Search for scalar leptoquarks in the acoplanar jet topology in ppbar collisions at sqrt(s)=1.96 TeV

A search for leptoquarks has been performed in 310 pb-1 of data from ppbar collisions at a center-of-mass energy of 1.96 TeV, collected by the D0 detector at the Fermilab Tevatron Collider. The topology analyzed consists of acoplanar jets with missing transverse energy. The data show good agreement with standard model expectations, and a lower mass limit of 136 GeV has been set at the 95% C.L. for a scalar leptoquark decaying exclusively into a quark and a neutrino.

topology analyzed consists of acoplanar jets with missing transverse energy. The data show good agreement with standard model expectations, and a lower mass limit of 136 GeV has been set at the 95% C.L. for a scalar leptoquark decaying exclusively into a quark and a neutrino.
PACS numbers: 13.85.Rm Many extensions of the standard model (SM) that attempt to explain the apparent symmetry between quarks and leptons predict the existence of leptoquarks (LQ) [1]. These new particles are scalar or vector bosons that carry the quantum numbers of a quark-lepton system. They are expected to decay into a quark and a charged lepton with a branching fraction β, or into a quark and a neutrino with a branching fraction (1 − β). At pp colliders, leptoquarks can be pair produced, if sufficiently light, primarily by qq annihilation and gluon-gluon fusion, with a production cross section independent of the unknown leptoquark-quark-lepton coupling. For β = 0, the resulting final state consists of a pair of acoplanar quark jets with missing transverse energy, / E T , carried away by the two neutrinos.
In this Letter, a search for leptoquarks that decay into a quark and a neutrino, using data collected at a centerof-mass energy of 1.96 TeV with the D0 detector during Run II of the Fermilab Tevatron Collider, is reported. The production cross section for vector leptoquark pairs is larger than that for scalar leptoquarks, but it is model dependent. The interpretation of the results is therefore presented in terms of scalar leptoquark masses. The most constraining 95% C.L. lower mass limit for β = 0, previous to this search, was 117 GeV, obtained by the CDF Collaboration with 191 pb −1 of Run II data [2].
A detailed description of the D0 detector can be found in Ref. [3]. The central tracking system consists of a silicon microstrip tracker and a fiber tracker, both located within a 2 T superconducting solenoidal magnet. A liquid-argon and uranium calorimeter covers pseudorapidities up to |η| ≈ 4.2, where η = − ln [tan (θ/2)] and θ is the polar angle with respect to the proton beam direction. The calorimeter consists of three sections housed in separate cryostats: the central one covers |η| ∼ < 1.1, and the two end sections extend the coverage to larger |η|. The calorimeter is segmented in depth, with four electromagnetic layers followed by up to five hadronic layers. It is also segmented in projective towers of size 0.1 × 0.1 in η-φ space, where φ is the azimuthal angle in radians. Calorimeter cells are formed by the intersections of towers and layers. Additional energy sampling is provided by scintillating tiles between cryostats. An outer muon system, covering |η| < 2, consists of a layer of tracking detectors and scintillation trigger counters in front of 1.8 T iron toroids, followed by two similar layers beyond the toroids.
For this search, data collected with a jets + / E T trigger have been analyzed. At the first level, this trigger selects events in which at least three calorimeter trigger towers of size ∆φ × ∆η = 0.2 × 0.2 record a transverse energy in excess of 5 GeV. At the second and third trigger levels, requirements are placed on / H T , the vector sum of the jet transverse momenta (/ H T = | jets − → p T |). Coarse jets are reconstructed from trigger towers at the second level, while the full detector information is used at the third level. The / H T thresholds are 20 and 30 GeV at the second and third levels, respectively. The trigger efficiency is larger than 98% for events fulfilling the selection criteria of this analysis. Data quality requirements on the performance of each detector subsystem yielded an available integrated luminosity of 310 pb −1 .
