Search for the standard model Higgs boson in the H to ZZ to ll tau tau decay channel in pp collisions at sqrt(s)=7 TeV

A search is reported for the standard model Higgs boson in the decay mode H to ZZ to tau plus lepton pairs, where the leptons are either electrons or muons, in proton-proton collisions at sqrt(s)=7 TeV, corresponding to an integrated luminosity of 4.7 inverse femtobarn collected with the CMS detector at the LHC. No evidence is found for a significant deviation from the background expectation. An upper limit four to twelve times larger than the predicted value is set at 95% confidence level for the product of the standard model Higgs boson production cross section and decay branching fraction in the mass range 190


Introduction
The inner tracker measures charged particle tracks within the range |η| < 2.5. It consists of 1 440 silicon pixel and 15 148 silicon strip detector modules, and provides an impact parameter resolution of ∼15 µm and a transverse momentum resolution of about 1.5% for 100 GeV particles. The reconstructed tracks are used to measure the location of interaction vertices. The spatial resolution of the reconstruction is ≈25 µm for vertices with more than 30 associated tracks [39].
The muon barrel region is covered by drift tubes, and the endcap regions by cathode strip chambers. In both regions, resistive plate chambers provide additional coordinate and timing information. Muons are reconstructed in the range |η| < 2.4, with a typical p T resolution of ≈1% for p T ≈ 40 GeV.

Event Selection and Monte Carlo Samples
At the trigger level, the selected events are required to have at least two muons, one with p T > 13 GeV (p T > 17 GeV for the end of the data-taking period when the instantaneous luminosity was highest) and the other with p T > 8 GeV, or at least two electrons, one with p T > 17 GeV and the other with p T > 8 GeV.
Algorithms for identifying muons and electrons, collectively referred to as leptons, are based on the tracker, the muon systems and the calorimeters [40,41]. Since the ZZ final state is expected to have only a small contribution from background processes, the algorithms are tuned to maximize the lepton-reconstruction efficiency, resulting in an increased lepton-misidentification rate. A particle flow (PF) technique [42] is used to form lepton-isolation quantities and is also used for τ h reconstruction. In the PF approach, information from all subdetectors is combined to reconstruct and identify particles produced in the collision. The particles are classified into mutually exclusive categories: charged hadrons, photons, neutral hadrons, muons, and electrons. These particles are used to reconstruct τ h with the "hadron plus strip" (HPS) algorithm [43] that is designed to optimize the performance of τ h identification and reconstruction by considering specific τ h decay modes. The neutrinos produced in all τ decays escape detection and are ignored in the τ h reconstruction. The algorithm provides high τ h identification efficiency, approximately 50% for the range of τ h energies relevant for this analysis, while keeping the misidentification rate for jets at the level of ≈1%, that is factor of three to four times lower with respect to other available algorithms [44].
