Physics Letters B 718 (2012) 369–390 Contents lists available at SciVerse ScienceDirect Physics Letters B www.elsevier.com/locate/physletb Search for the Standard Model Higgs boson produced in association with a vector boson and decaying to a b-quark pair with the ATLAS detector✩ .ATLAS Collaboration  a r t i c l e i n f o a b s t r a c t Article history: Received 1 July 2012 Received in revised form 14 September 2012 Accepted 23 October 2012 Available online 26 October 2012 Editor: H. Weerts Keywords: Standard Model Higgs boson ATLAS LHC This Letter presents the results of a direct search with the ATLAS detector at the LHC for a Standard Model Higgs boson of mass 110mH  130 GeV produced in association with a W or Z boson and decaying to b ¯b. Three decay channels are considered: ZH → +−b ¯b, WH → νb ¯b and ZH → ν ¯νb ¯b, where  corresponds to an electron or a muon. No evidence for Higgs boson production is observed in a dataset of 7 TeV pp collisions corresponding to 4.7 fb−1 of integrated luminosity collected by ATLAS in 2011. Exclusion limits on Higgs boson production, at the 95% confidence level, of 2.5 to 5.5 times the Standard Model cross section are obtained in the mass range 110–130 GeV. The expected exclusion limits range between 2.5 and 4.9 for the same mass interval. © 2012 CERN. Published by Elsevier B.V. 1. on (L tio w fe co be tio fid th co by th th du st bo ad su vi ki m H ✩  0 ht Open access under CC BY license.Introduction The search for the Standard Model (SM) Higgs boson [1–3] is e of the most important endeavours of the Large Hadron Collider HC). The H → b ¯b decay corresponds to the highest branching ra- for a low-mass Higgs boson in the SM. Observing this decay ould provide direct sensitivity to the Higgs boson coupling to rmions. The results of searches in various channels using data rresponding to an integrated luminosity of up to 4.9 fb−1 have en reported recently by both the ATLAS and CMS collabora- ns [4,5]. The Higgs boson has been excluded at the 95% con- ence level (CL) below 114.4 GeV by the LEP experiments [6], in e regions 100–106 GeV and 147–179 GeV at the Tevatron p ¯p llider [7], and in the regions 112.9–115.5 GeV and 127–600 GeV the LHC experiments [4,5]. This Letter reports on a search for e SM Higgs boson performed for the H → b ¯b decay mode, over e mass range 110–130 GeV where this decay mode dominates. Due to the large backgrounds present in the dominant pro- ction process gg → H → b ¯b, the analysis reported here is re- ricted to Higgs boson production in association with a vector son, WH and ZH [8–12], where the vector boson provides an ditional final state signature, allowing for significant background ppression. An additional handle against the backgrounds is pro- ded by exploiting the better signal-over-background level of the nematic regions where the weak bosons have high transverse omenta [13]. These channels are also important contributors to iggs boson searches at CMS [14] and the Tevatron [7]. © CERN for the benefit of the ATLAS Collaboration. This Letter presents searches in the ZH → +−b ¯b, WH → νb ¯b and ZH → ν ¯νb ¯b channels, where  is either an electron or a muon, including electrons and muons from tau lepton decays. The data used were recorded by the ATLAS experiment during the 2011 LHC run at a centre-of-mass energy of √ s = 7 TeV and correspond to integrated luminosities of 4.6 to 4.7 fb−1 [15,16], depending on the analysis channel. The leptonic decay modes of the weak bosons are selected to suppress backgrounds containing only jets in the fi- nal state. In the ZH → ν ¯νb ¯b channel, the multijet background is suppressed by requiring a large missing transverse energy. 