is the LQ branching ratio in the eq channel. This improves the results obtained with a lower luminosity sample from Run II of the Tevatron. Lower limits on vector LQ masses with different couplings from 357 GeV to 464 GeV for beta=0.5 are also set using this analysis.

(Dated: July 6, 2009) A search for pair production of first-generation leptoquarks (LQ) is performed with data collected by the D0 experiment in pp collisions at √ s = 1.96 TeV at the Fermilab Tevatron Collider. In a sample of data corresponding to ∼ 1 fb −1 the search has been performed on the final states with two electrons and two jets or one electron, two jets and missing transverse energy. We find our data consistent with standard model expectations. The results are combined with those found in a previous analysis of events with two jets and missing transverse energy to obtain scalar LQ mass limits. We set 95% C.L. lower limits on a scalar LQ mass of 299 GeV, 284 GeV and 216 GeV for β = 1, β = 0.5 and β = 0.02 respectively, where β is the LQ branching ratio in the eq channel. This improves the results obtained with a lower luminosity sample from Run II of the Tevatron. Lower limits on vector LQ masses with different couplings from 357 GeV to 464 GeV for β = 0.5 are also set using this analysis. Leptoquarks are conjectured particles, predicted by many extensions [1] of the standard model (SM). In such exotic models, transitions between the leptonic and baryonic sectors would be allowed. Thereby, the detection of leptoquarks (LQ) could be, among others, the signature of compositeness, supersymmetric couplings in R-parity violating models, Grand Unification models, or technicolor. Leptoquarks can be scalar or vector fields. It is generally assumed that there is no intergenerational mixing, because it is severely constrained by low-energy experiments, and that first-generation LQs couple only to e or ν e and to u or d quarks. At the Fermilab Tevatron Collider, pair production of leptoquarks can proceed through quark-antiquark annihilation (dominant for M LQ ≥ 100 GeV) or through gluon fusion, therefore being independent of the LQ − e − q Yukawa coupling λ. Thus the production cross section for scalar leptoquarks only depends on the strong coupling constant and on the leptoquark mass. In the vector leptoquark case, the production cross section also depends on the anomalous couplings κ G and λ G of the LQ to the gluon. At the CERN e + e − Collider (LEP), pair production of leptoquarks could have occurred in e + e − collisions via a virtual γ or a Z boson in the s-channel. Experiments at the Fermilab Tevatron Collider [2,3] and at the LEP Collider [4] set lower limits on the masses of leptoquarks. The H1 and ZEUS experiments at the DESY e ± p collider HERA published [5] lower limits on the mass of a firstgeneration LQ that depend on the coupling λ. In the case of single LQ production at LEP or at the Tevatron, the mass limits depend also on λ [6]. The branching ratio for LQ or LQ decay into a charged lepton and a quark is denoted as β, so 1 − β is the branching ratio of the reaction LQ → ν + q. The branching ratios of the three decay modes LQLQ → eeqq, LQLQ → eνqq and LQLQ → ννqq are then equal to β 2 , 2β(1 − β) and (1 − β) 2 , respectively.
In this Letter, we present a search for first-generation leptoquarks for two cases: when both leptoquarks decay to an electron and a quark and when one of the leptoquarks decays to an electron and a quark and the other to a neutrino and a quark. The corresponding final states consist of two electrons and two jets (eejj) and one electron, two jets and missing transverse energy (eνjj).
This study is performed on data collected with the D0 detector [7] in pp collisions at √ s = 1.96 TeV during Run II of the Tevatron Collider. The D0 detector comprises three main elements. A magnetic central tracking system, which consists of a silicon microstrip tracker and a central fiber tracker, is located within a 2 T superconducting solenoidal magnet. Three liquid-argon/uranium calorimeters, a central section (CC) covering pseudorapidities [8] |η| up to ∼ 1 and two end calorimeters (EC) extending coverage to |η| ≃ 4, are housed in separate cryostats. Scintillators between the CC and EC cryostats provide a sampling of developing showers for 1.1 < |η| < 1.4. A muon system is located outside the calorimeters and covers the region |η| < 2. The luminosity is measured using plastic scintillator arrays placed in front of the EC cryostats. The data samples for the eνjj and eejj analyses are selected with combinations of single electron and electron plus jets triggers. The corresponding integrated luminosity is ∼1 fb −1 .