The offline analysis utilized jets reconstructed with the iterative midpoint cone algorithm [4] with a cone size of 0.5. The jet energy calibration was derived from the transverse momentum balance in photon+jet events. Only jets with p T > 15 GeV that passed general quality criteria, based on the jet longitudinal profile in the calorimeter, were selected for this analysis. The missing transverse energy was calculated from all calorimeter cells, corrected for the energy calibration of reconstructed jets and for the momentum of reconstructed muons.
The sample of approximately 14 million events collected with the jets + / E T trigger was reduced by requiring the following preselection criteria to be satisfied: at least two jets; / H T > 40 GeV; / E T > 40 GeV, where, in contrast to / H T , information from energy not belonging to reconstructed jets is taken into account; and ∆Φ < 165 • , where ∆Φ is the acoplanarity of the two leading jets, i.e., the two jets with the largest transverse momenta, defined as the difference between their azimuthal angles. To ensure that the selected events were well contained in the detector, the position of the interaction vertex along the beam direction was required to be within 60 cm of the detector center.
Events in which the presence of obvious calorimeter noise could be detected were rejected. The inefficiency associated with this procedure was measured using events collected at random beam crossings (zero-bias events), and events collected with an unbiased trigger and containing exactly two jets back-to-back in azimuth. At this stage, 306, 937 events survived.
Signal efficiencies and SM backgrounds have been evaluated using a full geant [5] based simulation of events, with a Poisson average of 0.8 minimum-bias events superimposed, corresponding to the luminosity profile in the data sample analyzed. These simulated events were reconstructed in the same way as the data. The jet energies further received calibration corrections and an additional smearing to take into account residual differ-ences between data and simulation, as determined with photon+jet events. The instrumental background due to jet energy mismeasurements in QCD multijet production was estimated directly from the data.
The SM processes expected to yield the largest background contributions are vector boson production in association with jets, among which Z → νν is irreducible. Vector boson pair production and top quark production have also been considered. All of these processes were generated with alpgen 1.3 [6], interfaced with pythia 6.202 [7] for the simulation of initial and final state radiation and for jet hadronization. The parton distribution functions (PDFs) used were CTEQ5L [8]. The nextto-leading order (NLO) cross sections for vector boson production in association with jets were calculated with mcfm 3.4.4 [9] and the CTEQ5M PDFs.
The production of scalar leptoquarks via the processes qq or gg → LQLQ was simulated with pythia and the CTEQ5L PDFs. The chosen leptoquark masses ranged from 80 to 140 GeV, in steps of 5 GeV. For each mass, 10,000 events were generated. The NLO leptoquark pair production cross sections were calculated using a program based on Ref. [10], with CTEQ6.1M PDFs [11]. For the mass range considered, they vary from 52.4 to 2.38 pb. These nominal values were obtained for a renormalization and factorization scale equal to the leptoquark mass.
The selection criteria for this analysis are listed in Table I, together with the numbers of events surviving at each step and with the cumulative efficiency for a leptoquark mass of 140 GeV. The jet kinematic cuts C1 to C4 reject a large fraction of the SM and instrumental backgrounds. They take advantage of the central signal production and decay by requiring that |η det | be smaller than 1.5 for the two leading jets, where η det is the pseudorapidity measured from the detector center. Cut C5, where EMF is the fraction of jet energy contained in the electromagnetic section of the calorimeter, rejects jets likely due to photons or electrons.
In cut C6, the total transverse energy of the charged particles emanating from the interaction vertex and associated with a jet, as measured in the tracking system, is compared to the jet transverse energy recorded in the calorimeter. The charged particle fraction CPF, i.e. the ratio of these two quantities, is expected to be close to zero either if a wrong interaction vertex was selected, in which case it is unlikely that the charged tracks truly associated with the jet will come from the selected vertex, or if the jet is a fake one, e.g. due to calorimeter noise, in which case there should be no real charged tracks associated with it. The efficiency of this jet confirmation procedure was determined using events containing two jets back-to-back in azimuth.
