-
In addition to the SM fields, an extra
${ {SU(2)_L}}$ triplet Higgs field is introduced, providing a solution to the neutrino mass origin, which is the so called type-II seesaw mechanism. We denote the triplet complex scalar fields Δ with hypercharge$ Y = 2 $ and SM doublet Higgs Φ as$ \Delta(3,+2) = \begin{pmatrix} \dfrac{1}{\sqrt{2}}\Delta^+ & \Delta^{++}\\ \Delta^0 & -\dfrac{1}{\sqrt{2}}\Delta^+ \end{pmatrix}\,, $
$ \Phi(2,+1) = \begin{pmatrix} \phi^+ \\ \phi^0 \end{pmatrix}\,, $
(1) where
$\Delta^0 = (\delta + {\rm i} \eta)/\sqrt{2}$ and$ \phi^0 = (\phi_R + i \phi_I)/\sqrt{2} $ . The most general scalar potential involving Δ and Φ can be written as$ \begin{aligned}[b] V(\Delta ,\,\Phi) = &m_{\phi}^2 \,\Phi^{\dagger}\Phi + m_{\Delta}^2 {\rm{Tr}} (\Delta^{\dagger} \Delta) + \rho_{\phi} \,(\Phi^{\dagger}\Phi)^2 \\&+ \frac{1}{2} \rho_1 \,[{\rm{Tr}} (\Delta_k^{\dagger} \Delta_l)]^2 + \frac{1}{2}\rho_2 \,{\rm{Tr}} (\Delta^{\dagger} \Delta^{\dagger})\,{\rm{Tr}} (\Delta \Delta)\\&+ \sigma_1 \,\Phi^{\dagger}\Phi\,{\rm{Tr}}(\Delta^{\dagger} \Delta) + \sigma_2 \,\Phi^{\dagger}\Delta^{\dagger} \Delta \Phi \\&+ (\mu\,\Phi^{\dagger}\Delta \tilde{\Phi}+ {\rm{h.c.}})\,, \end{aligned} $
(2) where
$\tilde{\Phi} = {\rm i}\tau_2 \Phi^*$ . The VEVs of the Higgs fields are given by$ \begin{aligned}[b]& \langle \Delta \rangle = \frac{1}{\sqrt{2}} \begin{pmatrix} 0 & 0\\ v_{\Delta} & 0 \end{pmatrix}\,, \\& \langle \Phi\rangle = \frac{1}{\sqrt{2}} \begin{pmatrix} 0 \\ v \end{pmatrix}\,. \end{aligned} $
(3) Without loss of generality, we assumed all parameters and VEVs are real. The triplet Higgs VEV is naturally chosen as
$ v_{\Delta} \sim {\rm{keV}} $ , well below the constraints from the electroweak precision test [31]. The conditions for minimization of potential are$ \begin{align} m_{\phi}^2 + \rho_{\phi} v^2 + \left(\frac{1}{2} \sigma_1 v_{\Delta} + \sqrt{2} \mu\right)\,v_{\Delta} & = 0\,, \end{align} $
(4) $ \begin{align} m_{\Delta}^2 v_{\Delta} + \rho_1 v_{\Delta}^3 + \left(\frac{1}{2} \sigma_1 v_{\Delta} + \sqrt{2} \mu\right)\, v^2 & = 0\,. \end{align} $
(5) After the Higgs fields Δ and Φ acquire VEVs due to the spontaneous symmetry breaking, the mass squared matrices of CP-even Higgs fields
$ (\phi_R,\delta) $ , CP-odd Higgs fields$ (\phi_I,\eta) $ , singly-charged Higgs fields$ (\phi^+,\Delta^+) $ , and double-charged Higgs fields$ \Delta^{\pm\pm} $ are given respectively by$ \begin{align}\\[-5pt] (M^0_{R})^2& = \begin{pmatrix} m_{\phi}^2 + 3 \rho_{\phi} v^2 + \sqrt{2}\mu v_{\Delta} + \dfrac{1}{2}\sigma_1 v_{\Delta}^2 & \sigma_1 v v_{\Delta}+ \sqrt{2}\mu v \\ \sigma_1 v v_{\Delta}+ \sqrt{2} \mu v & m_{\Delta}^2 + \dfrac{1}{2} \sigma_1 v^2 + \dfrac{3}{2} \rho_1 v_{\Delta}^2\\ \end{pmatrix}\,, \end{align} $ (6) $ \begin{align} (M^0_{I})^2& = \begin{pmatrix} m_{\phi}^2 + \rho_{\phi} v^2 - \sqrt{2}\mu v_{\Delta} + \dfrac{1}{2}\sigma_1 v_{\Delta}^2 & \sqrt{2} \mu v \\ \sqrt{2} \mu v & m_{\Delta}^2 + \dfrac{1}{2} \sigma_1 v^2 + \dfrac{1}{2}\rho_1 v_{\Delta}^2\\ \end{pmatrix}\,, \end{align} $