Electrons are defined as clusters of energy deposition in the calorimeters with a high fraction (> 90%) deposited in the electromagnetic (EM) sections. The energy cluster must be isolated from other energy deposits in the calorimeter [9] and matched with a charged particle with transverse momentum p T > 5 GeV. A condition on the value of an electron likelihood based on a shower shape parameter and conditions on the number of tracks in the vicinity of the electron are applied. Electrons that fulfill all the above criteria except the likelihood condition are classified as loose electrons. Those which satisfy all criteria are referred to as tight electrons.
Jets are reconstructed with an iterative cone algorithm [10] with radius of 0.5 and a minimal distance R > 0.5 [9] from any EM object. The jet energy scale (JES) corrections were derived from the transverse momentum balance in photon-plus-jet events. The missing transverse energy E / T is calculated from all calorimeter cells, and corrected for the jet energy scale and for the transverse momenta of reconstructed muons.
Scalar LQ Monte Carlo samples with masses from 140 GeV to 320 GeV have been generated with pythia [11] using the CTEQ6L1 [12,13,14] parton density functions (PDFs). Two processes are generated: qq → LQLQ (dominant for LQ masses above 100 GeV) and gg → LQLQ [15]. The LQs are treated as resonances and their isotropic decay mode is to a u quark and an electron. The pythia code has therefore been slightly modified to allow that one of the LQs decays into a d quark and a neutrino. The LQ → ql vertex depends on the Yukawa coupling λ which affects the width of the LQ. We have taken λ equal to the electromagnetic coupling √ 4πα. The next-to-leading order (NLO) cross section of scalar LQ pair production has been calculated in Ref. [16]. To generate the vector leptoquarks, the model described in Ref. [17] and implemented in comphep [18] is used. In this model, the leading order (LO) cross section depends on the LQ mass and on the anomalous couplings of the LQ to the gluon, κ G and λ G . In the following, three types of couplings have been considered: "MC" coupling {κ G = 1, λ G = 0}, "YM" coupling {κ G = 0, λ G = 0} and "MM" coupling {κ G = −1, λ G = −1}. We have generated pairs of vector leptoquarks with masses between 200 GeV and 480 GeV that decay, as in the scalar LQ case, into an electron and a quark or into a neutrino and a quark, and we have also used a λ = √ 4πα. The main SM backgrounds relevant to these final states are the associated production of jets with Z/γ * or W boson and top quark pairs in dilepton or semi-leptonic chan-nels. Less important contributions come from Z/γ * → τ τ (τ → e), single top quark decaying into e or τ , and Diboson final states including jets. Most of the samples were generated with alpgen [19] interfaced with pythia for parton showering and hadronization. Exceptions are the diboson and single top processes, which were generated with the pythia and the singletop [20] event generators, respectively. The PDFs used are CTEQ6L1. The alpgen inclusive W/Z production cross section is normalized to the NLO theoretical prediction using Kfactors derived by comparing the LO and NLO cross sections in mcfm [21]. All the SM generated backgrounds are normalized to the integrated luminosity of data sample.
Signal and background Monte Carlo samples are processed through a geant-based [22] simulation of the D0 detector and the same reconstruction program as used for the collider data. To model the effects of detector noise and multiple pp interactions, each Monte Carlo event is overlaid with a data event from a random pp crossing. Monte Carlo samples pass the same selection criteria as the data samples. But since the efficiency of these selections is different for data and for Monte Carlo, efficiency corrections are applied to the simulated events: the trigger probability (η and p T -dependent efficiencies for the chosen single electron triggers), a correction for the efficiencies of the jet selection, an η and φ dependent correction of the electron selection efficiency, and a correction to reproduce the luminosity profile of the data and the distribution along beam axis of the event primary vertex.
In the eejj analysis, events are selected with at least two isolated electrons satisfying tight identification criteria, with p T ≥ 25 GeV and at least one of the two detected in the central part of the calorimeter (|η| ≤ 1.1) The selected events must also contain one or more jets with p T ≥ 25 GeV and |η| ≤ 2.5. In addition to the main SM backgrounds described above, an instrumental background consists of multijet processes (MJ), and is due to the misidentification of jets as electrons. This contribution is extracted from data. A specific sample containing events with two "fake" electrons and at least one additional jet, where a "fake" electron is an isolated cluster in the calorimeter with the usual EM fraction value for a loose electron but shower shape conditions relaxed, is used to reproduce the shapes of the kinematical distributions. The normalization of the total expected background to the number of data events in two regions of the M ee spectrum (50 < M ee ≤ 80 GeV and 80 < M ee ≤ 102 GeV) gives the MJ and Z/γ * + jets sample contributions. The tt and Diboson contributions are normalized to the luminosity. Two normalization factors are extracted and further used to determine the number of background events in the sample obtained when the requirement of a second jet with p T ≥ 25 GeV is added.
