Search for charged Higgs bosons produced via vector boson fusion and decaying into a pair of W and Z bosons using proton-proton collisions at √ s = 13 TeV

A search for charged Higgs bosons produced via vector boson fusion and decaying into W and Z bosons using proton-proton collisions at √ s = 13 TeV is presented. The data sample corresponds to an integrated luminosity of 15.2 fb − 1 collected with the CMS detector in 2015 and 2016. The event selection requires three leptons (electrons or muons), two jets with large pseudorapidity separation and high dijet mass, and missing transverse momentum. The observation agrees with the standard model prediction. Limits on the vector boson fusion production cross section times branching fraction for new charged physical states are reported as a function of mass from 200 to 2000 GeV and interpreted in the context of Higgs triplet models.

Searches for charged Higgs bosons (H ± ) at the LHC currently focus on the production and the decay to fermions [22][23][24][25][26][27][28][29][30], well motivated by the minimal supersymmetric standard model [31]. In this model, the H ± tb coupling is the strongest irrespective of the mass of the charged Higgs boson (m(H ± )) and tan β, the ratio of the vacuum expectation values (vevs) of the two Higgs doublets. Couplings to vector bosons are however largely suppressed in these models.
Fermiophobic charged Higgs bosons appear in Higgs sectors extended by SU(2) triplets. Couplings to W and Z bosons are introduced at tree level. A prominent example is the Georgi-Machacek (GM) model [32], where real and complex Higgs triplets with high vevs are arranged to preserve custodial symmetry. In such models, the charged Higgs bosons are produced via vector boson fusion (VBF) and the couplings depend on m(H ± ) and the parameter sin θ H , or s H , where s 2 H denotes the fraction of the W boson mass squared generated by the vev of the triplets. The leading Feynman diagram for the production and decay is shown in Fig. 1. In this Letter, we discuss the search for charged Higgs bosons that are produced via VBF and decay via couplings to W and Z bosons. The analysis is performed on a sample of protonproton collisions collected at √ s = 13 TeV center-of-mass energy by the CMS experiment at the LHC. The data sample corresponds to integrated luminosities of 2.3 and 12.9 fb −1 recorded during the years 2015 and 2016, respectively. The search is performed using W and Z bosons decaying into electrons and muons. The event selection requires two jets with large pseudorapidity separation and high dijet mass to select a VBF topology. The data are compared to the predictions of the GM model for a charged Higgs boson mass range of 200 < m(H ± ) < 1000 GeV. In addition, an exclusion limit on the VBF production cross section times branching fraction (B) for 200 < m(H ± ) < 2000 GeV is derived. A similar search was performed by the ATLAS collaboration in proton-proton collisions at √ s = 8 TeV in the semi-leptonic (WZ → qq ) final state [33].
The signal samples are produced with MADGRAPH5 aMC@NLO v2.2.2 [34]. The WZ production in association with two jets involving exclusively weak interactions at tree level is generated using MADGRAPH5 aMC@NLO and is referred to as EW WZ background. Two jet associated WZ production with both the strong and electroweak interaction vertices at tree level is simulated using POWHEG 2.0 [35][36][37][38] and is denoted as QCD WZ background. The Z+jets, Zγ, tt, ttV, VVV, where V refers to a W or Z boson, are produced using MADGRAPH5 aMC@NLO . The gg → ZZ process is generated with MCFM [39], while ZZ production via qq annihilation is simulated at next-to-leading order (NLO) with POWHEG. The PYTHIA 8 [40] package is used for parton showering, hadronization, and the underlying event simulation with parameters affecting the underlying event simulation set to the CUETP8M1 tune [41,42]. The NNPDF 3.0 [43] set is used as the default set of parton distribution functions (PDFs). For all processes, the detector response is simulated using a detailed description of the CMS detector, based on the GEANT4 package [44], and event reconstruction is performed with the same algorithms as used for data. The simulated samples include additional interactions per bunch crossing (pileup) matching the observed multiplicity in the data of about 11 and 20 interactions per bunch crossing in 2015 and 2016, respectively.
Details of the CMS detector, its performance and the definition of the coordinate system can be found in Ref. [45]. The detector features a superconducting solenoid with a diameter of 6 m, providing a magnetic field of 3.8 T, and surrounding a silicon pixel and strip tracking detector, a lead tungstate electromagnetic calorimeter and a brass scintillator hadronic calorimeter. Gas ionization detectors embedded into the steel-flux return yokes, the muon system, are installed around the solenoid. The subdetectors are composed into a barrel and two endcap sections. The hadron forward calorimeter provides calorimetry to pseudorapidities from |η| > 3 to |η| < 5. A particle-flow technique is employed to identify and reconstruct the individual particles emerging from each collision [46,47].