2. The ATLAS detector The ATLAS detector [17] consists of four main subsystems. An inner tracking detector is immersed in the 2 T magnetic field produced by a superconducting solenoid. Charged particle posi- tion and momentum measurements are made by silicon detec- tors in the pseudorapidity1 range |η| < 2.5 and by a straw tube tracker in the range |η| < 2.0. Calorimeters cover |η| < 4.9 with a variety of detector technologies. The liquid-argon electromag- netic calorimeter is divided into barrel (|η| < 1.475) and end- cap (1.375 < |η| < 3.2) sections. The hadronic calorimeters (using 1 ATLAS uses a right-handed coordinate system with its origin at the nominal in- teraction point (IP) in the centre of the detector and the z-axis coinciding with the axis of the beam pipe. The x-axis points from the IP to the centre of the LHC ring, and the y-axis points upward. Cylindrical coordinates (r, φ) are used in the trans- verse plane, φ being the azimuthal angle around the beam pipe. The pseudorapidity is defined in terms of the polar angle θ as η = − ln tan(θ/2). For the purpose of the fiducial selection, this is calculated relative to the geometric centre of the detector;E-mail address: atlas.publications@cern.ch. otherwise, it is relative to the reconstructed primary vertex of each event. 370-2693 © 2012 CERN. Published by Elsevier B.V. tp://dx.doi.org/10.1016/j.physletb.2012.10.061 Open access under CC BY license. 370 ATLAS Collaboration / Physics Letters B 718 (2012) 369–390 liquid argon or scintillating tiles as active materials) surround the electromagnetic calorimeter and cover |η| < 4.9. The muon spec- trometer measures the deflection of muon tracks in the field of three large air-core toroidal magnets, each containing eight super- conducting coils. It is instrumented with separate trigger chambers (covering |η| < 2.4) and high-precision tracking chambers (cover- ing |η| < 2.7). 3. Data and Monte Carlo samples The collision data used in this analysis are selected such that all elements of the ATLAS detector were delivering high-quality data. In the ZH → +−b ¯b and the WH → νb ¯b analyses, events were primarily collected using single-lepton triggers with a transverse momentum (pT) threshold of 20 GeV for electrons, which was raised to 22 GeV as the instantaneous luminosity increased, and 18 GeV for muons. In the ZH → +−b ¯b analysis, these triggers were supplemented with a di-electron trigger with a threshold of 12 GeV. The lepton trigger efficiency is measured using a sample of Z → +− events. The resulting efficiencies, relative to the offline selection, are close to 100% for ZH → e+e−b ¯b and WH → eνb ¯b. The efficiencies are around 95% for the ZH → μ+μ−b ¯b chan- nel and 90% for the WH → μνb ¯b channel, due to the lower angular coverage of the muon trigger chambers with respect to the precision tracking chambers. The missing transverse energy (EmissT ) trigger used for the ZH → ν ¯νb ¯b channel has a threshold of 70 GeV and an efficiency above 50% for EmissT above 120 GeV. This efficiency exceeds 99% for EmissT above 170 GeV. The effi- ciency curve is measured in a sample of W → μν + jet events collected using muon triggers, which do not rely on the presence of EmissT . The Monte Carlo (MC) simulation predicts the trigger ef- ficiency to be 5% higher than that observed in collision data for 120 GeV EmissT < 160 GeV and agrees for E miss T  160 GeV. A cor- rection factor of 0.95 ± 0.01 is therefore applied to the MC in the lower EmissT region, and no trigger efficiency correction is applied elsewhere. Due to practical constraints, several MC generators were used to simulate signal and background processes. The WH and ZH signal processes are modelled using MC events produced by the Pythia [18] event generator, interfaced with the MRST modified leading-order (LO*) [19] parton distribution functions (PDFs), us- ing the AUET2B tune [20] for the parton shower, hadronization and multiple parton interactions. The total cross sections for these channels, as well as their corresponding uncertainties, are taken from the LHC Higgs Cross Section Working Group report [21]. Dif- ferential next-to-leading order (NLO) electroweak corrections as a function of the W or Z transverse momentum have also been applied [22,12]. The Higgs boson decay branching ratios are cal- culated with Hdecay [23]. The background processes are modelled with several different event generators. The Powheg [24–26] generator, in combination with MSTW 2008 NLO PDFs [27] and interfaced with the Pythia program for the parton shower and hadronization, is used to sim- ulate W+  1b jet events. The Sherpa