Heavy Higgs bosons and the 2 TeV W′ boson

The hints from the LHC for the existence of a W′ boson of mass around 1.9 TeV point towards a certain SU(2)L × SU(2)R × U(1)B−L gauge theory with an extended Higgs sector. We show that the decays of the W′ boson into heavy Higgs bosons have sizable branching fractions. Interpreting the ATLAS excess events in the search for same-sign lepton pairs plus b jets as arising from W′ cascade decays, we estimate that the masses of the heavy Higgs bosons are in the 400-700 GeV range.


Introduction
Using LHC data at √ s = 8 TeV, the ATLAS and CMS Collaborations have reported deviations from the Standard Model (SM) of statistical significance between 2σ and 3σ in several final states, indicating mass peaks in the 1.8-2 TeV range [1][2][3][4][5][6][7][8]. The cross sections required for producing these mass peaks are consistent with the properties of a W boson in an SU(2) L × SU(2) R × U(1) B−L gauge theory with right-handed neutrinos that have Dirac masses at the TeV scale [9].
The spontaneous breaking of SU(2) × SU(2) × U(1) gauge groups requires an extended Higgs sector. For large regions of parameter space, the W boson has large branching fractions into heavy scalars from the Higgs sector [10,11]. We show here that the W boson hinted by the LHC data may also decay into H + A 0 and H + H 0 , where H + , A 0 and H 0 are heavy spin-0 particles present in Two-Higgs-Doublet models. We compute the branching fractions for these decays and present some evidence that signals for the W → H + A 0 /H 0 processes may already be visible in the 8 TeV LHC data.
There are numerous and diverse studies of SU(2) L × SU(2) R × U(1) B−L models, spanning four decades [12][13][14][15][16][17][18][19][20][21][22][23]. An interesting aspect of the left-right symmetric models is that they can be embedded in the minimal SO(10) grand unified theory. This scenario must be significantly modified due to the presence of Dirac masses for right-handed neutrinos required by the CMS e + e − jj events. 1 The theory introduced in [9] involves at least one vectorlike fermion transforming as a doublet under SU(2) R . This may be part of an additional SO(10) multiplet, but it may also be associated with entirely different UV completions.

JHEP10(2015)118
In section 2 we write down the Higgs potential and analyze its implications for the scalar spectrum. In section 3 we derive the interactions of the W boson with scalars, and compute all W branching fractions. The couplings of heavy Higgs bosons to quarks are discussed in section 4. LHC signals of heavy scalars produced in W decays are the subject of section 5. Our conclusions are summarized in section 6.

Extended Higgs sector
The Higgs sector of the SU(2) L ×SU(2) R ×U(1) B−L gauge theory discussed in [9] consists of two complex scalar fields: an SU(2) R triplet T of B −L charge +2, and an SU(2) L ×SU(2) R bidoublet Σ of B − L charge 0.

Scalar potential
The renormalizable Higgs potential is given by where the triplet-only potential is The bidoublet-only potential V (Σ) is chosen such that by itself it does not generate a VEV for Σ. This is discussed later, together with V (T, Σ), which collects all the terms that involve both scalars. For M 2 T > 0 and λ T + λ T > 0, the T scalar acquires a VEV, which upon an SU(2) R × U(1) B−L transformation can be written as where the minimization of V (T ) gives This breaks SU(2) R ×U(1) B−L down to the SM hypercharge gauge group, U(1) Y , leading to large masses for the W and Z bosons. The value of the T VEV is related to the parameters of the W boson. In the next section we will show that the parameters indicated by the LHC mass peaks near 2 TeV imply u T ≈ 3-4 TeV. The triplet field includes 6 degrees of freedom, and can be written as T = (T 1 , T 2 , T 3 ), with T i (i = 1, 2, 3) complex scalars, or more explicitly as where σ i are the Pauli matrices, and the factors of √ 2 are required for canonical normalization of the kinetic terms. The fields of definite electric charge, which are combinations of the T i components, include three Nambu-Goldstone bosons (G ± R , G − R ). These become the longitudinal degrees of freedom of the W ± and Z bosons. The three remaining fields are a real scalar T 0 , a doubly-charged scalar T ++ and its charge conjugate state T −− . These have masses given by For quartic couplings in the 0.1-1 range and in the absence of fine tuning, the T 0 and T ++ particles have masses comparable to, or heavier than W . The bidoublet-only potential includes the following terms: where Σ is the charge conjugate state of the bidoublet. Other terms, such as Tr[(Σ † Σ) 2 ] can be written as linear combinations of the terms in eq. (2.7). We take M 2 Σ ,M 2 Σ > 0, which is a sufficient condition for the minimum of V (Σ) to be at Σ = 0. The mixed terms, which involve both the T and Σ scalars, will induce a nonzero VEV. In terms of fields of definite electric charge, the bidoublet scalar can be written as and its charge conjugate state is The mixed terms in the potential are: Likewise, the squared mass for (Σ 0 1 , Σ − 1 ) turns negative, due to the second term in eq. (2.10), when −λ T Σ u 2 T + M 2 Σ < 0. In addition, a Σ 0 1 Σ 0 2 term is induced forλ T Σ = 0. Thus, for a range of parameters, the VEV of Σ takes the form where v H 174 GeV is the electroweak scale. We are interested in the case where u T /v H ∼ 20. The effect of the Σ VEV on the T 0 and T ++ masses and couplings is thus negligible. At energy scales below the T 0 and T ++ masses, the scalar sector consists only of Σ, which is the same as two Higgs doublets.