Electrons are reconstructed within |η| < 2.5. The reconstruction combines the information from clusters of energy deposits in the electromagnetic calorimeter and the trajectory in the tracker [48]. The selection criteria depend on transverse momentum p T and |η|, and on a categorization based on observables sensitive to the amount of bremsstrahlung emitted. Muons are reconstructed within |η| < 2.4 [49]. The reconstruction combines the information from both the tracker and the muon spectrometer. Leptons are required to be isolated from other charged and neutral particles in the event. The lepton relative isolation is defined as the ratio of the p T sum of charged hadrons and neutral particles within a cone of radius ∆R = √ (∆η) 2 + (∆φ) 2 < 0.4 (where φ is the azimuthal angle in radians) around the lepton and the lepton p T . The relative isolation, corrected for pileup contributions, is required to be less than 6.5% (15%) for electrons (muons). Overall efficiencies of the reconstruction, identification, and isolation requirements for the prompt leptons are measured in data in several bins of p T and |η| using a "tag-andprobe" technique [50] applied to a sample of leptonically decaying Z boson events.
Jets are reconstructed using the anti-k T clustering algorithm [51] with a distance parameter R = 0.4, as implemented in the FASTJET package [52, 53], and jet energy corrections are applied [54,55]. To suppress the top quark background contribution in its decay to b quarks, the combined secondary vertex b tagging algorithm [56, 57] requirement is used corresponding to an efficiency of about 45% with a light flavor quark misidentification probability of 0.1%.
The missing transverse momentum vector p miss T is defined as the negative vectorial sum of the momenta of all reconstructed particles in an event projected onto the plane perpendicular to the beams, corrected for the pileup contribution [58]. Its magnitude is referred to as the p miss T . Events are selected by the trigger system requiring the presence of one or two high p T electrons or muons. The trigger efficiency is greater than 99% for events that pass all other selection criteria explained in the following. The selection of events aims to single out three-lepton events with the VBF topology. The event selection requires three lepton (electron or muon) candidates that meet the isolation and identification requirements. Two leptons are required to have p T > 20 GeV and the third lepton is required to have p T > 10 GeV. Events with an additional fourth lepton with p T > 10 GeV are rejected. Events are required to have at least two jets with p T > 30 GeV, and |η| < 4.7. The VBF topology is exploited by requiring that the two jets of highest p T have large dijet mass, m jj > 500 GeV, and large pseudorapidity separation, ∆η jj > 2.5. To reconstruct a Z boson candidate a pair of same-flavor and opposite-charge leptons is required to have a dilepton invariant mass within 15 GeV of the nominal Z boson mass [59]. When there are two or more candidate pairs, the one with the mass closest to the nominal Z boson mass is chosen. The remaining lepton is associated with the W boson decay, and it is required to have p T > 20 GeV. The p miss T in the event is required to be larger than 30 GeV to select W boson decays. To reject the top-quark background, the event must not have jets passing the b-tagging selection.
After these requirements, the signal efficiency is about 10-15%, depending on m(H ± ). For extraction of the signal, the shape of the distribution of the transverse mass variable (m T ) obtained from the WZ system is used where the p T (W) is reconstructed from the vectorial sum of p miss T and the lepton p T , and E T (W) is calculated from the scalar sum of the lepton transverse energy and p miss T . A combination of methods using control samples in the data and detailed simulation studies is used to estimate background contributions. The following background categories are considered: WZ, ZZ → 4 , VVV, Zγ and processes with nonprompt leptons.
The QCD and EW WZ background constitutes about 80% of the total expected SM background yield. The normalization of the QCD WZ background is obtained from a backgrounddominated sideband defined by the dijet variables outside of the search region where the expected signal yield is negligible: 100 GeV < m jj < 500 GeV and |∆η jj | < 2.5. In this phase-space region, expected background contributions from ZZ → 4 , VVV productions and nonprompt leptons are subtracted from the number of events in data. The simulated sample of QCD WZ processes is then normalized to match the observed number of events in this control region. The estimated normalization of events is consistent with the SM prediction obtained using the POWHEG NLO cross section calculation. The EW WZ background is obtained at leading-order using the MADGRAPH5 aMC@NLO simulation and contributes about 30% to the overall WZ background processes.