generator [28] is used to simulate Z+  1b jet and Z+  1c jet events. The Alpgen gen- erator [29] interfaced with the Herwig program [30] is used to simulate W+  1c jet, W+  1 light jet (i.e. not a c or b jet) and Z+  1 light jet events. The above background simulations include γ ∗ production and Z/γ ∗ interference where appropriate. The MC@NLO generator [31], using CT10 NLO PDFs [32] and inter- faced to Herwig, is used for the production of top-quarks (single- top and top-quark pair production). The Herwig generator, is used to simulate the diboson (Z Z , W Z and WW ) samples. The Her- wig generator uses the AUET2 tune [33] for the parton shower and hadronization model, relies on MRST LO* PDFs (except for top production) and is in all cases interfaced to Jimmy [34] for the modelling of multiple parton interactions. The diboson cross sec- tions normalized to NLO QCD computations [35,36]. MC samples are passed through the full ATLAS detector simulation [37] based on the Geant4 [38] program. 4. Reconstruction and identification of physics objects Events are required to have at least one reconstructed primary vertex with three or more associated tracks with pT > 0.4 GeV in the inner detector. If more than one vertex is reconstructed, the primary vertex is chosen as the one with the highest sum of the squares of the transverse momenta of all its associated tracks. Electron candidates are reconstructed from energy clusters in the electromagnetic calorimeter and are required to pass identifi- cation criteria based on the shower shape. Central electrons must have a matching track in the inner detector that is consistent with originating from the primary vertex and requirements are placed on track quality and track-cluster matching [39]. Further track and cluster related identification criteria are applied to electron candi- dates in order to reduce background from jets being misidentified as electrons. The criteria are tighter for W decays, where the back- ground is larger. Muons are found offline by searching for tracks reconstructed in the muon spectrometer with |η| < 2.7. The charged leptons that are used to reconstruct the vector bo- son candidate are required to satisfy pT > 20 GeV in the ZH → +−b ¯b channel, while this cut is increased to pT > 25 GeV in the WH → νb ¯b channel in order to be above the trigger threshold, and maintain a high trigger efficiency. In both cases, the leptons must be central (|η| < 2.47 for electrons and |η| < 2.5 for muons) and have a matching track in the inner detector (with a coverage up to |η| < 2.5) that is consistent with originating from the pri- mary vertex. In order to suppress background from semileptonic heavy- flavour hadron decays, the leptons are required to be isolated. In the ZH → +−b ¯b and WH → νb ¯b channels the sum of the transverse momenta of all charged tracks (other than those of the charged leptons) reconstructed in the inner detector within a cone of R = √ ( η)2 + ( φ)2 < 0.2 from each charged lepton is re- quired to be less than 10% of the transverse momentum of the lepton itself. In the WH → νb ¯b channel, the isolation require- ment is strengthened by requiring in addition that the sum of all transverse energy deposits in the calorimeter within a cone of R < 0.3 from the charged lepton be less than 14% of the trans- verse energy of the lepton itself. In order to suppress the top-quark background in the ZH → ν ¯νb ¯b channel, events containing electrons with |η| < 2.47 and pT > 10 GeV, or muons with |η| < 2.7 and pT > 10 GeV are re- moved. Similar requirements are applied on any additional lepton reconstructed in the WH → νb ¯b channel, but the minimum lep- ton pT is increased to 20 GeV if the additional lepton has the same charge as, or a different flavour than the signal lepton. Events with forward electrons [39] (2.47 < |η| < 4.5) with pT > 20 GeV are also removed in the WH → νb ¯b channel. Jets are reconstructed from energy clusters in the calorime- ter using the anti-kt algorithm [40] with a radius parameter of 0.4. Jet energies are calibrated using pT- and η-dependent correc- tion factors based on MC simulation and validated with data [41]. A further correction is applied when calculating the di-jet invari- ant mass, as described in Section 5 below. The contribution from jets originating from other collisions in the same bunch crossing is reduced by requiring that at least 75% of the summed trans- verse momentum of inner detector tracks (with pT > 0.4 GeV) associated with the jet are compatible with originating from the ATLAS Collaboration / Physics Letters B 718 (2012) 369–390 371 primary vertex. Furthermore, a jet is required to have no identi- fied electron within R  0.4. Only jets with pT > 25 GeV and within the acceptance of the inner detector (|η| < 2.5) are used to reconstruct Higgs boson candidates. Events containing additional jets are rejected in the WH → νb ¯b analysis, to suppress back- grounds characterized by additional hadronic activity. To do this, jets are counted using the following criteria: pT > 20 GeV and |η| < 4.5. Jets which originate from b quarks can be distinguished from other jets by the relatively long lifetime of hadrons containing b quarks. Such jets are primarily identified (“b-tagged”) by recon- structing one or more secondary decay vertices from tracks within the jet, using either an inclusive vertex reconstruction algorithm or a cascade b → c-hadron decay chain vertex fit, or by combining the distances of closest approach to the primary event vertex (impact parameters) of tracks in the jet [42–45]. The information from the vertex and impact parameter based algorithms is combined into a single discriminant w by using an artificial neural network, which is trained based on a set of samples of simulated events, such that a jet with higher w is more likely to be a b jet. A selection cut on w is applied, resulting in an efficiency of about 70% for identi- fying true b jets, of about 20% for c jets and about 0.8% for light jets, as evaluated in simulated t¯t events. The b-tagging efficiency and rejection factors in the simulation are corrected to the respec- tive measurements in data by the use of appropriate scale factors. These correspond to corrections of around 5 to 15% for b jets, 20% for c jets, and around 50% for light jets. The EmissT magnitude and direction are measured from the vec- tor sum of the transverse momentum vectors associated with clusters of energy reconstructed in the calorimeters with |η| < 4.9 [46]. A correction is applied to the energy of those clusters that are associated with a reconstructed physical object (jet, electron, τ -lepton, photon). Reconstructed muons are also included in the sum, and any calorimeter energy deposits associated with them are excluded. To supplement the calorimeter-based definition of EmissT in the ZH → ν ¯νb ¯b channel, the track-based missing trans- verse momentum, pmissT , is calculated from the vector sum of the transverse momenta of inner detector tracks associated with the primary vertex [47]. 5. Event selection Events in the ZH → +−b ¯b channel are required to contain exactly two same-flavour leptons. The two leptons must be op- positely charged in the case of muons. This is not required for electrons since energy losses from showering in material in the inner detector lead to a higher charge misidentification probabil- ity. The invariant mass of the lepton pair must be in the range 83 GeV 25 GeV. A requirement on the transverse mass2 of mT > 40 GeV is imposed to suppress the mul- tijet background. The ZH → ν ¯νb ¯b selection requires EmissT > 120 GeV. Require- ments of pmissT > 30 GeV and on the difference in azimuthal angle between the directions of EmissT and p miss T , φ(E miss T , p miss T ) < π/2, are imposed to suppress events with poorly measured EmissT . These help to suppress the multijet background, which is dominated by 2 The transverse mass (mT) is defined from the transverse momenta and the azimuthal angles of the charged lepton (pT and φ ) and neutrino (pνT and φ ν ): mT = √ 2pTp ν T (1− cos(φ  − φν)), where pνT = E miss T . one or more jets being mismeasured by the calorimeter. A cut on the difference in azimuthal angle between EmissT and the nearest jet min( φ(EmissT , jet)) > 1.8 is applied to further reduce the