Search for pair-produced dijet resonances in four-jet final states in pp collisions at sqrt(s) = 7 TeV

A search for the pair production of a heavy, narrow resonance decaying into two jets has been performed using events collected in sqrt(s) = 7 TeV pp collisions with the CMS detector at the LHC. The data sample corresponds to an integrated luminosity of 5.0 inverse femtobarns. Events are selected with at least four jets and two dijet combinations with similar dijet mass. No resonances are found in the dijet mass spectrum. The upper limit at 95% confidence level on the product of the resonance pair production cross section, the branching fractions into dijets, and the acceptance varies from 0.22 to 0.005 pb, for resonance masses between 250 and 1200 GeV. Pair-produced colorons decaying into q q' are excluded for coloron masses between 250 and 740 GeV.


1
The high center-of-mass energy provided by the Large Hadron Collider (LHC) offers opportunities to search for physics beyond the standard model (SM) and in particular to search for new strongly interacting particles. Searches for new resonances in the dijet mass spectrum for the jets with the highest transverse momentum (p T ) have been performed at both the Tevatron [1] and the LHC [2][3][4][5][6][7]. These searches were not optimized for pair-produced particles, which are predicted by some models [8][9][10]. The ATLAS experiment has performed a search for pair production of dijet resonances in four-jet events [11,12] and excludes a scalar gluon at 95% confidence level (CL) in the mass range between 100 and 287 GeV.
In this Letter, we report the results of a search for the pair production of a narrow resonance, which decays into two jets, using the dijet mass spectrum in four-jet final states, measured with the Compact Muon Solenoid (CMS) detector in pp collisions at the LHC at √ s = 7 TeV. This search focuses on the high mass range between 250 and 1200 GeV. We define the common dijet mass as the average of the invariant masses of the two dijets with the smallest difference in invariant mass, rejecting events where the difference exceeds 15% of the average value. Models [8] predicting pair production through gg interactions of color-octet vectors, also called colorons (C), and color-octet scalars (S 8 ) would give rise to such a signature. In addition to qq decays, the coloron can also decay into a pair of color-octet scalars if the mass of the coloron is greater than twice the mass of the two scalars. Pair-produced colorons decaying to quarkantiquark pairs (gg → CC → qqqq) [8] are used as the benchmark signal in this analysis, although we separately consider the possibility of decays to S 8 S 8 . As a third possibility, we consider an R-parity violating SUSY model [13,14] in which pair-produced top squarks (stops) each decay to qq, as this leads to a similar final state.
The CMS detector is a multipurpose apparatus and is described in detail in Ref. [15]. The CMS coordinate system has its origin at the center of the detector, the z axis along the direction of the counterclockwise circulating proton beam, the y axis normal to the LHC plane pointing vertically upward, and the x axis radially inward toward the center of the LHC ring. We define φ to be the azimuthal angle, θ the polar angle, and η = − ln[tan(θ/2)] the pseudorapidity. The central feature of the CMS apparatus is a superconducting solenoid with a 6 m internal diameter, operating at a central field strength of 3.8 T. Within the field volume are, in order of increasing radius, a silicon pixel and strip tracker, a high-granularity PbWO 4 crystal electromagnetic calorimeter (ECAL), and a brass/scintillator hadron calorimeter (HCAL). All three systems have both barrel and endcap components, with |η| = 2.5 (3.0) the outermost extent of the tracker (calorimeters). The ECAL and HCAL cells are grouped into towers, projecting radially outward from the origin. Outside of the field volume an iron/quartz-fiber hadron calorimeter covers the forward region (3 < |η| < 5). A muon system encloses the central and endcap calorimeters out to |η| = 2.4.
The data sample used for this analysis was collected in 2011 and corresponds to an integrated luminosity of 5.0 fb −1 . The triggers used for the analysis require the presence of at least four jets, based on information from the calorimeters. Each jet must have |η| less than 3.0 and p T greater than 70 or 80 GeV, depending on the running period. This trigger is 99.5% efficient for events with four leading (highest p T ) jets, each with a transverse momentum exceeding 110 GeV.