(7) $ \begin{align} (M^+)^2& = \begin{pmatrix} m_{\phi}^2 + \rho_{\phi} v^2 + \dfrac{1}{2}(\sigma_1 + \sigma_2)v_{\Delta}^2& \mu v - \dfrac{\sqrt{2}}{4} \sigma_2 v v_{\Delta} \\ \mu v - \dfrac{\sqrt{2}}{4} \sigma_2 v v_{\Delta} & m_{\Delta}^2 + \dfrac{1}{2} \sigma_1 v^2+ \dfrac{1}{4}\sigma_2 v^2+ \dfrac{1}{2} \rho_1 v_{\Delta}^2\\ \end{pmatrix}\,, \end{align} $
(8) $ (M^{++})^2 = m_{\Delta}^2 + \frac{1}{2}\sigma_1 v^2+ \frac{1}{2}\sigma_2 v^2+ \frac{1}{2} (\rho_1 + 2 \rho_2)v_{\Delta}^2\,. $
(9) By diagonalizing the above mass squared matrices, we obtain seven physical states: CP-even Higgs bosons h and H, CP-odd Higgs boson A, singly-charged Higgs boson
$ H^{\pm} $ , and double-charged Higgs boson$ H^{\pm\pm} $ . The coupling parameters in the scalar potential are expressed in terms of the masses of physical particles and mixing angle α between two CP-even Higgses,$ \begin{align} &\rho_{\phi} = \frac{1}{v^2}\left(m_{h}^2c_\alpha^2 + m_{H}^2 s_\alpha^2 \right) \,, \end{align} $
(10) $ \begin{align} &\rho_1 = \frac{1}{2 v_{\Delta}^3}\left(\sqrt{2} \mu v^2 + 4 m_{h}^2 s_\alpha^2 v_{\Delta} + 4 m_{H}^2 c_\alpha^2 v_{\Delta}\right)\,, \end{align} $
(11) $ \begin{align} &\rho_2 = \frac{1}{2 v_{\Delta}^3}\left(\sqrt{2} \mu v^2 + 2m_{H^{\pm\pm}}^2 v_{\Delta} -\sigma_2 v^2 v_{\Delta} \right) \,, \end{align} $
(12) $ \begin{align} &\sigma_1 = \frac{1}{vv_{\Delta}}\left(2\left(m_{h}^2 - m_{H}^2 \right) s_\alpha c_\alpha- \sqrt{2} \mu v \right) \,, \end{align} $
(13) $ \begin{align} &\sigma_2 = \frac{1}{v^4 + 6 v^2 v_{\Delta}^2 + 8 v_{\Delta}^4}\left(4m_{H^{\pm}}^2( v^2 + 4 v_{\Delta}^2) - 8 m_A^2(v^2 + 2 v_{\Delta}^2) \right) \,, \end{align} $
(14) $ \begin{align} &\mu = -\frac{2\sqrt{2}m_A^2 v_{\Delta}}{v^2+4v_{\Delta}^2}\,, \end{align} $
(15) where we denote
$ c_\alpha\equiv\cos \alpha $ and$ s_\alpha\equiv\sin \alpha $ . The triplet field Δ can couple with a pair of lepton doublets$ L_L^i = (\nu_L^i,\ell_L^i)^T $ through the Yukawa terms, which are written as$ \begin{aligned}[b] {\cal{L}}_{\Delta} = &\frac{1}{\sqrt{2}}(Y_{\Delta})_{ij}\overline{(L_L^i)^C}i\tau_2\Delta L_L^j + {\rm{h.c.}} = \frac{1}{\sqrt{2}}(Y_{\Delta})_{ij}\overline{(\nu_L^i)^C}\Delta^0 \nu_L^j \\&- \frac{1}{2}(Y_{\Delta})_{ij}\overline{(\nu_L^i)^C}\Delta^+ \ell_L^j - \frac{1}{2}(Y_{\Delta})_{ij}\overline{(\ell_L^i)^C}\Delta^+ \nu_L^j\\ &- \frac{1}{\sqrt{2}}(Y_{\Delta})_{ij}\overline{(\ell_L^i)^C} \Delta^{++} \ell_L^j + {\rm{h.c.}}, \end{aligned} $