mul- tijet background. The transverse momentum of the vector boson, pVT , is recon- structed from the two leptons in the ZH → +−b ¯b channel, from the lepton and EmissT in the WH → νb ¯b channel and from E miss T in the ZH → ν ¯νb ¯b channel. Events in all channels are required to contain exactly two b- tagged jets, of which one must have pT > 45 GeV and the other pT > 25 GeV. If pVT is less than 200 GeV the two b-tagged jets are required to have a separation of R > 0.7, to reduce W + jet and Z + jet backgrounds. Additionally, in the ZH → ν ¯νb ¯b chan- nel a cut on the separation between the two jets of R < 2.0 ( R < 1.7) for pVT < 160 GeV (p V T > 160 GeV) is applied to reduce the multijet background. Events in the ZH → +−b ¯b channel may contain additional non-b-tagged jets, while in the WH → νb ¯b and ZH → ν ¯νb ¯b channels, events with additional jets are rejected to further suppress top-quark background. In the WH → νb ¯b analysis, where the top-quark background is dominant, events con- taining additional jets with |η| < 4.5 and pT > 20 GeV are rejected, while in the ZH → ν ¯νb ¯b channel the selection is restricted to jets with |η| < 2.5 and pT > 25 GeV. In the ZH → ν ¯νb ¯b analysis, further cuts are applied on the azimuthal angle between EmissT and the reconstructed transverse momentum of the b ¯b system, φ(b ¯b, EmissT ), to further reject mul- tijet background. The ZH → ν ¯νb ¯b signal, where the Higgs and Z bosons recoil against each other, is characterized by large values of this angle. The cuts of φ(b ¯b, EmissT ) > 2.7 for 120 < p V T < 160 GeV and φ(b ¯b, EmissT ) > 2.9 for p V T  160 GeV were established from MC-based optimization studies. A search for H → b ¯b decays is performed by looking for an excess of events above the background expectation in the invari- ant mass distribution of the b-jet pair (mb ¯b). The value of the reconstructed mb ¯b is scaled by a factor of 1.05, obtained from MC-based studies, to account on average for e.g. losses due to soft muons and neutrinos from b and c hadron decays. To in- crease the sensitivity of the search, this distribution is examined in bins of pVT . As the expected signal is characterized by a rela- tively hard pVT spectrum, the signal to background ratio increases with pVT . The ZH →  +−b ¯b and WH → νb ¯b channels are exam- ined in four bins of the transverse momentum of the reconstructed W or Z boson, given by: pVT < 50 GeV, 50  p V T < 100 GeV, 100 pVT < 200 GeV and p V T  200 GeV. In the ZH → ν ¯νb ¯b search three bins are defined: 120 < pVT < 160 GeV, 160 p V T < 200 GeV and pVT  200 GeV. The expected signal to background ratios for a Higgs boson signal with mH = 120 GeV vary from about 1% in the lowest pVT bins to about 10–15% in the highest p V T bins. For this Higgs boson mass, 5.0% and 2.4% of the ZH → +−b ¯b and WH → νb ¯b events are expected to pass the respective analysis selections, with negligible contributions from other final states. On the other hand, the ZH → ν ¯νb ¯b analysis has a non-negligible con- tribution from WH → νb ¯b: 2.1% of the ZH → ν ¯νb ¯b signal and 0.2% of the WH → νb ¯b signal are expected to pass the analysis selection. 6. Background estimation Backgrounds are estimated using a combination of data-driven and MC-based techniques. Significant sources of background in- clude top, W + jet, Z + jet, diboson and multijet production. The dominant background in the ZH → +−b ¯b channel is Z + jet production. In the WH → νb ¯b channel both the top-quark and 372 ATLAS Collaboration / Physics Letters B 718 (2012) 369–390 Fi ch di an an W th bo ex is je no to is sa or po di bu Z fr an je tiv ov w st ang. 1. (a) The dilepton invariant mass distribution in the ZH → +−b ¯b channel, (b) the missing transverse energy without the mT requirement in the WH → νb ¯b annel, (c) the azimuthal angle separation between EmissT and p miss T and (d) the minimum azimuthal separation between E miss T and any jet in the ZH → ν ¯νb ¯b channel. All stributions are shown for events containing two b-tagged jets. The various Monte Carlo background distributions are normalized to data sidebands and control distributions d the multijet