Cut C7 was applied to suppress further the instrumental background, which is enriched in multijet events by the acoplanarity requirement. The efficiency of such a jet multiplicity cut is sensitive to the modeling of initial and final state radiation (ISR/FSR). To verify the simulation of these effects, (Z → ee)+ ≥ 2-jet events were selected in the data, and compared to a simulation by alpgen for the production of (Z → ee)+2-jets, with ISR/FSR jets added by pythia. The two leading jets were required to fulfil criteria similar to those used in the analysis, and the numbers of events with additional jets were compared between data and simulation, as well as the p T spectra of those jets. The small deficit observed in the simulation, located mostly at p T < 20 GeV, was used to correct the signal and background simulations, and the statistical power of this test was taken as a systematic uncertainty. After cut C8, the level of the instrumental background is largely reduced and is similar to the level of the SM backgrounds. The final / E T cut value (cut C14) was optimized as explained below.
Cuts C9, C10 and C11, reject a large fraction of the events originating from W/Z+jet processes. In cut C9, an electron with p T > 10 GeV is declared isolated if the calorimeter energy in a cone of radius 0.4 in η-φ around the electron direction does not exceed the energy contained in the electromagnetic layers inside a cone of radius 0.2 by more than 15%. In cut C10, a muon with p T > 10 GeV is declared isolated if the calorimeter energy in a hollow cone with inner and outer radii 0.1 and 0.4 around the muon direction is smaller than 2.5 GeV, and if the sum of the transverse energies of charged tracks, other than the muon, in a cone of radius 0.5 is smaller than 2.5 GeV. In cut C11, a charged track with p T > 5 GeV is declared isolated if no charged track with p T > 0.5 GeV is found within a hollow cone of radii 0.1 and 0.4 around the track considered. This cut was specifically designed to reject hadronic decays of τ -leptons; the use of a hollow, rather than full cone renders it efficient also in case for data (points with error bars), for SM backgrounds (shaded histograms), and for a 140 GeV LQ signal (hatched histograms). In the ∆Φmax − ∆Φmin distribution, cuts C1 to C11 are applied. The excess in data beyond 120 • is attributed to the non-simulated instrumental background. In the ∆Φmax + ∆Φmin distribution, the cut ∆Φmax − ∆Φmin < 120 • (C12) has been applied in addition. The locations of cuts C12 and C13 are indicated by arrows in (a) and (b), respectively. of decays into three charged particles.
The angular correlations between the jet and / E T directions are used to suppress both the instrumental and SM backgrounds. To this end, the minimum ∆Φ min (/ E T , any jet) and maximum ∆Φ max (/ E T , any jet) of the azimuthal angle differences between the / E T direction and the direction of any of the two jets are combined as shown in Fig. 1. It can be seen in Fig. 1a that cut C12 rejects most of the remaining instrumental background, which is responsible for the excess beyond 120 • . Cut C13, which suppresses SM backgrounds at the expense of a moderate reduction of the signal efficiency, was optimized as explained below. The variable ∆Φ max + ∆Φ min is the one which discriminates best the signal and the irreducible background from (Z → νν)+2-jets. Its effect is demonstrated in Fig. 1b.
Finally, the / E T and ∆Φ max + ∆Φ min cuts were optimized for a 140 GeV LQ mass so as to minimize the cross section expected to be excluded in the absence of signal. Cut C8 was removed, and / E T cut values ranging from ET for data (points with error bars), for SM backgrounds (heavy-shaded histograms), for the instrumental background (labeled QCD, light-shaded histograms), and for a 140 GeV LQ signal. In (a), all cuts except C8 and C14 are applied, the LQ signal is shown as a hatched histogram, and the insert shows how the instrumental background contribution is estimated from power law (solid curve) and exponential (dashed curve) fits. The / ET distribution in (b) is after all cuts, with the same shading code but with the signal contribution now displayed on top of all backgrounds. 60 to 90 GeV were probed in 10 GeV steps. The cut on ∆Φ max + ∆Φ min was varied between 260 • and 300 • in steps of 10 • . For each set of cuts, the instrumental background was estimated as explained below. The systematic uncertainties discussed further down were taken into account in the calculation of the expected limits. The optimal set of cuts reported as C13 and C14 in Table I selects 86 data events.