Events are required to have at least one Z → + − candidate, denoted by Z 1 , with the leptons of opposite charge, one with p T > 20 GeV and another with p T > 10 GeV, and with |η| < 2.4 for the muons and |η| < 2.5 for the electrons. Both leptons are required to have a combined PF relative isolation I PF rel < 0.25, which is defined as: where p charged T is the scalar sum of the charged hadrons p T , and E γ T and E neutral T correspond, respectively, to the sum of the transverse energies of the photons and neutral hadrons, all measured in the isolation cone of ∆R < 0.4 around the lepton direction, where ∆R = (∆η) 2 + (∆φ) 2 . The contribution from neutrals is corrected for the effect of overlapping pp interactions. The value of the correction is estimated by scaling the sum of the p T of all charged particles not associated with the vertex in the isolation cone (p PU T ) by a factor 0.5, which is approximately the ratio of neutral to charged hadron yields in hadronization processes.
For the second Z, denoted by Z 2 , the selection requirements depend on the final state. If the final state is τ µ τ e , the lepton p T values are required to exceed 10 GeV. The remaining criteria are identical to those for Z 1 . Since τ h s have much larger misidentification rates than the other leptons, the isolation requirement based on I PF rel for the muons and electrons in the final states τ µ τ h and τ e τ h is changed to 0.15 and 0.1, respectively. In a study of inclusive Z → ττ production [45], it was demonstrated that modifying the muon and electron isolation requirements is a more effective way to reduce background in such final states than requiring tighter isolation on τ h . The τ h are required to have p T > 20 GeV and |η| < 2.3, and to pass a loose HPS workingpoint requirement. If the Z 2 decays to τ + h τ − h , both τ h are required to pass a medium working point of the HPS algorithm. The loose (medium) working point requires the scalar sum over the charged hadrons p T and the neutral hadrons E T in the isolation cone, to be less than 2 GeV (1 GeV). The loose (medium) working point corresponds to a probability of approximately 1% (0.5%) for jets to be misidentified as τ h . Using the medium instead of loose working point leads to a decrease in the τ h reconstruction efficiency from ≈50% to ≈40%.
The visible invariant mass of the reconstructed Z 2 → τ + τ − is required to be 30 < m ττ < 80 GeV, and this criterion is used for most of the final states. The upper bound reduces contributions from Z 2 → , where a muon or an electron is not well reconstructed, and misidentified as a τ h . For the Z 2 → τ e τ µ final state, the upper bound is increased to 90 GeV, as this state is not produced in Z 2 → decays. Leptons from the same Z are required to be separated by ∆R > 0.4 for Z 1 , and by ∆R > 0.5 for Z 2 . The two reconstructed Z 1 and Z 2 are required to be separated by ∆R > 0.5.
A set of Monte Carlo (MC) event samples is used to simulate signal and background events. The Drell-Yan background, + − in association with jets, is simulated with the next-to-leading order (NLO) MC generator POWHEG 2.0 [46][47][48]. The QCD multijet, W and diboson WZ backgrounds are simulated with PYTHIA 6.424 [49]. The ZZ background is simulated with PYTHIA 6.424 and MADGRAPH [50]. The tt samples are simulated with MADGRAPH. The τ-lepton decays are generated with TAUOLA [51]. The Higgs boson samples are generated using POWHEG 2.0, which incorporates NLO gluon fusion (gg → H) and vector-boson fusion (qq → qqH). All events are processed through a detailed simulation of the CMS detector based on GEANT4 [52] and reconstructed with the same algorithms that are used for data.