For offline reconstruction, we employ the CMS particle-flow algorithm [16] in the region |η| ≤ 2.5 to reconstruct objects used in jet determinations. This algorithm uses calorimeter information and reconstructed tracks to identify electrons, muons, photons, and both neutral and charged hadrons. Jets are reconstructed from particle-flow objects using the anti-k T algorithm with a distance parameter 0.5 [17,18]. Jet energy is corrected to account for the nonlinearities and nonuniformities in the response of the calorimeters, as determined from Monte Carlo (MC) simulation, test beam, and collision data [19]. Additional corrections accounting for the effect of multiple pp collisions per bunch crossing are also applied [20,21].
We require events to have at least one good primary vertex with a z position within 24 cm of the center of the detector and with a transverse distance from the beam spot of less than 2 cm. A set of jet quality criteria are applied to remove possible instrumental and non-collision backgrounds [22]. All data as well as all simulated signal events passing these selection criteria also satisfy standard jet identification requirements [23]. We require that events have at least four jets, each with p T > 110 GeV and |η| < 2.5. We require the two jets in each possible pair to have a separation ∆R jj = (∆η) 2 + (∆φ) 2 ≥ 0.7. This ensures a negligible overlap between the jets. We calculate the dijet-mass combinations from the four leading jets and choose the one with the smallest ∆m/m avg , where ∆m is the mass difference between the two dijets and m avg is their average mass. We require ∆m < 0.15m avg , which is approximately three times the dijet mass resolution of 4.5%.
The benchmark signal events are simulated using the MADGRAPH v5 [24] event generator with the CTEQ6L1 parton distribution functions (PDF) [25], and PYTHIA v6.4.26 [26] parton showering and hadronization. The generated events are further processed through a GEANT4 [27] simulation of the CMS detector. The assumed width of the simulated coloron resonance is negligible compared with the experimental resolution. The dominant background arises from QCD processes resulting in four or more jets. Studies of this background are performed using a sample of simulated QCD events also generated using MADGRAPH.
For each dijet we define a quantity ∆ as the difference between the scalar sum of the transverse momenta of the two jets in the dijet and the average pair mass in the event: Figure 1 shows the distribution of ∆ versus m avg for simulated QCD background events as well as for coloron signal events. Due to the selection requirements, we observe a broad structure at around m avg = 300 GeV from QCD events [28]. To remove this structure, thus leaving a smoothly falling dijet mass spectrum, we require ∆ > 25 GeV for each of the two dijets in the event. This requirement reduces the QCD background by more than an order of magnitude while retaining approximately 25% of the signal. Figure 2 shows the paired dijet mass spectrum in data with all the selection criteria applied. The observed mass spectrum extends up to 1200 GeV. We obtain a prediction for the QCD background by fitting the data to a smooth parametrization: where P 0 , P 1 , P 2 , and P 3 are free parameters. This functional form has been used in previous searches for dijet resonances [4]. The fit to the data and the normalized QCD simulation are given in Fig. 2  The signal shapes from colorons and stops have negligible difference and we use a single parameterization for both. It is modeled by the sum of two separate Gaussian functions: one Gaussian describes the core and the other the tails, with widths and normalizations determined from a fit to simulated signal events at each assumed mass value. The dijet mass resolution described by the rms of the core Gaussian is approximately 4.5%, the dijet mass tail described by the rms of the other Gaussian is between 150 and 250 GeV, and the fraction of the core Gaussian varies from 85% at 300 GeV to 45% at 1000 GeV. The signal acceptance, listed in Table 1, varies from 0.4% for a coloron with mass 200 GeV to 12.1% for a coloron with mass 1000 GeV. The acceptance for the stop signal is larger than that for the coloron signal because the stop production model includes qq interactions and has a different final state angular distribution.
