[Up] [Previous section] [Next section]
Recently D0 published [1] an analysis based on data with integrated luminosities between 44 and 56~pb^-1, depending on the channel. We observed a top quark signal in both dilepton and single-lepton channels. The dilepton candidates possessed two isolated leptons, at least two jets, and large missing transverse energy; the single-lepton candidates had one isolated lepton, large missing E_T, and a minimum of three jets (with muon tag) or four jets (without tag). We imposed a minimum requirement in all channels on H_T, the scalar sum of the transverse energies E_T of the jets (for the single-lepton and mumu + jets channels) or the scalar sum of the E_T's of the leading electron and the jets (for the emu + jets and ee + jets channels).
For the dilepton channels, the main backgrounds arise from Z and continuum Drell-Yan production (Z,gamma^* -> e e, mu mu, and tautau), vector boson pairs (W W, W Z), heavy flavor (bbar b and cbar c) production, and backgrounds with jets misidentified as leptons. For the single-lepton channels, the main backgrounds are from W + jets, Z + jets, and multijet production with a jet misidentified as a lepton. From all seven channels, we observed 17 events with an expected background of 3.8+-0.6 events.
Our measured cross section as a function of the top quark mass hypothesis is shown here.
Assuming a top quark mass of 199 GeV/c^2, the production cross section is 6.4+-2.2 pb. The error in the cross section includes an overall 12% uncertainty in the luminosity. The probability of an upward fluctuation of the background to 17 or more events is 2times 10^-6, which corresponds to 4.6 standard deviations for a Gaussian probability distribution. We have calculated the probability for our observed distribution of excess events among the seven channels and find that our results are consistent with top quark branching fractions at the 53% CL. Thus, we observe a statistically significant excess of events and the distribution of events among the seven channels is consistent with top quark production.
To measure the top quark mass, single-lepton + four-jet events were subjected to 2-constraint kinematic fits to the hypothesis tbar t-> W^+W^-bbar b->ellnu qbar qbbar b. With the H_T requirement removed and additional kinematic requirements loosened, allowing a substantial background contribution at lower mass, 27 single-lepton + four-jet events remained of which 24 were fitted successfully. An unbinned likelihood fit, incorporating top quark and background contributions, with the top quark mass allowed to vary, was performed on the fitted mass distribution. The background contributions were constrained to be consistent with our background estimates, but the result of the fit did not change significantly when the background normalization was left unconstrained. The best fit top quark mass was 199^+19_-21 (stat.) GeV/c^2. This is consistent with the result obtained from the standard event selection. The total systematic error is 22 GeV/c^2, dominated by the uncertainty in the jet energy scale.
All simulations described here for top physics include detailed ISAJET/GEANT modeling of radiative effects and interactions in material, etc., and use full pattern recognition of tracks with raw hits as input. These simulations yield a p_T resolution for magnetic tracking of electrons from t -> e (t -> b -> e) of 0.002p^2_T (0.003p^2_T). This is independent of and in addition to the approx 0.15p_T^{1/2} resolution available both in Runs I and II from the central electromagnetic calorimeter. With the calorimeter continuing to provide the best information above p_Tapprox 20 GeV/c, the main advantages of magnetic tracking for electrons lie in new capabilities for triggering and for electron identification using E/p, and in identification of the electron charge.
The benefits of magnetic tracking within the central volume enclosed by the calorimeter are dramatic for muons, which will enjoy slightly better momentum resolution from the solenoid than do the electrons. In Run I the muon p_T resolution is bounded at low p_T by approx 0.2p_T due to Coulomb scattering in the steel filter. At high p_T it is limited to approx 0.003p_T^2 by muon drift chamber resolution including alignment. With the addition of central magnetic tracking in Run II, the muon p_T resolution will improve by a factor of about two at high p_T. At lower p_T the muon resolution will improve even more, by a factor of approximately 100/p_T(GeV/c).
In Run II the new muon A-layer counters will permit muon identification and momentum reconstruction down to p_T approx 1.5, GeV/c. This will extend acceptance for soft muon b tags well below the present approx 4 GeV/c. Triggering on all top final states containing such tags (dilepton, lepton+jets, and all jets) will be enhanced by the new ability in Run II to trigger on muons above 6--10 GeV over most of the eta acceptance of the calorimeter. This will become possible owing to the new dedicated muon trigger counters.
The new preshower detector, combined with the E/p match afforded by magnetic tracking, will allow DO in Run II to carry out b tagging in the electron channel as well. Preliminary studies indicate that the fine segmentation of the EM calorimeter permits soft electrons from b-decays to be found with good efficiency (simge 50%). The preshower and E/p match will serve to reduce the presently large backgrounds.
We have studied b jet tagging in t tbar events using a simple signed impact parameter technique. First we generate 638 t tbar (t -> b e nu) events (m_t=160 GeV) imposing no restrictions on the composition of the b jets. As usual these events were fully GEANT simulated and pattern recognized although, conservatively, only the barrel silicon and fiber detector hits were used. Each reconstructed track was then associated with one ISAJET jet based on a cone cut of 0.45 in eta-phi space.