Higgs bosons
Using the notation of Two-Higgs-Doublet models [25], the components of Σ defined in eq. (2.8) are related to the Higgs doublets H 1 and H 2 as follows: , (2.12) where h 0 is the SM-like Higgs boson, and G ± , G 0 are the Nambu-Goldstone bosons, which become the longitudinal W and Z. We have not included here the effects of the CPviolating phase α Σ , which would lead to H 0 − A 0 mass mixing. The measurements of h 0 couplings are in good agreement with the SM predictions, implying an alignment limit [26], α = β − π/2. In addition, the observed ATLAS events [1,2] consistent with the W → W Z process indicate sin 2β 0.8 (see ref. [9]), which gives 0.5 tan β 2.
We have obtained a Two Higgs Doublet model with a potential formed of the 4 quartic terms of V (Σ), and 3 independent mass terms from V (Σ) + V (T, Σ) with T replaced by its VEV. The most general renormalizable potential for two Higgs doublets includes three more quartic terms. The allowed potential in our model is a special case of the general Two Higgs Doublet model, where λ 1 = λ 2 = λ 3 ≡ λ Σ and λ 6 = λ 7 ≡λ Σ , using the standard notation [25]. The usual −m 2 12 H 1 H 2 mass mixing term arises here from the Tr( Σ † Σ) term in V (Σ), and the last term in V (T, Σ), so that The squared masses of the charged Higgs boson and of the CP-odd scalar take a simple form: 14) The value of m 12 given by eq. (2.13) may be comparable to M W , as both are controlled by the SU(2) R breaking VEV, u T . At the same time, the weak scale is an order of magnitude smaller than u T , which requires some tuning of the parameters in the potential; a similar tuning could lead to m 12 M W . Thus, theoretically M H ± and M A may be anywhere between the weak scale and ∼ M W .
Various searches for the heavy Higgs bosons set mass limits substantially above the mass of the SM-like Higgs boson (M h = 125 GeV). It is sufficient then to expand in (v H /m 12 ) 2 1. The heavy CP-even scalar, H 0 , has a squared mass Note that the relative mass splittings between H ± , A 0 and H 0 are small. These four states approximately form a heavy SU(2) W doublet of zero VEV.

JHEP10(2015)118
The mass of the SM-like Higgs boson is related to the quartic couplings by This provides an estimate of the typical values of the quartic couplings: for tan β → 1, 2λ Σ +λ Σ +λ Σ ≈ 0.5. The departure from the alignment limit is given by (2.17)

W decays into heavy Higgs bosons
Given the Higgs sector discussed in the previous section, which breaks the SU(2) L × SU(2) R × U(1) B−L gauge symmetry down to the U(1) em group of electromagnetism, we can now derive the W couplings to bosons and the ensuing decay widths.