After the requirements of two electrons and two jets, 448 events remain in the data sample, with 449 ± 13 predicted background events of which 91% originates from the Z/γ * → e + e − samples. The dielectron invariant mass M ee and the transverse scalar energy S T (see Fig. 1), defined as the scalar sum of the p T of the two electrons and the two highest E T jets, are used as discriminant variables in this analysis. Most Z/γ * → e + e − events are concentrated around the mass of the Z boson (80 < M ee < 102 GeV), and the multijet contribution populates the region S T < 300 GeV. To suppress as much background as possible while minimizing any reduction of signal acceptance, the selections on the M ee and S T variables have been optimized. For different sets of requirements on these variables, we combine the numbers of expected signal and background events, and their uncertainties, from the bins of the average electron-jet invariant mass distribution to calculate the expected upper limit on the cross section at 95% C.L. We used a modified frequentist CLs method, based on a likelihood ratio as described in Ref. [23]. The effects of systematic uncertainties on the signal and background, taking into account correlations, are included in the resulting limits. The best sensitivity is obtained for M ee > 110 GeV and S T > 400 GeV. After all selections, no data events remain, for an expected SM background of 1.51 ± 0.12(stat) ± 0.04(syst) events (see Table I). The acceptance for a scalar LQ with a mass between 250 GeV and 300 GeV varies between 20% and 23%. The acceptances for the vector LQs are similar. In the eνjj analysis we select events containing exactly one isolated electron satisfying tight identification criteria with p T ≥ 25 GeV and |η| < 1.1, and with E / T ≥ 35 GeV. The selected events must also contain at least two high p T jets with |η| < 2.5, with the leading jet having p T ≥ 40 GeV and the second leading jet having p T ≥ 25 GeV. A veto on a second tight electron with |η| < 2.5 guarantees that there is no overlap with the eejj analysis. Multijet processes again contribute to an instrumental background. A fake electron could be present due to misidentification of one jet, and the E / T could be due to the resolution of the jet energy measurement. Events with ≥ 3 jets can thus be reconstructed as eνjj events. In these events, the E / T tends In order to model the multijet contribution, a sample containing events with one "fake" electron and ≥ 2 additional jets is created. The number of multijet background events is computed using the method described in Ref. [24]. Two samples of events are used, the first one contains events with a loose electron and the second one, which is a subsample of the first one, is composed of events with a tight electron. Using the number of events in these two samples together with the efficiencies for a real and a "fake" electron to pass the likelihood condition, referred to as ǫ SM and ǫ MJ respectively, we can determine the number of MJ events. We measure ǫ SM as the ratio of the number of Monte Carlo events which pass the likelihood condition over the number of Monte Carlo events which fail it and correct for differences between data and simulation. We measure ǫ MJ directly from data assuming that the low E / T region (E / T ≤ 10 GeV) is dominated by the multijet background after subtracting a small contribution of real electrons determined from Monte Carlo. The number of Monte Carlo W + jets events is normalized to data within a range of the transverse invariant mass of the electron and the E / T where the expected number of LQs is very small: M T (e, E T ) ≤ 100 GeV. There is good agreement between data and expected SM background both in number of events and in the shape of the distributions. The M T (e, E T ) distribution is shown in Fig. 2 with the signal for a scalar LQ sample for M LQ = 250 GeV superimposed. The number of data events that pass the selection criteria is equal to 3563 which is in good agreement with the total expected background of 3549 ± 68 events, of which 87% come from W + jets events. A cut M T (e, E T ) ≥ 130 GeV strongly reduces this background. Other discriminants are the p T distributions of the decay products of the two LQs. We determined the best p T cuts as described in the eejj analysis, but using the S T distribution, where S T is the sum of the p T of the electron, the p T of the two leading jets, and E / T . The best expected cross section limits are obtained for a cut of 80 GeV on both the p T of the electron and the E / T , and the cuts p T (leading jet) > 40 GeV and p T (second jet) > 25 GeV. After all selections, 8 events remain, for an expected SM background of 9.8 ± 0.8(stat) ± 0.8(syst) events (see Table II). The acceptances are similar for scalar or vector LQs. They range from 18.5% to 20% for a LQ mass varying between 250 GeV to 300 GeV. In Fig. 3, the distributions of the masses M (e, jet) and M T (E / T , jet) are shown. The signal for a scalar LQ sample for M LQ = 250 GeV has been superimposed. The agreement is good between data and the SM expectations, both in number of events and in the shape of the distributions.