The distributions of signed impact parameter delta and of its significance for all tracks associated with b jets in both tbar t events and QCD jet events were studied. The distribution of delta for tracks associated with b's is found to be comparable in these two samples. Likewise, the distribution of delta for tracks in non-b jets in the top events (from the W from t decay) was found to be comparable to that for tracks with zero decay lifetime in QCD dijets. This assures us that our Run II tracking resolutions will not be significantly degraded in busier events with larger total p_T.
Using the W+jets simulation program VECBOS, with supplementary QCD evolution and hadron fragmentation performed by ISAJET, we also produced a 100 event background sample. Thereafter both the top and W+jets events were simulated and reconstructed identically. To obtain a first estimate of performance, we considered a jet to be b lifetime tagged if at least three of its tracks possessed a (positive) impact parameter significance of at least 3sigma. By itself, this requirement preserved more than 50% of the top signal events, while eliminating all but 2+-1.4% of the W+jets background. As shown here, further requiring the minimum tagged jet E_T to exceed 30 GeV reduced the background to 1+-1% while retaining 46% of the original top signal. We emphasize that these performance figures are a conservative first estimate made with a very simple algorithm. No cuts, for example on the longitudinal event vertex position, were made especially for this study.
The following assumes an integrated luminosity to tape of 2 fb^-1 delivered to the upgraded D0 detector. We consider a center of mass energy limited to the present sqrt{s}=1.8 TeV (though raising it to 2.0 TeV would increase the yields by 30%). We normalize to the measured background subtracted top event rate, which corresponds to a cross section for 200 GeV top of 6.4 pb. For simplicity, where appropriate we apply the same ``standard'' cuts recently used for our initial top quark observation.
Further improvements in Run II top analysis will include a factor of two reduction in dilepton background from the electron sign selection and the more precise rejection of Z -> mumu, and an increase in soft mu tagging efficiency from the lower muon p_T threshold.
Together the Run II improvements will make possible an increase from approx 0.3/pb^-1 to approx 0.5/pb^-1 in the yield of background subtracted top events passing final cuts, without any corresponding background increase. With an integrated luminosity of 2 fb^-1, we expect a final sample of approx 1000 background subtracted top candidates with signal to background ge5:1. Either the yield or the S:B, and probably both, can be enhanced by further optimization of the cuts and selection algorithms. With these statistics, the error on the measured top cross section clearly will be dominated by systematics.
A W mass precision approaching approx 35 MeV may be achieved shortly after the end of the decade from the Tevatron and LEP200 experiments. Our goal is to measure the top mass in D0 to a precision of 5 GeV, or 2.5%. The sensitivity of the Higgs mass to m_t and m_W is such that an error on the top mass much larger than this will dominate the uncertainty on m_H, while smaller errors on m_t do not significantly improve the precision with which m_H is constrained. Our 5 GeV goal corresponds to a Higgs mass uncertainty of delta m_H/m_H ~ 0.8.
When simulated Monte Carlo top to lepton+jets events are subjected to a conventional 2C fit, the Tevatron collider experiments obtain a mass lineshape which has a approx 15% rms width, and they observe a flattening of the response function <m_t>/m_gen approx 0.6, where <m_t> is the mean fit mass and m_gen is the generated top mass. (This flattening arises from the effects of gluon radiation and jet misassignment.) If background is neglected, from these elementary considerations one expects a random error in the fit mass of roughly 15%/0.6sqrt{N} = 25%/sqrt{N} for an N event signal.
At present we quote a 10% random error on m_t based on fitting 24 ``loose cut'' candidates divided equally between signal and background. This gives 35%/sqrt{N}; the inflation arises mainly from the effects of background in the fit. %(After fitting N=12 candidates %plus 7 background events, CDF's random error is 4.5%, or 16%/sqrt{N}). Even this conservative estimate results in a random error of only 1.3% with the full Run II top statistics. Clearly, then, the Run II top mass precision will be dominated by systematic errors. At present in DO, foremost among those is the energy scale error of 10%. We have already exhibited a 2.3sigma mass peak in the m_t-m_W plane; in Run II we plan to use a similar W mass peak to fix the hadronic calorimeter energy scale for top mass reconstruction. We expect less than a 2% statistical error on this method of calibration.
Among the residual systematic errors, perhaps the least reducible is the difference between the actual pattern of gluon radiation and that predicted by the available parton shower Monte Carlo calculations. At present we observe >2% differences in the means of distributions of fit mass for HERWIG and ISAJET top samples; the peaks of these distributions, to which fits can be especially sensitive, exhibit still larger shifts. Uncertainties associated with radiative corrections will be more severe at the LHC; even after experiments begin collecting data there, the facility of choice for precise top mass measurement may continue to be the Tevatron.
D0 possesses a unique advantage for controlling errors due to radiative corrections. This is its fine calorimeter granularity, surpassing that of any existing or planned hadron collider detector. We plan to use this detailed calorimeter information to measure the initial and final state radiation and to recombine the latter. Our techniques in this area are still primitive. However, we are already able to demonstrate clustering algorithms able to detect, on average, >2 extra gluons per top event, and mass fitting algorithms able to identify and recombine these gluon jets. Relative to the conventional practice of fitting only the four leading jets, these algorithms improve the mass resolution and, more important, reduce the systematic uncertainties associated with these radiative corrections. Our detailed GEANT simulation of the calorimeter will also contribute to reducing these uncertainties, particularly as we gain more experience in benchmarking the simulations against real data.