The ZZ → 4 , VVV and Zγ contributions are estimated from simulated samples, with corrections to the lepton reconstruction, trigger and selection efficiencies, and momentum scale and resolution measurements estimated from data control samples. The overall expected contribution from these processes to the total background yield is about 10%, and the uncertainties in the estimates are dominated by the statistical component introduced by the number of simulated events passing the event selection requirements. The ZZ → 4 background is largely reduced by the p miss T requirement and the veto on events containing an additional lepton. The qq → ZZ process is normalized to the next-to-next-to-leading order (NNLO) cross section prediction with a K-factor of 1.1 [60] and gg → ZZ at NLO with a K-factor of 1.7 [61].
The main contributions to nonprompt leptons are from Z+jets and top-quark (tt and tW) events, where at least one of the jets or a constituent is misidentified as an isolated lepton. The dominant background at the final-selection level is Z+jets. According to the simulation, fewer than 10% of the background events with at least one nonprompt lepton come from top quark processes. Data control samples are used to estimate this background. Lepton candidates selected with loose identification requirements are defined in a sample of events dominated by dijet production. The efficiency for candidates to pass the full lepton selection criteria is measured and is parametrized as a function of p T and η. The calculated efficiencies are used as weights to extrapolate the yield of the sample of loose leptons to the sample of fully selected leptons. The background estimation method is validated on a nonprompt lepton W+jets and tt enriched sample, selected by inverting the Z boson mass or b tagging criteria, where good agreement between the data and prediction is observed.
Residual discrepancies in the lepton reconstruction, identification and trigger efficiencies between data and simulation are corrected by applying data-to-simulation scale factors measured using Z+jets events in the Z boson peak region [50] recorded with unbiased triggers. These factors depend on the lepton p T and η and are within 2% of unity for electrons and muons. The experimental uncertainties in the lepton momentum scale and resolution, p miss T modeling, and jet energy scale are applied in simulated events by smearing and scaling the relevant observables and propagating the effects to the kinematic variables used in the analysis, in particular m T . Uncertainties in the lepton momentum scale and resolution are smaller than 1% per lepton depending on the p T and η of the lepton, and the effect on the yields at the analysis selection level is less than 1%. The uncertainties in the jet energy scale and resolution result in a 5% uncertainty in the signal yields. The uncertainty in the resolution of the p miss T measurement is 10%. Randomly smearing the measured p miss T by one standard deviation of the resolution gives rise to a 5% variation in the estimation of signal yields after the full selection. Uncertainties of 2.3% and 2.5% are assigned to the integrated luminosity measurements in the years 2015 and 2016, respectively [62, 63]. The effect of higher-order corrections to the signal cross section in the GM model is taken from Ref. [64]. The theoretical uncertainty is dominated by missing higher-order EW corrections estimated to be 7%. Uncertainties due to PDF choice and renormalization and factorization scales are 2-3% and less than 1%, respectively.
The uncertainty in the estimation of the expected number of QCD WZ events is 12%, which is estimated from the measured yields in the two-jet control region. An uncorrelated uncertainty of 30% is assigned on the normalization of WZ events produced via EW processes, estimated from variations of the renormalization and factorization scale. The total uncertainty in the prediction of the nonprompt background varies bin-by-bin in the m T distribution between 30% and 80%, dominated by the low number of nonprompt leptons passing the sideband selection. A summary of the relative systematic uncertainties in the estimated signal and background yields is shown in Table 1.   10.0 ± 0.8 59.9 ± 3.5 Signal (m(H ± ) = 700 GeV) 0.9 ± 0.1 4.7 ± 0.5 In summary, we present a search for charged Higgs bosons produced via vector boson fusion and decaying into W and Z bosons in proton-proton collisions at √ s = 13 TeV based on a sample corresponding to an integrated luminosity of 15.2 fb −1 . Events are required to have three leptons (electrons or muons), two jets with large pseudorapidity separation and high dijet mass, and missing transverse momentum. The number of events observed in the signal region agrees with the standard model prediction. The first limits on σ VBF (H ± ) B(H ± → WZ) at √ s = 13 TeV are obtained. The results are interpreted in the Georgi-Machacek model for which the most stringent limits are derived.
We congratulate our colleagues in the CERN accelerator departments for the excellent performance of the LHC and thank the technical and administrative staffs at CERN and at other CMS institutes for their contributions to the success of the CMS effort. In addition, we gratefully acknowledge the computing centers and personnel of the Worldwide LHC Computing Grid for delivering so effectively the computing infrastructure essential to our analyses.     [31] S. Dimopoulos and H. Georgi, "Softly broken supersymmetry and SU (5)