Couplings of the W to bosons
The SU(2) L and SU(2) R gauge bosons include electrically-charged states, W ±µ L and W ±µ R , respectively. The kinetic terms for the T and Σ scalars give rise to the following mass terms for the charged gauge boson where g 2 L and g 2 R are the SU(2) L × SU(2) R gauge couplings. Since we are interested in the case M W M W , we diagonalize the above mass matrix by keeping only the leading term in (M W /M W ) 2 . All equations that follow are valid up to corrections of order (M W /M W ) 2 ≈ 0.2%, where we used M W ≈ 1.9 TeV. The SU(2) L gauge coupling is given by the SM SU(2) W gauge coupling g, while the SU(2) R gauge coupling is given by the W coupling g R to the ud quarks in the gauge eigenstate basis: where the range for g R is required in order to explain [9] the LHC mass peaks near 2 TeV. The physical bosons, W and W , are admixtures of W ±µ L and W ±µ R : The W ±µ L − W ±µ R mixing angle θ + satisfies For M W ≈ 1.9 TeV and g R ≈ 0.45-0.6 (as determined in [9], by comparing the W production cross section to the CMS dijet excess [5,6]), we find the SU(2) R breaking scale u T ≈ 3-4 TeV. This set of parameters is compatible with constraints from electroweak precision observable [27,28]. The W W Z interactions are given by Here c W ≡ cos θ W is the usual SM parameter, and [µ, ν] represents commutation (µν − νµ) of the two Lorentz indices. Both W + L W − L Z and W + R W − R Z terms contribute to the above W W Z interactions through W L -W R mixing, with coupling strengths gc W and gs 2 W /c W , respectively. Note that the coefficient of eq. (3.6) can also be written as g R (M W /M W ) 2 sin 2β/c W . Comparing this form with eq. (4) of [9] gives ξ Z = sin 2β.
The W interactions with a W and a neutral Higgs boson arise from the Σ kinetic term: also a W ± interaction with ZH ∓ , which gets contributions from the kinetic terms of Σ and T : The W ± couplings to H ∓ and one of the neutral Higgs bosons are the following: (3.9) In the alignment limit with tan β → 1 the above W ± H ∓ h 0 coupling vanishes, while the W ± H ∓ H 0 coupling reaches its maximum. Finally, the W ± H ∓ A 0 coupling is

W branching fractions
The dominant decay modes of W are into fermion pairs, mainly quark-antiquark pairs. W can also decay into bosons, leading to challenging and interesting phenomenology. Amongst many possible bosonic decay modes, W → W Z and W → W h 0 are inevitable, originating from the kinetic term of the bidoublet scalar field Σ. Unlike other bosonic decays discussed below, the widths of these two modes do not depend on the unknown masses of heavy scalars. Eqs. (3.6) and (3.7) imply where terms of order (M W,Z,h /M W ) 2 have been neglected, and the approximate relation c W M Z M W has been used. We use full expressions for numerical study later in this section, which lead to a slight difference between the partial widths of these two W decay modes, arising from differences in underlying dynamics and kinematics.
The equivalence theorem requires that W decays into fields that are part of the same Higgs doublet (the longitudinal Z and h 0 for α → β − π/2 in this case) have equal decay widths up to electroweak symmetry breaking effects and phase-space factors. As mentioned in ref. [9], in the alignment limit Γ(W → W Z) Γ(W → W h 0 ).
Besides h 0 and the longitudinal W and Z, the bidoublet field Σ includes the heavy scalars H ± , H 0 , A 0 . The range of allowed masses for these particles spans more than an order of magnitude, from the weak scale to the SU(2) R breaking scale. If they are lighter than the W boson, then W decays may provide the main mechanisms for production of these scalar particles at hadron colliders. The W decays into a heavy scalar and an electroweak boson have widths where φ labels the heavy neutral scalars, and ξ φ follows from the couplings in eq. (3.7): and a similar replacement for W ± → ZH ± , where the function F is defined by and f is the phase-space suppression factor: Similarly, the W decay width into a charged Higgs boson and the SM-like Higgs particle is The terms of order M h /M W , neglected here, are taken into account by replacing the last factor in the above equation by For M H ± < M W /2, the W can also decay into a pair of heavy scalars, with widths Note that the leading order in M 2 H ± /M 2 W is independent of tan β. The dominant decay modes of W are into quarks, and have the following widths: for light flavors, and for heavy flavors. QCD corrections increase these two widths by about 3%. The decay widths into a τ lepton or an electron and the N τ right-handed neutrino are where s θe is the coefficient of the g R W νē R γ ν N τ R interaction term in the Lagrangian, and the analogous coefficient for the muon satisfies s θµ s θe . The baseline W model used in [9] to explain the excess events near 2 TeV reported in several channels by ATLAS and CMS, including e + e − jj [3], has s θe ≈ 0.5, and is consistent with all current constraints on flavor-changing processes. 2 We emphasize that taking into account the e + e − jj excess was crucial in identifying our baseline W model (without it, leptophobic W models [30,31] are interesting alternatives). If the W boson will be discovered in Run 2, then the eejj process will allow various tests of the underlying couplings [32].
The W → jj and W → tb channels have a combined branching fraction of approximately 86%, as shown in figure 1. The branching fractions for W → τ N τ R and W → eN τ R add up to 6% for the benchmark value of M N τ R = 1 TeV. The remaining decay modes are into bosons, which together have a branching fraction of approximately 8% for M H ± = 500 GeV, as indicated by the dashed line in figure 1. Note that the difference between tan β = 1 and 2 cannot be resolved, as expected from eq. (3.19).
Eqs. (3.11)- (3.22) show that the W widths into bosons depend on the almost degenerate masses of the heavy scalars, and on sin 2β whose value can be between 0.8 and 1, as favored by the CMS and ATLAS mass peaks near M W = 1.9 TeV discussed in [9]. The dependence on the CP-even Higgs mixing angle α is very weak, due to the alignment limit eq. (2.17), which implies cos(α + β) sin 2β and cos(α − β) 1. We show the W branching fractions in figure 2 as a function of M H ± M A M H 0 , for sin 2β = 1 (i.e., tan β = 1) and sin 2β = 0.8 (i.e., tan β = 2 or 1/2). The W branching fractions to W h 0 and W Z are almost constant because these partial widths do not depend on the masses of the heavy scalars. The branching fractions of W into H ± H 0 and H ± A 0 are equal because of the alignment condition and the approximate mass degeneracy. Similarly, decay widths of W H 0 , W A 0 , H ± Z and H ± h 0 are equal, suppressed by cos 2 2β, and with a different phase-space function than the two-heavy-scalar modes. We can see from figure 2 that the tan β = 1 case is simple: W Z, W h 0 , H + H 0 and H + A 0 modes are maximal, and the other -9 -JHEP10(2015)118 decay modes vanish. When tan β deviates from unity, the other four decay modes start to emerge with sub-percent level branching fractions.