background is entirely estimated from data as described in the text. The vertical dashed lines correspond to the values of the cuts applied in each analysis, d the horizontal arrows indicate the events selected by each cut. + jet production are important. In the ZH → ν ¯νb ¯b channel, ere is a significant contribution from top, W + jet, Z + jet and di- son production. Multijet production is a negligible background, cept for the WH → νb ¯b channel. The flavour composition of the W + jet and Z + jet backgrounds determined partially from data. The shapes of the mb ¯b distribution of the top, W + jet and Z + t backgrounds are taken from MC simulation, with the respective rmalizations being determined from data. The ratio of single-top top-pair production is taken from NLO QCD computations [48]. The flavour composition of the W + jet and Z + jet samples determined using templates produced from three exclusive MC mples containing at least one true b jet, at least one true c jet, only light jets. The relative normalizations of the three com- nents are adjusted by fitting the distribution of the b-tagging scriminating variable w found in MC simulation to the distri- tion found in control data samples dominated by W + jet and + jet events. For the Z + jet sample this is a Z reconstructed om 2 electrons or muons and 2 jets. The W + jet sample is a W d 2 jets with an additional cut on the invariant mass of the 2 ts of less than 80 GeV to reduce top background. Once the rela- e normalizations of the flavour components have been fixed, the erall normalizations are determined from data in a separate step. Sidebands in the mb ¯b distribution, defined by selecting events In addition, two control regions which are dominated by top- quark production are used to further constrain the normaliza- tion of the top background. The ZH top control region selects events from the sidebands of the Z boson mass peak: m ∈ [60 GeV,76 GeV]∪ [106 GeV,150 GeV] with EmissT > 50 GeV, while the WH top control region selects W + 3 jet events with two b- tagged jets. The normalizations of the Z + jet, W + jet and top-quark back- grounds are determined in the ZH → +−b ¯b or WH → νb ¯b channels, by simultaneous fits to the sidebands of the mb ¯b distri- butions, and either the ZH or WH top control regions defined above. In the WH sideband fit, the normalizations of the top- quark, the W + 2 jet and the W + 3 jet distributions are varied. In the ZH sideband fit, the normalizations of the top-quark and Z + jet backgrounds are left floating. The normalizations of the re- maining sub-leading backgrounds are left fixed in the fit at their expectation values from Monte Carlo predictions, except for mul- tijet production which is estimated from data. The relative data to MC normalization factors for top-quark background agree with unity to within 20% in both the ZH → +−b ¯b or WH → νb ¯b sideband fits. The normalization of the top-quark background in the ZH → +−b ¯b signal region is based on the ZH sideband and control region fit result. The normalization of the top-quark background in the WH → νb ¯b and ZH → ν ¯νb ¯b signal regions isith mb ¯b < 80 GeV or 150 GeV < mb ¯b < 250 GeV along with the andard event selection, are used to normalize the Z + jet, W + jet d top backgrounds. based on the WH sideband and control region fit result. Monte Carlo simulation is used to estimate the shape of the Z + jet (W + jet) background, while its normalization is determined in the ATLAS Collaboration / Physics Letters B 718 (2012) 369–390 373 Fi an th w Z th an tip ad to ag w fr ba mg. 2. The invariant mass mb ¯b for ZH →  +−b ¯b shown for the different pZT bins: (a) 0 < p Z T < 50 GeV, (b) 50 p Z T < 100 GeV, (c) 100 p Z T < 200 GeV, (d) p Z T  200 GeV d (e) for the combination of all pZT bins. The signal distributions are shown for mH = 120 GeV and are enhanced by a factor of five for visibility. The shaded area indicates e total uncertainty on the background prediction. For better visibility, the signal histogram is stacked onto the total background, unlike the various background components hich are simply overlaid in the distribution. H → +−b ¯b (WH → νb ¯b) sidebands to the signal regions of all ree channels. The MC to data normalization factors for W + jet d Z+ jet range from 0.8 to 2.4 depending on jet flavour and mul- licity. The normalization factors are applied to the MC in several ditional control samples with selections to enhance the Z , W or p-quark contributions. After these corrections are applied, good reement is found with the data in both shape and normalization ithin the statistical and systematic uncertainties. The backgrounds from multijet events are estimated entirely + − to contribute less than 1% and is therefore neglected. Multijet EmissT templates for the WH → νb ¯b channel are obtained by selecting events with lepton candidates failing the charged lepton analysis selection, but satisfying looser lepton selections. The normalization is determined by fitting these templates to the EmissT distribution. A 30% uncertainty is determined from a comparison between the normalized templates and the data in a multijet-dominated control region, defined by requiring EmissT < 25 GeV and mT < 40 GeV. In the ZH → ν ¯νb ¯b channel, the multijet background is esti- om collision data. For the ZH →   b ¯b channel, the multijet ckground normalization is determined from the sidebands of the  distribution in events containing at least two jets, and is found mated using three control regions defined using two variables, φ(EmissT , p miss T ) and min( φ(E miss T , jets)), which showed no ap- preciable correlation. The ratio of events with φ(EmissT , jet) > 1.8 374 ATLAS Collaboration / Physics Letters B 718 (2012) 369–390 Fi an th w to w w in in 12 sp of in ce twg. 3. The invariant mass mb ¯b for WH → νb ¯b shown for the different p W T bins: (a) 0 < p W T < 50 GeV, (b) 50 p W T < 100 GeV, (c) 100 p W T < 200 GeV, (d) p W T  200 GeV d (e) for the combination of all pWT bins. The signal distributions are shown for mH = 120 GeV and are enhanced by a factor of five for visibility. The shaded area indicates e total uncertainty on the background prediction. For better visibility, the signal histogram is stacked onto the total background, unlike the various background components hich are simply overlaid in the distribution. those with min( φ(EmissT , jet)) < 1.8 is determined for events ith φ(EmissT , p miss T ) > π/2. This ratio is then applied to events ith φ(EmissT , p miss T ) < π/2 to estimate the multijet background the signal region. Upper estimates of the multijet contamination the signal region are found to be 0.85, 0.04 and 0.26 events for 0 < pVT < 160 GeV, 160 p V T < 200 GeV and p V T  200 GeV, re- ectively. The accuracy of the estimate is limited by the number events in the control regions. The distribution of m in the ZH → +−b ¯b channel is shown by Z + jet with smaller contributions from top-quark and dibo- son production. The EmissT distribution in the WH → νb ¯b chan- nel is shown in Fig. 1(b) after all requirements, except for the mT and EmissT cuts. The signal region is seen to have large con- tributions from top-quark production and W + jet, with smaller contributions from the multijet background, Z + jet and dibo- son production. Figs. 1(c) and 1(d) show the φ(EmissT , p miss T ) and min( φ(EmissT , jet)) distributions for the ZH → ν ¯νb ¯b channel, af- ter all requirements except for those applied to these variables.Fig. 1(a) after all analysis requirements have been applied (ex- pt for the di-lepton mass cut), including the requirement of o b-tagged jets. The signal region is seen to be dominated The multijet background shape in Fig. 1(c) is obtained from data events with min( φ(EmissT , jet)) < 0.4, after subtracting the re- maining backgrounds, and normalized to the data in the region ATLAS Collaboration / Physics Letters B 718 (2012) 369–390 375 Fi co un ar de is iz tiv co th di ul da 7. fe le th an th re be th 3% de th je ofg. 4. The invariant mass mb ¯b for ZH → ν ¯νb ¯b shown for the different p Z T bins: (a) 120 < p Z T < 160 GeV, (b) 160 p Z T < 200 GeV, (c) p Z T  200 GeV and (d) for the mbination of all pZT bins. The signal distributions are shown for mH = 120 GeV and are enhanced by a factor of five for visibility. The