(16) where C denotes the charge conjugation. When the triplet Higgs field Δ acquires VEV, the lepton number is broken, and the neutrino mass matrix is then generated from Eq. (16)
$ \begin{equation} (M_{\nu})_{ij} = (Y_{\Delta})_{ij}v_{\Delta}\,. \end{equation} $
(17) The Yukawa coupling matrix
$ Y_{\Delta} $ is fixed by the three neutrino masses and the Pontecorvo-Maki-Nakagawa-Sakata (PMNS) mixing matrix$ U_{{\rm{PMNS}}} $ $ \begin{equation} Y_{\Delta} = \frac{1}{v_{\Delta}}U^*_{{\rm{PMNS}}} M_{\nu}^{{\rm{diag}}} U^{\dagger}_{{\rm{PMNS}}}\,, \end{equation} $
(18) where
$ M_{\nu}^{{\rm{diag}}} $ denotes the diagonal neutrino mass matrix. The PMNS matrix can be parameterized as [1]$ \begin{aligned}\\[-5pt] U_{{\rm{PMNS}}} = \begin{pmatrix} c_{12} c_{13} & s_{12} c_{13} & s_{13}{\rm e}^{-{\rm i}\delta} \\ -s_{12}c_{23}-c_{12}s_{23}s_{13}{\rm e}^{{\rm i}\delta} & c_{12}c_{23}-s_{12}s_{23}s_{13}{\rm e}^{{\rm i}\delta} & s_{23}c_{13} \\ s_{12}s_{23}-c_{12}c_{23}s_{13}{\rm e}^{{\rm i}\delta} & -c_{12}s_{23}-s_{12}c_{23}s_{13}{\rm e}^{{\rm i}\delta} & c_{23} c_{13} \end{pmatrix}\,, \end{aligned} $ (19) where
$ c_{ij}\equiv \cos\theta_{ij} $ ,$ s_{ij}\equiv\sin\theta_{ij} $ , and δ is the the Dirac CP phase. For simplicity, we have set the Majorana phases in the PMNS matrix to be zero. The values of neutrino oscillation data we used are listed in Table 1. The cosmological upper limit on the sum of three light neutrino masses is given by the Planck data at$ 95{\text{%}} $ confidence level,$ \sum m_{i} \leq 0.12\ {\rm{eV}} $ [33]. The exact values of each neutrino mass and even the mass ordering still remain unknown. In this work, we have set the lightest neutrino mass to be zero for both normal hierarchy (NH) and inverted hierarchy (IH), and then, the other two neutrino masses can be fixed by the oscillation data. For the quasi-degenerate hierarchy (DH) case, we assume that the sum of the three neutrinos masses is 0.12 eV. After confirming the neutrino masses, the Yukawa coupling matrices in Eq. (18) for three different hierarchy cases are given respectively byQuantity Normal ordering Inverted ordering $\theta_{12} /(^{\circ})$ 33.44 33.45 $\theta_{23} /(^{\circ})$ 49.0 49.3 $\theta_{13} /(^{\circ})$ 8.57 8.61 $\delta_{CP} /(^{\circ})$ 195 286 $\Delta m_{21}^2 /{\rm{eV} }^2$ $ 7.42 \times 10^{-5} $ $ 7.42 \times 10^{-5} $ $\Delta m_{31}^2 /{\rm{eV} }^2$ $ 2.514 \times 10^{-3} $ $ -2.497 \times 10^{-3} $ Table 1. Global fit of neutrino oscillation parameters, taken from [32].