Quark masses
The SU(2) L × SU(2) R × U(1) B−L gauge symmetry allows quark masses to be generated by Yukawa couplings to the bidoublet Σ: Here i, j = 1, 2, 3 label the SM fermion generations, is the corresponding SU(2) R doublet. The Yukawa couplings y ij andỹ ij are complex numbers. The mass terms for the up-and down-type quarks are then These terms highlight a problem. As we are interested in tan β = O(1), the above terms generically induce masses of the same order of magnitude for up-and down-type quarks. Even though the fermion mass hierarchies are not understood in the SM, they can be fitted by appropriately small Yukawa couplings. In the case of eq. (4.2), the known ratio of the bottom and top quark masses can be fitted only by tuning various parameters. For example, m b v H can be achieved by imposingỹ 33 /y 33 − tan β. In the absence of an explanation for the tuning of these independent parameters, it is useful to explore alternative mechanisms for fermion mass generations.
Consider the case where y ij andỹ ij are negligibly small. A different type of gaugeinvariant operator that can generate the quark masses is where the SU(2) R indices of the scalar triplet T are contracted with T † such that T † T belongs to the triplet representation of SU(2) R . The flavor-dependent coefficients C ij are complex dimensionless parameters, and form a 3 × 3 mass matrix C. We assume that the analogous operator with Σ replaced by Σ has negligibly small coefficients. Replacing T by its VEV given in eq. (2.3), the dimension-6 operator of eq. (4.3) generates a 3 × 3 mass matrix for the up-type quarks, while the down-type quarks remain massless at this stage. There is, however, an analogous effective operator with T replaced by T ≡ σ 2 T * σ 2 , This generates a mass matrix only for down-type quarks:

JHEP10(2015)118
Given that the coefficientsC ij are different than C ij , we obtain mass matrices for the upand down-type quarks that are independent of each other, as in the SM. Various possibilities for the origin of the effective operators (4.3) and (4.5) will be discussed elsewhere. The operators (4.3) and (4.5) have an additional useful property: they separate the contributions of the two Higgs doublets to the quark masses. Notice that the embedding (2.12) of H 1 and H 2 in Σ implies Σ = ( H 2 , H 1 ), where H 2 = iσ 2 H * 2 . From eqs. (4.3) and (4.5) then follows We have obtained the Yukawa couplings of the Two-Higgs-doublet model of Type II. This automatically avoids tree-level flavor-changing processes (see [33] and references therein). The couplings of the heavy Higgs bosons to the physical eigenstates of the quarks are proportional to the quark masses, with overall coefficients of sin β for up-type quarks and cos β for down-type quarks. For 0.5 tan β 2 and close to the alignment limit, as indicated [9] by the W signals near 2 TeV, the branching fractions for H + → tb and H 0 , A 0 → tt are almost 100%.