shaded area indicates the total certainty on the background prediction. For better visibility, the signal histogram is stacked onto the total background, unlike the various background components which e simply overlaid in the distribution. fined by φ(EmissT , p miss T ) > π/2. In Fig. 1(d), the multijet shape obtained from events with φ(EmissT , p miss T ) > π/2 and normal- ed to data events with min( φ(EmissT , jet)) < 0.4. It can be seen that the requirements on these variables effec- ely reduce the multijet background. The signal region has large ntributions from Z + jet and top, with smaller contributions from e W + jet, diboson production and multijet backgrounds. For all stributions, the data are reasonably well described by MC sim- ation and the multijet background, which was determined from ta. Systematic uncertainties The sources of systematic uncertainty considered are those af- cting the various efficiencies (reconstruction, identification, se- ction), as well as the momentum or energy of physics objects, e normalization and shape of the mb ¯b distribution of the signal d background processes, and the integrated luminosity. Among ese, the leading instrumental uncertainties for all channels are lated to the uncertainty on the b-tagging efficiency, which varies tween 5% and 19% depending on the b-tagged jet pT [44], and e jet energy scale (JES) for b-tagged jets which varies between and 14% depending on the jet pT and η [49]. The pT depen- nce of the b-tagging efficiency has been considered, based on relative fraction of Z + c-jets and W + c-jets derived from the fit described in Section 6 by 30%. The uncertainties on the SM Higgs boson inclusive cross sec- tions are evaluated by varying the factorization and renormaliza- tion scales, and by taking into account the uncertainties on the PDFs, on the strong coupling constant and on the H → b ¯b branch- ing fraction. These uncertainties are estimated to be ≈ 4% for both WH and ZH production and are treated according to the rec- ommendations given in Refs. [21,50,51]. Additional uncertainties are considered, as a function of the transverse momentum of the W and Z bosons, which range from ≈ 4% to ≈ 8%, depending on channel and on the pWT or p Z T interval. These correspond to the dif- ference between the inclusive and differential electroweak correc- tions [22,12], and to differences in acceptance between the Pythia and Powheg + Herwig generators. The latter arise mainly from the perturbative QCD model uncertainty caused by rejecting events with three or more jets in the WH → νb ¯b and ZH → ν ¯νb ¯b anal- yses. The uncertainties on the normalizations of the Z + jet, W + jet and top-quark backgrounds are taken from the statistical uncer- tainties on the fits to control regions and mb ¯b sidebands (see Sec- tion 6) and from variations of the nominal fit result induced by the remaining sources of systematic uncertainty. The resulting nor- malization uncertainties are applied to the ZH → ν ¯νb ¯b channel. A correlation between the normalizations of the W + jet and top-e full covariance matrix of the measured b-tagging efficiency in t pT intervals [44]. The uncertainty on the flavour composition the Z + jet and W + jet background is estimated by varying the quark backgrounds is introduced by the simultaneous fit to the mb ¯b sidebands and the WH top control region in the WH → νb ¯b channel. This correlation is taken into account when transferring 376 ATLAS Collaboration / Physics Letters B 718 (2012) 369–390 Table 1 Number of data, simulated signal, and estimated background events in each bin of pVT for the WH → νb ¯b, ZH →  +−b ¯b and ZH → ν ¯νb ¯b channels. The signal corresponds to a Higgs boson mass of mH = 120 GeV. The number of events is shown for the full signal region (mb ¯b ∈ [80 GeV,150 GeV]). Background sources found to be negligible are signalled with “–”. Relative systematic uncertainties on the hypothesized signal and estimated total background are shown. bin ZH → +−b ¯b W H → νb ¯b ZH → ν ¯νb ¯b pVT [GeV] p V T [GeV] p V T [GeV] 0–50 50–100 100–200 >200 0–50 50–100 100–200 >200 120–160 160–200 >200 Number of events for 80