$ \begin{aligned}[b] Y_{\Delta} =& \dfrac{10^{-2}\, {\rm{eV}}}{v_{\Delta}}\\&\times\begin{pmatrix} 0.35+0.06\,{\rm i} & -0.25-0.14\,{\rm i} & -0.74-0.12\,{\rm i} \\ -0.25-0.14\,{\rm i} & 3.11+0.02\,{\rm i} & 2.13-0 \,{\rm i} \\ -0.74-0.12\,{\rm i} & 2.13-0 \,{\rm i} & 2.40-0.01\,{\rm i} \end{pmatrix} \;{\rm{(NH)}}\,, \end{aligned} $
(20) $ \begin{aligned}[b] Y_{\Delta} =& \dfrac{10^{-2}\, {\rm{eV}}}{v_{\Delta}}\\&\times\begin{pmatrix} 4.92+0\,{\rm i} & 0.56-0.14\,{\rm i} & 0.45-0.13\,{\rm i} \\ 0.56-0.14\,{\rm i} & 2.24+0.03\,{\rm i} & -2.46-0.03 \,{\rm i} \\ 0.45-0.13\,{\rm i} & -2.46-0.03 \,{\rm i} & 2.92-0.02\,{\rm i} \end{pmatrix} \;{\rm{(IH)}}\,, \end{aligned} $
(21) $ \begin{aligned}[b] Y_{\Delta}=& \dfrac{10^{-2}\, {\rm{eV}}}{v_{\Delta}}\\&\times\begin{pmatrix} 7.14+0.10\,{\rm i} & -0.15-0.46\,{\rm i} & -0.16-0.40\,{\rm i} \\ -0.15-0.46\,{\rm i} & 8.01+0.05\,{\rm i} & 0.74+0.04 \,{\rm i} \\ -0.16-0.40\,{\rm i} & 0.74+0.04\,{\rm i} & 7.80+0.03\,{\rm i} \end{pmatrix} \; {\rm{(DH)}}\,. \end{aligned} $
(22) The kinetic terms in the Lagrangian for the SM doublet Φ and triplet Δ are given by
$ \begin{equation} {\cal{L}}_{{\rm{gauge}}} = (D_{\mu}\Phi)^{\dagger}D^{\mu}\Phi + {\rm{Tr}}\,[(D_{\mu}\Delta)^{\dagger}(D^{\mu}\Delta)]\,, \end{equation} $
(23) where
$D_{\mu}\Phi = \partial_{\mu}\Phi-\dfrac{{\rm i}}{2}gW_{\mu}^a\tau^a \Phi - \dfrac{{\rm i}}{2}g^{\prime}B_{\mu}\Phi$ and$D_{\mu}\Delta = $ $ \partial_{\mu}\Delta-\dfrac{{\rm i}}{2}g[W_{\mu}^a\tau^a, \Delta] - g^{\prime}B_{\mu}\Delta$ . The doubly charged scalar$ (H^{\pm\pm}) $ can decay into a singly charged scalar and a W boson$ (H^{\pm}W^{\pm}) $ , a pair of leptons$ (\ell^{\pm}\ell^{\pm}) $ , or a pair of W bosons$ (W^{\pm}W^{\pm}) $ . The channel$ H^{\pm\pm} \to H^{\pm} W^{\pm} $ can be kinematically forbidden by taking$ m_{H^{\pm\pm}} \simeq m_{H^{\pm}} $ . The branching ratio of channel$ H^{\pm\pm} \to \ell^{\pm}\ell^{\pm} $ and channel$ H^{\pm\pm} \to W^{\pm}W^{\pm} $ depends on the value of the triplet Higgs VEV, and the$ H^{\pm\pm} $ dominantly decays into leptons when the triplet Higgs VEV$ v_{\Delta} < 10^{-5} $ GeV [34]. With the assumption of$ m_{H^{\pm}} \simeq m_{H} \simeq m_{A} $ and$ v_{\Delta} < 10^{-5} $ GeV,$ H^{\pm} $ , H and A dominantly decay into$ \ell^{\pm}\nu $ ,$ \nu\nu $ and$ \nu\nu $ , respectively. -
In this work, we consider a hybrid triplet portal model with two Higgs triplets to explain cosmic ray electron and positron excesses and compare this scenario with the single triplet Higgs model. Models with two triplet Higgs fields have been proposed to address the neutrino mass matrix pattern and leptogenesis [35–37] and were explored for the context of collider experiments [36,38]. We denote two complex triplet Higgs fields