With approximately 1000 background-subtracted tbar t candidates, it will be possible to test the standard model predictions of the decay properties of the top. The electronic and muonic branching ratios will be measured to 15% each, providing a key check on the Standard Model. The rate of double vs. single b lifetime tags should determine the t -> b branching fraction to 5%, and thus the CKM matrix element V_tb to less than 3%. By fitting the angular distribution of the lepton from decay of W from top, it should be possible to measure the approx 70% branching ratio of top to longitudinally polarized W_L with a 3-5% error, thus limiting the V+A coupling at the W vertex.
Single t production has been studied using a fast simulation which includes parameterized D0 detector resolutions and assumes a b-tagging efficiency of 50% per jet and a lepton reconstruction efficiency of 75%. (These are the current Run I efficiencies; the improvements for Run II have not been taken into account here). The expected dominant process for single top quark production in pbar p collisions is via W t-channel production, e.g. u bbar -> d tbar with the bbar supplied by the antiproton sea. For a 200 GeV top quark, this process accounts for three quarters of the 1.7 pb single top cross section, yielding 3400 events in 2 fb^-1 at sqrt{s=1.8 GeV. Of these, approx 750 are tagged by e or mu semileptonic decay. After efficiencies and cuts, about 400 events are expected in a 100 GeV band centered at the top mass above a approx 300 event background. The dominant background processes are pbar p -> Wbbar b and pbar p -> tbar t. The distribution of reconstructed top mass for signal and backgrounds, after cuts, is shown here.
With 1 fb^-1 of data it will be possible to measure the single top cross section with a precision of about 10%. The full width of the top is not measurable in the t tbar state, but can be extracted directly (within theoretical uncertainties) from the single top cross section; a approx 20% top width measurement should be feasible. Single top events will also provide an interesting cross-check of the top mass, since the multi-jet combinatoric problems are much less severe here.
The present limit on the mass of the charged Higgs boson predicted by minimal supersymmetry (m_H^+- > 45 GeV/c^2) is from LEP measurements. By the time the DO Upgrade run is underway, we expect this limit to increase to simge 100 GeV/c^2, the exact value depending on the maximum energy reached by LEP 200. Assuming that the charged Higgs is lighter than the top quark, one can search for top quark decays into a charged Higgs and a b quark in addition to its usual decay into W and a b quark. The charged Higgs will have two predominant decay modes, into cbar s and into tau nu, the relative branching ratios depending on the tan(beta) parameter of the minimal supersymmetric model. The Higgs decay will not contribute to the DO dilepton top search channels of ee, emu and mumu except by the secondary decay of the tau, which we neglect in the present calculation. The lepton + jets channels e + jets and mu + jets will have contributions from both W decays and from charged Higgs decays.
There are several ways of searching for the charged Higgs in top quark decays.
The top quark's large mass makes it unique among the elementary fermions in having a coupling to the SM Higgs of order unity; it is the only fermion not to be approximately massless on the scale of electroweak symmetry breaking. It therefore offers a dramatic example of broken flavor symmetry and a window into possible physics beyond the standard model --- for example, to scenarios where electroweak symmetry is dynamically broken at a scale of order the top quark mass. Examples of such models are multiscale technicolor [2] color octet vector mesons V_8 associated with tbar t condensation [3] and electroweak isoscalar quarks [4].
The cross section, tbar t invariant mass, and centre-of-mass angular distributions may all be used to search for new physics.
The tbar t cross section is enhanced by most of these models. For example, in multiscale technicolor, a color-octet technipion eta_T which occurs with mass 400--500 GeV can easily double the tbar t rate. If the eta_T -> tbar t partial width is not too large, (e.g. the coupling C_t ~ 1/3) then a peak at the eta_T mass will easily be seen in the tbar t invariant mass distribution (with a natural FWHM of about 20 GeV) with 1 fb^-1 of integrated luminosity. If C_t ~ 1 then the eta_T -> tbar t peak is much broader, and the mass peak is too wide for direct observation; the enhancement of the tbar t cross section will still be obvious.
Subsystem invariant masses may also be sensitive to new physics. For example, a color-octet technirho rho_T may be produced and decay via rho_T -> W pi^+-_T, with pi^+-_T -> t bar b -> W b bar b, the same final state as in tbar t production. Searches for processes such as this will be possible with 1 fb^-1 of data.
Another sensitive distribution is the centre-of-mass angle of the top quarks. The Tevatron has a distinct advantage in this case, as the annihilating quark and antiquark directions are strongly correlated with those of the proton and antiproton. Measurements of forward-backward asymmetry (A_FB) are therefore much cleaner than at the LHC, where tbar t production is dominated by gg fusion. It is found [5] that sufficient to distinguish, at the 5sigma (statistical) level, the A_FB arising from V_8-enhanced top production from the standard QCD mechanisms.