We search for pair production of dijet resonances by fitting the background parameterization in Eq. (1) plus a signal hypothesis at an assumed mass to the data. The signal magnitude is a free parameter in the fit for each fixed value of signal mass. We restrict the fit to dijet masses above 220 GeV in order to avoid the threshold due to the jet p T requirement. With this requirement the dijet mass spectrum is described by both the simulation and the background parameterization. The maximum value of the likelihood obtained from a background-only fit is denoted by L 0 and the likelihood from a background plus signal fit by L S . The local significance is defined as −2 ln(L 0 /L S ). The results of this analysis are shown in Fig. 2. We find that the largest fluctuation in the pair-produced dijet mass spectrum occurs for a hypothetical resonance mass of 1130 GeV and has a local statistical significance of 2.6σ. The global significance is reduced to 1.2σ after taking into account the trials factor [29] within the full mass range of this search. We conclude that there is no evidence for pair-produced narrow dijet resonances in the data.
Upper limits are placed on the product of the production cross section of the pair-produced res- onances, the square of the branching fraction to dijets, and the detector acceptance. The dijet mass for the limit is required to be above 250 GeV to ensure a full coverage of the low mass tail of the resonance between 220 and 250 GeV. To set upper limits we use a modified-frequentist method (CL s ) [30,31]. We fit the signal+background hypothesis to the data, allowing both the signal strength and the background parameters free to vary. The sources of systematic uncertainties consist of a luminosity uncertainty of 2.2% [32] and a signal acceptance uncertainty of 10%. The latter is determined by the jet energy scale uncertainty (2.2%) and the jet energy resolution uncertainty (10%) [19]. The variation in expected signal yields due to PDF uncertainties is negligible. The uncertainties in the luminosity, the signal acceptance due to jet energy scale and resolution, and the parameters of the background function are all treated as nuisance parameters and expressed as log-normal distributions with their central values and uncertainties. The observed and expected limits are calculated using the CL s method with a one-sided profile likelihood test statistic. Figure 3 shows the observed and expected 95% CL limits, the 1σ and 2σ uncertainty bands around the expected limits, and predictions from the coloron and SUSY models. The observed limit on the product of the resonance pair production cross section, the branching fractions into dijets, and the acceptance varies from 0.22 to 0.005 pb for resonance masses between 250 and 1200 GeV. The limits are generally applicable for pair-produced resonances, each decaying to dijets, and they are compared with calculations for the coloron model [8] described above. At 95% CL we exclude pair production of colorons with mass m C in the range 250 < m C < 740 GeV, assuming that colorons have flavor-universal couplings and decay only into qq [10]. Assuming the branching fraction of colorons into qq is reduced due to competition with a C → S 8 S 8 channel where m S 8 = 150 GeV and tan θ = 0.3 (the suppression factor of gluon coupling to qq compared with the analogous QCD coupling) [10], we exclude pair production in the range 250 < m C < 580 GeV. This analysis is not sensitive to the pair-produced S 8 , where the color-octet scalars decay exclusively to qq. We also compare the results with those of a SUSY model for pair-produced stops, where the stops decay exclusively to qq and R-parity is violated [13,14]. The calculation is done at next-to-leading order (NLO) with next-to-NLO corrections [33][34][35][36][37].  In summary, a search for pair production of a narrow dijet resonance has been performed with the CMS detector using 5.0 fb −1 of √ s = 7 TeV pp collisions produced at the LHC. The paired dijet mass spectrum is found to be a smooth distribution and is in agreement with the predictions of the standard model. Upper limits are reported on the product of the production cross section, the branching fractions into dijets, and the acceptance of a pair-produced dijet resonance having a width negligible compared with the experimental resolution. At 95% CL, the pair production of colorons is excluded for coloron masses between 250 and 740 GeV assuming that a coloron decays 100% into qq, or between 250 and 580 GeV assuming that coloron decays into qq compete with decays into S 8 S 8 . The search significantly extends previous results [12].
We congratulate our colleagues in the CERN accelerator departments for the excellent performance of the LHC machine. We thank the technical and administrative staff at CERN and other CMS institutes, and acknowledge support from BMWF and FWF (  [6] ATLAS Collaboration, "ATLAS search for new phenomena in dijet mass and angular distributions using pp collisions at √ s = 7 TeV", (2012). arXiv:1210.1718. Submitted to JHEP.