$ \Delta_1 $ and$ \Delta_2 $ with hypercharge$ Y = 2 $ as$ \begin{aligned}[b] \Delta_{1}(3,+2) =& \begin{pmatrix} \dfrac{1}{\sqrt{2}}\Delta_{1}^+ & \Delta_{1}^{++}\\ \Delta^0_{1} & -\dfrac{1}{\sqrt{2}}\Delta_{1}^+ \end{pmatrix}\,, \\ \Delta_{2}(3,+2) =& \begin{pmatrix} \dfrac{1}{\sqrt{2}}\Delta_{2}^+ & \Delta_{2}^{++}\\ \Delta^0_{2} & -\dfrac{1}{\sqrt{2}}\Delta_{2}^+ \end{pmatrix}\,, \end{aligned} $
(24) where
$\Delta_{1,2}^0 = (\delta_{1,2}^0 + {\rm i} \eta_{1,2})/\sqrt{2}$ . The general scalar potential of these two triplet Higgs fields$ \Delta_1 $ and$ \Delta_2 $ and the SM doublet Higgs fields Φ can be written as$ \begin{aligned}[b] V(\Delta_k ,\,\Phi) = &m_{\phi}^2 \,\Phi^{\dagger}\Phi + m_{kl}^2 {\rm{Tr}} (\Delta_k^{\dagger} \Delta_l) + \rho_{\phi} \,(\Phi^{\dagger}\Phi)^2 \\&+ \frac{1}{2} \rho_{kl} \,[{\rm{Tr}} (\Delta_k^{\dagger} \Delta_l)]^2 + \frac{1}{2}\rho^{\prime}_{kl} \,{\rm{Tr}} (\Delta_k^{\dagger} \Delta_l^{\dagger})\,{\rm{Tr}} (\Delta_k \Delta_l)\\ & + \rho_{\Delta} \,{\rm{Tr}} (\Delta_1^{\dagger} \Delta_1)\,{\rm{Tr}} (\Delta_2^{\dagger} \Delta_2) + \rho_{\Delta}^{\prime} \,{\rm{Tr}} (\Delta_1^{\dagger} \Delta_2)\,{\rm{Tr}} (\Delta_2^{\dagger} \Delta_1) \\&+ \sigma_{kl}\,\Phi^{\dagger}\Phi\,{\rm{Tr}}(\Delta_k^{\dagger} \Delta_l) + \sigma^{\prime}_{kl}\,\Phi^{\dagger}\Delta_k^{\dagger} \Delta_l \Phi\\ &+(\mu_k\,\Phi^{\dagger}\Delta_k \tilde{\Phi}+{\rm{h.c.}})\,, \end{aligned} $
(25) where
$ k,l = 1,2 $ . The VEVs of the Higgs fields are given by$ \begin{equation} \langle \Delta_{1,2} \rangle = \frac{1}{\sqrt{2}}\begin{pmatrix} 0 & 0\\ v_{1,2} & 0 \end{pmatrix}\,. \end{equation} $
(26) The orders of magnitude for the parameters in the potential are assumed:
$ \begin{aligned}[b] & m_{\phi}, m_{kl} \sim v; \qquad \rho_{\phi}, \rho_{kl}, \rho^{\prime}_{kl}, \rho_{\Delta}, \rho_{\Delta}^{\prime}, \sigma_{kl}, \sigma^{\prime}_{kl} \sim 1; \\\\ & |\mu_{k}| \ll 1; \qquad v_1,v_2 \ll v\,. \end{aligned} $
(27) For convenience, we define some matrices as
$ \begin{aligned}[b] &{\boldsymbol{m}} = (m_{kl})\,, \qquad {\boldsymbol{\rho}} = (\rho_{kl})\,, \qquad {\boldsymbol{\rho^{\prime}}} = (\rho^{\prime}_{kl})\,, \\ & {\boldsymbol{\sigma}} = (\sigma_{kl})\,, \qquad {\boldsymbol{\sigma^{\prime}}} = (\sigma^{\prime}_{kl})\,,\\ &{\boldsymbol{\mu}} = \dbinom{\mu_1}{\mu_2}\,, \qquad {\boldsymbol{V}} = \dbinom{v_1}{v_2}\,. \end{aligned} $
(28) With the approximation of
$ v_1,v_2 \ll v $ , the conditions for minimization of potential are$ \begin{aligned}[b] m_{\phi}^2 + \rho_{\phi}v^2 + \sqrt{2}{\boldsymbol{\mu}}^T {\boldsymbol{V}} = &0\,,\\ \left({\boldsymbol{m}}^2+\frac{v^2}{2}{\boldsymbol{\sigma}}\right){\boldsymbol{V}}+\frac{\sqrt{2}}{2}{\boldsymbol{\mu}}v^2 = &0\,. \end{aligned} $
(29) The mass squared matrices of CP-even Higgs