Background Estimates and Systematic Uncertainties
The major irreducible source of background to the H → ZZ → + − τ + τ − process is from SM ZZ → + − τ + τ − production. The ZZ contribution is estimated from data by scaling the prediction from simulation to the well measured inclusive Z production cross section. The number of estimated ZZ events, N est ZZ , can be written as: where N obs Z is the number of observed events from inclusive Z production, A Z is their estimated acceptance from a MC simulation, including all selection requirements, and rescaled by measured data/MC correction factors, A ZZ is the acceptance for ZZ events, σ SM Z is the SM cross section for inclusive Z production, and σ SM ZZ is the SM cross section for ZZ production calculated with MCFM [53].
The other major background contributions arise from the production of Z and WZ in association with jets, as well as tt and QCD multijet production. The latter two backgrounds are small. In all these cases, a jet or non-isolated lepton is misidentified as a τ h , τ e or τ µ . The probability

Results
for jets to be misidentified as τ h is measured using + − τ h τ h events in data in which the Z 1 passes all selection requirements, but no requirement is applied on τ h isolation, and the two τ h candidates are required to have the same charge. This region is dominated by Z+jets events. The τ h misidentification rate is defined as the ratio of the number of τ h candidates that pass the HPS loose or medium working-point requirements, to the initial number of τ h candidates, and is measured as a function of the p T for each τ h . To estimate the number of background events in the signal region, the measured misidentification rate is applied to events that pass all selection requirements, including the opposite-charge requirement for the Z 2 , but requiring the τ h candidates to not be isolated.
The misidentification rate for τ e and τ µ in the µµτ µ τ e and eeτ µ τ e final states is estimated using events in which the Z 1 passes all selection requirements, and the event contains an additional muon or electron. No isolation requirements are applied to it. The misidentification rate is defined for τ e and τ µ in the same way as described above for τ h and applied to µµτ µ τ e and eeτ µ τ e events that pass all the selection requirements, but requiring τ e or τ µ to not be isolated. Isolated muons and electrons from H → ZZ → 4 and ZZ → 4 production can also be misidentified as τ h . Events are rejected if they are also identified as ZZ → 4 events with criteria described in Ref. [16].
Theoretical uncertainties on the Higgs boson cross section (17-20%) and branching ratio (2%) are taken from Ref. [17]. Recent studies [17, 54,55] show that current MC simulations do not describe the correct Higgs boson mass line shape above ≈300 GeV. This effect amounts to an additional uncertainty on the theoretical cross section, and hence on the limits, of about 4% at m H = 300 GeV and 10-30% for m H of 400-600 GeV. The main uncertainty on the estimate of the ZZ background arises from the theoretical uncertainty on the ZZ production cross section and is taken from Ref. [56]. The uncertainties on the other backgrounds, Z+jets, WZ+jets, and tt reflect the uncertainties on the measured values of the misidentication rates and the limited statistics of the control regions in the data. The uncertainty on integrated luminosity of the data sample is 4.5% [57]. Systematic uncertainties on trigger efficiency (1%) and on lepton identification efficiency and isolation are evaluated from data. The uncertainties associated with lepton identification and isolation are 1-2% for muons and electrons, and 6-7% for τ h . Uncertainties on energy scales, 3% for τ h and 1-2.5% for electrons, contribute to variation in the shape of the mass spectrum.

Results
Ten + − τ + τ − candidates are observed in eight search channels, while 11.60 ± 0.54 (stat.) ± 1.62 (syst.) background events are expected. Table 1 compares the estimated number of background events to the number of events observed in the signal region. The distribution of the reconstructed invariant mass summed over all eight + − τ + τ − decay channels is shown in Fig. 1. The shape of the background is taken from the MC simulation, with each component normalized to the corresponding estimated value from Table 1. The expected mass distributions for the SM Higgs boson with a mass of m H = 200 GeV and 400 GeV are also shown in Fig. 1. The reconstructed masses are shifted with respect to the generated values by ≈ 30% due to the undetected neutrinos in τ decays. As a result, the H → ZZ → + − τ + τ − mass resolution is 10-15%, depending on the final state, and is almost independent of m H .
The product of the acceptance and branching fraction for the individual τ-decay channels ranges between 0.01-0.02 for m H = 200 GeV, and increases by a factor of three to four for m H = 400-450 GeV. This behaviour is expected. The final-state leptons produced in τ decays of more massive Higgs bosons have higher momenta than those from direct Z → production Table 1: The estimated yields of ZZ and other background events obtained from data, as described in the text, are shown for each decay channel and are summed in the total background yield ("Total backgr."), and compared to the number of events observed in the signal region. The total uncertainty is the sum in quadrature of statistical and systematic uncertainties. The number of signal events expected for the SM   In Fig. 2 the expected and observed upper limits at 95% CL on the product of the Higgs boson production cross section and decay branching fraction normalized to the SM expectation are presented as a function of m H . The limits are calculated with the modified frequentist construction CL s [58][59][60] based on the shape of the + − τ + τ − invariant mass distributions by including all eight individual channels in the likelihood combination. The green and yellow bands represent the one-and two-standard-deviation variations from the expected limit. The systematic uncertainties are introduced in the form of nuisance parameters with log-normal probability density functions. The upper limit on the cross section is approximately a factor four to twelve larger than the SM Higgs boson production cross section in the range of 190 < m H < 600 GeV.

Summary
A search for the standard model Higgs boson has been performed in the decay mode H → ZZ → + − τ + τ − using CMS data corresponding to an integrated luminosity of 4.7 fb −1 . No evidence is found for a significant deviation from the background expectation. An upper limit four to twelve times larger than the predicted value is set at 95% confidence level for the product of the standard model Higgs boson production cross section and decay branching fraction in the mass range 190 < m H < 600 GeV. This is the first Higgs boson search performed in the H → ZZ → + − τ + τ − channel. [40] CMS Collaboration, "Performance of muon identification in pp collisions at √ s = 7 TeV", CMS Physics Analysis Summary CMS-PAS-MUO-10-002, (2010).