The values of the systematic uncertainties are summarized in Table III  lution and the jet identification efficiency (Jet ID). The systematic uncertainty from the correction of the electron identification efficiency (EM ID) is evaluated from the uncertainty on the Monte Carlo/data correction factors and by choosing another parametrization of the correction. Other systematics uncertainties affect the luminosity, or are computed by measuring the effect of the PDF choice on the signal acceptances using a different PDF set (20-eigenvector basis CTEQ6.1M NLO PDF). The uncertainties due to the propagation into the analyses of uncertainties on the parameters used in the background normalization are referred as background normalization in Table III. The SM uncertainties are the combined relative uncertainties on the expected background due to uncertainties on the cross sections of the SM processes and to different modeling of jet radiation in the W + jets process. The uncertainties that are shown on the same row are treated as correlated in the determinations of the limits. No deviations from the SM predictions were observed in our data in either the eejj final state or in the eνjj final state and for each individual channel we determined cross section limits on the pair production of a firstgeneration scalar LQ at 95% C.L. The results are shown in Fig. 4 where the expected and observed cross section limits measured in the eejj and eνjj final states are displayed as a function of the LQ mass, assuming β = 1 and β = 0.5 respectively. On the same figure the scalar LQ pair production NLO cross sections, calculated for different values of the renormalization and factorization scales (µ = M LQ , M LQ /2 and 2M LQ ) are also shown.
In D0 analysis [25], using a sample of 2.5 fb −1 of data with acoplanar jets and missing transverse energy, a search for the pair production of first generation scalar leptoquarks both decaying in νq has shown no evidence of this production. We combined these three analyses to determine expected and observed cross section limits as a function of β and M LQ . We used the modified frequentist CLs method referenced in the eejj analysis and the JES, PDF and luminosity systematics uncertainties are treated as correlated errors. As an example, the values of the observed cross section limits are given in Table IV for β = 1 and β = 0.5. For each value of β, the limit is the LQ mass value where the experimental cross section limit and the theoretical cross section are equal. The expected and observed mass limits for factorization and renormalization scales µ equal to M LQ are summarized in Table V. They are shown in the β -M LQ plane in Fig. 5 together with the limits obtained in each final state analysis. The theoretical uncertainty on the observed mass limit, reflecting the PDF, normalization and factorization scale uncertainties, is also shown.
To compute the limit on vector LQ cross sections, as the vector and scalar LQ acceptances are very similar, we use the selections which have been found optimal in the search for a scalar LQ. The expected and observed cross section limits for each of the two final states eejj and eνjj, assuming β = 1 and β = 0.5 respectively, are shown in Fig. 6 as a function of the LQ mass. The vector LQ pair production LO cross sections are also shown, for each of the three couplings. They are calculated for different values of the renormalization and factorization scales µ = M LQ , M LQ /2 and 2M LQ . We combine these results to get expected and observed cross section limits as a function of β. The values of these limits obtained for β = 1 and β = 0.5 are given in Table IV. The expected and observed mass limits for a factorization and renormalization scales equal to M LQ are summarized in Table V. They are shown in the β -M LQ plane in Fig. 7 for the three couplings. The hatched areas show the effect of the theoretical uncertainties on the observed exclusions.
In this analysis of the D0 Run II dataset corresponding to an integrated luminosity of about 1 fb −1 , we have excluded a first-generation scalar LQ with mass varying between 216 GeV for β = 0.02 to 299 GeV for β = 1 assuming µ = M LQ . For µ = 2M LQ , the mass limits range from 206 GeV to 292 GeV. These results improve bounds given in previous LQ searches at Tevatron [2,3] by ≃ 50 GeV. We have also excluded vector LQs for different couplings. As an example for β = 0.5 and µ =