$ (\phi_R,\delta_1, \delta_2) $ , CP-odd Higgs$ (\phi_I,\eta_1, \eta_2) $ , singly-charged Higgs$ (\phi^+,\Delta_1^+, \Delta_2^+) $ , and double-charged Higgs$ (\Delta_1^{++}, \Delta_2^{++}) $ are given respectively by$ \begin{align} (M^0_{R})^2_{3 \times 3}& = \begin{pmatrix} m_{\phi}^2 + 3 \rho_{\phi} v^2 + \sqrt{2}\mu_1v_1 + \sqrt{2}\mu_2 v_2 & ({\boldsymbol{\sigma}} {\boldsymbol{V}} v+ \sqrt{2}{\boldsymbol{\mu}} v )^T \\ {\boldsymbol{\sigma}} {\boldsymbol{V}} v+ \sqrt{2}{\boldsymbol{\mu}} v & {\boldsymbol{m}}^2 + \dfrac{1}{2} {\boldsymbol{\sigma}} v^2\\ \end{pmatrix}\,, \end{align} $ (30) $ \begin{align} (M^0_{I})^2_{3 \times 3}& = \begin{pmatrix} m_{\phi}^2 + \rho_{\phi} v^2 - \sqrt{2}\mu_1v_1 - \sqrt{2}\mu_2v_2 & (\sqrt{2}{\boldsymbol{\mu}}v )^T \\ \sqrt{2}{\boldsymbol{\mu}}v & {\boldsymbol{m}}^2 + \dfrac{1}{2} {\boldsymbol{\sigma}} v^2\\ \end{pmatrix}\,, \end{align} $
(31) $ \begin{align} (M^+)^2_{3 \times 3}& = \begin{pmatrix} m_{\phi}^2 + \rho_{\phi} v^2& \left({\boldsymbol{\mu}} v - \dfrac{\sqrt{2}}{4}v {\boldsymbol{\sigma^{\prime}}}{\boldsymbol{V}}\right)^T \\ {\boldsymbol{\mu}} v - \dfrac{\sqrt{2}}{4}v {\boldsymbol{\sigma^{\prime}}}{\boldsymbol{V}} & {\boldsymbol{m}}^2 + \dfrac{1}{2} {\boldsymbol{\sigma}} v^2+ \dfrac{1}{4} {\boldsymbol{\sigma^{\prime}}}v^2\\ \end{pmatrix}\,, \end{align} $
(32) $ (M^{++})^2_{2 \times 2} = {\boldsymbol{m}}^2 + \frac{1}{2} {\boldsymbol{\sigma}} v^2+ \frac{1}{2} {\boldsymbol{\sigma^{\prime}}}v^2\,. $
(33) After diagonalization of these matrices, we obtain three Goldstone bosons and thirteen physical states. The transformations are parametrized by
$ \begin{aligned}[b]& \begin{pmatrix} h \\ H_1 \\ H_2 \end{pmatrix} = U_R \begin{pmatrix} \phi_R \\ \delta_1 \\ \delta_2 \end{pmatrix}\,,\qquad \begin{pmatrix} G^0 \\ A_1 \\ A_2 \end{pmatrix} = U_I \begin{pmatrix} \phi_I \\ \eta_1 \\ \eta_2 \end{pmatrix}\,,\\& \begin{pmatrix} G^{\pm} \\ H^{\pm}_1 \\ H^{\pm}_2 \end{pmatrix} = U_P \begin{pmatrix} \phi^{\pm} \\ \Delta_1^{\pm} \\ \Delta_2^{\pm} \end{pmatrix}\,,\qquad \begin{pmatrix} H^{\pm\pm}_1 \\ H^{\pm\pm}_2 \end{pmatrix} = U_{PP} \begin{pmatrix} \Delta^{\pm\pm}_1 \\ \Delta^{\pm\pm}_2 \end{pmatrix}\,. \end{aligned} $
(34) The Yukawa terms for triplet Higgses
$ \Delta_1, \Delta_2 $ coupled to leptons are given by$ \begin{aligned}[b] {\cal{L}}_{\Delta}^{\prime} =& \frac{1}{\sqrt{2}} (Y_{\Delta_1})_{ij} \overline{(L_L^i)^C}{\rm i}\tau_2\Delta_1 L_L^j \\&+ \frac{1}{\sqrt{2}} (Y_{\Delta_2})_{ij} \overline{(L_L^i)^C}{\rm i}\tau_2\Delta_2 L_L^j + \rm h.c.. \end{aligned} $
(35) When the triplet Higgs fields acquire VEVs, the neutrino mass matrix is generated,
$ \begin{equation} (M_{\nu})_{ij} = (Y_{\Delta_1})_{ij} v_1 + (Y_{\Delta_2})_{ij} v_2 \,. \end{equation} $
(36) Compared with the single triplet Higgs portal case, the coupling matrices