The instrumental background was estimated from exponential and power law fits to the / E T distribution (insert of Fig. 2a) in the range [40, 60] GeV, where the signal contribution is negligible, after subtraction of the SM expectation. Both fits were extrapolated beyond the / E T cut value, and the average of the two results was taken as the instrumental background estimate, with a systematic uncertainty accounting for the difference between the two fit results. The final / E T distribution is shown in Fig. 2b. The values of the SM and instrumental backgrounds are given in Table II. The largest background sources are, as expected, (Z → νν)+2-jets and II: Numbers of events expected from standard model, instrumental and total backgrounds; number of data events selected; and number of signal events expected for mLQ = 140 GeV, assuming the nominal production cross section. For the total SM and total backgrounds, as well as for the signal, the first uncertainties are statistical and the second systematic. The uncertainties on the individual SM backgrounds are statistical. The uncertainty on the instrumental background is mostly systematic from the difference between the power law and exponential fits. (W → ℓν)+jets (ℓ = e, µ, τ ).
The signal efficiencies at various stages of the analysis are given in Table I for m LQ = 140 GeV. The efficiency decreases together with the leptoquark mass, reaching 1.6% at 100 GeV. The number of signal events expected for a leptoquark mass of 140 GeV is indicated in Table II.
The following sources of systematic uncertainty are fully correlated between SM background and signal expectations: the relative jet energy calibration between data and simulation: +4 −8 % for the SM background and +6 −4 % for the signal; the relative jet energy resolution between data and simulation: +2 −4 % for the SM background and negligible for the signal; the efficiency of the jet multiplicity cut: ±3%, after corrections of −3% for the SM background and −2% for the signal; the trigger efficiency: ±2% after all selection cuts; and the integrated luminosity of the analysis sample: ±6.5%.
In addition to the +14 −13 % statistical uncertainty of the simulation, the normalization of the SM background expectation is affected by a ±12% uncertainty, as inferred from a comparison of data and simulated (Z → ee)+2-jet events selected with the same criteria for the jets as in the analysis sample. The uncertainty of ±1.2 events on the instrumental background was estimated from power law and exponential fits to the / E T distribution, as explained previously. As a check, the same procedure was applied to the events with ∆Φ max − ∆Φ min > 120 • , which are dominated by the instrumental background contribution. This showed that the high / E T tail is somewhat underestimated, possibly by as much as nine events, which leads to conservative results in terms of limit setting. Finally, the uncertainty on the signal efficiency due to the PDF choice was determined to be +6 −4 %, using the twenty-eigenvector basis of the CTEQ6.1M PDF set [11].
As can be seen in Table II and Fig. 2b, no significant excess of events is observed in the data above the background expectation. Therefore, given the number of selected events, the SM and instrumental background expectations, the integrated luminosity of 310 pb −1 , the signal selection efficiency as a function of the leptoquark mass, and the statistical and systematic uncertainties discussed above, a 95% C.L. upper limit on the cross section times (1 − β) 2 has been determined as shown in Fig. 3, using the modified frequentist CL s approach [12]. The expected limit in the absence of signal is also indicated. The nominal theoretical cross section for the pair production of scalar leptoquarks is also shown in Fig. 3. It was obtained based on Ref. [10] with CTEQ6.1M PDFs and for a renormalization and factorization scale µ rf equal to the leptoquark mass. The uncertainty associated with the PDF choice was estimated using the full set of CTEQ6.1M eigenvectors and combined quadratically with the variations obtained when µ rf was modified by a factor of two up or down. For a leptoquark mass of 140 GeV, the PDF uncertainty on the theoretical cross section amounts to +18 −13 % and the scale variation results in a change of +11 −13 %, the quadratic sum being +21 −19 %. Reducing the nominal cross section by this theoretical uncertainty, shown as the shaded band in Fig. 3, a lower mass limit of 136 GeV is derived at the 95% C.L. Masses smaller than 85 GeV, to which this analysis is not sensitive, have been excluded previously [2,13]. The cross section limit obtained here was combined with the results of the published D0 search for first-generation scalar lep-