Resonant top pair production could arise from the decays of massive Z-like bosons in extended gauge theories [1,2], KK states of the gluon [3] or Z [4], axigluons [5], topcolor [6], and other BSM theories. We study t-tbar invariant mass spectrum in 1 fb-1 of CDF Run II data and set model independent limits on new resonant t-tbar production mechanisms that would appear as bumps in invariant mass.
The data sample is standard b-tagged top mass selection requiring a central lepton with Et > 20 GeV, missing Et > 20 GeV, and 4 jets with |eta| < 2.0, of which 3 must have ET > 15 GeV, a fourth must have ET > 8 GeV, and at least one contains a secondary vertex b-tag. For the 955 pb-1 Run II data set, we find 347 candidate events.
The sample is dominated by SM t-tbar production, which is modeled with Herwig at mt = 175 GeV /c2. Contributions from smaller non-top backgrounds are modeled with standard techniques from the top cross section analysis. 73 ± 9 background events are expected.
Resonant t-tbar production is modeled by a heavy neutral boson with the same couplings as the Z0, Pythia process 141, decaying to 175 GeV/c2 top pairs. We simulated Z' masses between 450 and 900 GeV/c2 in increments of 50 GeV/c2. The full width of the Z' is narrow, 1.2% of the mass, and this electroweak model does not include any kind of resonant interference with the s channel gluon production, so the signal shape is totally dominated by resolution and combinatoric effects which have been studied in the top mass analysis and found to be well modeled in simulation.
We reconstruct the expected Mtt spectrum using the Chi2 based mass-fit algorithm with the additional constraint that mt = 175 GeV /c2. The lowest Chi2 solution is chosen for each event. In order to reject badly fit outliers inconsistent with the ttbar hypothesis, the reconstruction is required to have Chi2 < 50. The Chi2 < 50 cut removes 20 events leaving a final sample of 327 candidates.
A binned likelihood fit to the three components is performed over the range of Z' masses. The number of SM t-tbar events, non-top events, and Z'->t-tbar cross section are free parameters in the fit, with the non-top events Gaussian constrained to the Method II prediction. Monte Carlo pseudo-experiments with Z'cross sections between 0 and 4 pb are used to verify the performance of the algorithm and estimate sensitivity. We set 95% CL upper limits by integrating to the Z' cross section which gives 95% of the area of the posterior liklihood fuction. The expected 95% CL upper limit with statistical errors only is 1.47 +-0.5 pb for a Z' mass of 500 GeV/2.
The systematic shape uncertainties in the signal and background models are propagated into the measurement by marginalizing the likelihood. The uncertainty in the jet energy scale (JES) and top mass are the main culprits here, with smaller contributions from ISR, FSR, non-top background mix, and W-Q2. We generate separate pseudo samples with 1 sigma shifts in a given effect and calculate the offset in the fitted cross section for the null measurement. The offsets are added in quadrature and used as a cross-section dependent width of the Gaussian function which is convolved with the likelihood. The net effect of these systematic shape uncertainites is a small increase in the limits, roughly 0.2pb at MZ' = 450 GeV/c2, decreasing to 0.1pb at higher Z' masses.
The data and model spectra in the distribution above are subjected to this complete procedure. The measured and expected limits on Z'-like decays to t-tbar are shown in the Table and Figure immediately below, followed by another figure with theoretical interpretation. The measured limits are 1 sigma high in the low mass region (450 to 500 GeV/c2) and consistent with expectations at high mass. We find that the M(t-tbar) spectrum is consistent with the expected QCD result and shows no obvious evidence for additional resonant production mechanisms in top-pair production.
Our result is represented along with some theoretical models in the figure below. Here, the limit in the data is the solid black line and the shaded band around the grey line denotes the 1 sigma uncertainties around the expected upper limit. A leptophobic Z' predicted by the topcolor theory [6], shown as a dotted purple line, is ruled out below 720 GeV/c2. The dotted black curve at the bottom of the chart is the expected cross section for a massive Z' with standard model couplings, calculated with the Herwig Monte Carlo assuming a k-factor of 1.3 [7]. Interestingly, this generic and time-honored model, which underpins our efficiency estimate and analysis method, is still out of range of our sensitivity. The Tevatron cross-section for the KK gluon excitation in the RS model of Ref. 3 is shown as a blue dot-dash line [8]. Since the KK resonance is broad (approx 17% of the mass [3]), our limits derived in the ``narrow width'' assumption are not strictly applicable, but the figure suggests that the KK-top resonance signal at the Tevatron would be striking. Indeed, a recent paper rules out the model of Ref. 3 below 800 GeV/c2 based on the measured Tevatron top cross section alone [9]. The large width of the KK states, properly treated, will allow us to extend our sensitivity to higher masses, and this search will continue in larger Tevatron data sets.
References
1. A. Leike, Phys. Rep. 317, 143, 1999. arXiv:hep-ph/9805494
2. J. Rosner, CERN-TH/96-169, 1996.
3. B. Lillie, L. Randall, L.T.Wang, arXiv:hep-ph/0701166
4. T. Rizzo, Phys.Rev D61 (2000) 055005, arXiv:hep/ph/9909232
5. L. Sehgal, M. Wanninger, Phys. Lett B200, 211, 1988.
6. C. Hill, S. Park, PRD49, 4454, 1994.
7. Thanks to Sam Harper for Z' cross-sections.
8. B. Lillie, private communication
9. M. Guchait, F. Mahmoudi, K. Sridhar, arXiv:hep/ph/0703060
Michael Kagan Dan Amidei Last modified: 10/09/06