Signals of heavy scalars produced in W decays
Even though the W decays into heavy scalars have only percent-level branching fractions (see figure 2), the large W production cross section at the LHC [9] (200-350 fb at √ s = 8 TeV, and 1-2 pb at √ s = 13 TeV) makes these decays promising discovery channels. The pp → W → H ± A 0 and H ± H 0 processes lead to cascade decays that end up with 3W + 4b final states, including where = e or µ. The charge conjugate processes lead to a number of events smaller by a factor of almost 2, due to the smaller W − production in pp collisions. The cross section times branching fraction relevant for these processes is shown in figure 3 for the LHC at 8 and 13 TeV. We used there tan β = 1 and a W production cross section of 300 fb at √ s = 8 TeV for M W = 1.9 TeV (implying a cross section of 1.7 pb at √ s = 13 TeV). An additional W decay mode that contributes to final states with same-sign leptons or three leptons and b jets is The branching fraction for the N τ decay used here is B(N τ → τ − tb) ≈ 23%, while similar final states arise from B(N τ → e − tb) ≈ 8%. The cascade decays W + → e + N τ with N τ → τ tb or etb (the latter process is mentioned in [34]) also contribute.  The heavy scalars may also be produced directly, without W decays. A promising channel is the production of A 0 and H 0 in association with a tt pair [35,36]: Compared to the SM tth 0 process at large M h , each of the above two processes has a cross section scaled by (tan β) −2 . In figure 3 we show the cross section for this process, computed at leading order using MadGraph5 aMC@NLO [37] with the MSTWnlo2008 [38] parton distribution functions. We see that the processes listed in eqs. (5.1) and (5.3) have comparable cross section times branching fractions. The ATLAS search [39] for same-sign leptons and b-jets shows some deviations from the SM in two signal regions designed for four-top final states, with a pair of same-sign leptons and H T > 700 GeV. In the signal region "SR4t3" (exactly two b-jets and E T / > 100 GeV) there are 12 events with an expected background of 4.3 ± 1.1 ± 1.1 events. In the signal region "SR4t4" (3 or more b-jets) the search found 6 events with an expected background of 1.1 ± 0.9 ± 0.4 events. We estimate that the combination of these signals represents a ∼ 3σ excess over the SM background.
The processes shown in eqs. (5.1)-(5.3) provide a possible origin of this excess. The various contributions of heavy scalars and the right-handed neutrino to the two signal regions are shown in table 1 for tan β = 1 in the upper block and tan β = 2 in the lower block. We set M H ± = M H 0 = M A 0 = 500 GeV and 100% branching fractions of the heavy scalars into tt or tb. We estimate the efficiency by folding in the branching fractions to contributing final states, including combinatorial factors, and b-tagging efficiencies (using 70% for a single b tag). To obtain the predicted number of signal events in 20 fb −1 of data, we use a rough estimate of the acceptance, A = 50%. For both values of tan β, the expected  Table 1. Contributions from W cascade decays and ttA 0 /H 0 production to the same-sign leptons plus b-jets signals at the 8 TeV LHC. The last column gives the observed and expected number of events in the ATLAS search [39]. For the upper and lower blocks of the table tan β = 1 and 2, respectively. The range of predicted signal events corresponds to the 200-350 fb range for the W production rate. The parameters are fixed as follows: heavy scalar masses are 500 GeV, the N τ mass is 1 TeV, s θe = 0.5, and M W = 1.9 TeV.
number of signal events is compatible for both signal regions with the observed deviations above the background. We note that the W -produced events include more leptons with positive charge than with negative charge; the same feature is seen in the ATLAS events (tables 11 and 12 of [39]). If the heavy scalar masses are decreased to M H ± ≈ 400 GeV, the number of signal events in table 1 increases by a factor of about 3, so that the number of predicted events becomes too large (at tan β = 2 it may still be acceptable, given the uncertainties in the event selection efficiencies). For M H ± 700 GeV the number of predicted events becomes too small to account for the ATLAS excess. Thus, the preferred mass range for the heavy Higgs bosons is 400-700 GeV.
A CMS search [40] in a similar final state with same-sign leptons and b jets has yielded a smaller excess. The sum over the number of events with two or more b jets, large H T and high lepton p T from table 3 of [40] gives 11 observed events for a background of roughly 6 ± 2 events. The compatibility of the ATLAS and CMS results in these channels needs further scrutiny. We note though that the CMS event selection includes a veto of a third lepton, while the events reported by ATLAS include events with three leptons. 3