$ Y_{\Delta_1} $ and$ Y_{\Delta_2} $ are more flexible after fixing the VEVs of triplet Higgses. Gauge interactions of the SM Higgs and triplet Higgses are governed by$ \begin{aligned}[b] {\cal{L}}_{{\rm{gauge}}} =& (D_{\mu}\Phi)^{\dagger}D^{\mu}\Phi + {\rm{Tr}}\,\left[(D_{\mu}\Delta_1)^{\dagger}(D^{\mu}\Delta_1)\right]\\&+ {\rm{Tr}}\,\left[(D_{\mu}\Delta_2)^{\dagger}(D^{\mu}\Delta_2)\right]\,. \end{aligned} $
(37) The charged Higgs
$ H_{1,2}^{\pm} $ and$ H_{1,2}^{\pm\pm} $ dominantly decay into leptons in the scenario with the parameters set as follows: small VEVs of triplet fields$ v_{1,2} < 10^{-5} $ GeV, small mass splitting between two triplet Higgses$ m_{H_{i}^{\pm\pm}} \simeq m_{H_{i}^{\pm}} \simeq $ $ m_{H_{i}} \simeq m_{A_{i}}\; (i = 1,2) $ , and weak mixing between$ \Delta_1 $ and$ \Delta_2 $ .In addition to the triplet Higgs fields, we introduce a singlet real scalar field χ as a DM candidate. To ensure the candidate stability, the scalar field χ is assumed to be odd under
$ Z_2 $ symmetry, while all SM fields and the triplet Higgs fields are assumed to be even. The scalar potential relevant to DM χ is given by$ \begin{equation} V_{{\rm{DM}}} = m_{\chi}^2\chi^2 + \rho_{\chi}\chi^4 + \lambda_{\Phi}\Phi^{\dagger}\Phi \chi^2 + \lambda_{ij}\chi^2 {\rm{Tr}}(\Delta_i^{\dagger} \Delta_j)\,, \end{equation} $
(38) where
$ i, j = 1, 2 $ for the hybrid triplet Higgs portal model, and$ \lambda_{{ij}} = \lambda $ for the single triplet Higgs portal case. The coupling$ \lambda_{\Phi} $ is suppressed by DM direct detection experimental results. With the assumption of$ \lambda_{\Phi} \ll \lambda_{ij} $ , a pair of dark matter scalar fields χ dominantly annihilate into$ H_i^{\pm\pm}H_j^{\pm\pm} $ ,$ H_i^{\pm}H_j^{\pm} $ ,$ H_iH_j $ , and$ A_iA_j $ . -
It has been generally postulated that the Milky Way galaxy is located in a larger spherical DM halo. For the DM spatial distribution in the galaxy, we considered a generalized NFW profile [39,40] to describe,
$ \begin{equation} \rho_{{\rm{NFW}}}(r) = \rho_s\left( \frac{r_s}{r}\right)^{\gamma} \left(1+\frac{r}{r_s} \right)^{\gamma-3}\,, \end{equation} $
(39) where
$ \rho_s $ and$ r_s $ are typical density and typical radius respectively, whose values can fixed by the DM density at the location of the Sun (local density) and the total DM mass of the Milky Way, and γ is an undetermined parameter which mainly affects the DM density around the center of the Milky Way. According to the astrophysical observations to the rotation curve of the Milky Way, the local density of DM is roughly in the range of$ 0.2-0.6\ {\rm{GeV/cm^3}} $ [41–45], and the total DM mass contained in 60 kpc is estimated to be$ M_{60} = 4.7 \times 10^{11} M_{\odot} $ [46]. With the original NFW profile$ \gamma = 1 $ , we can fix the typical density and typical radius to be$ \rho_s = 0.184\, {\rm{GeV/cm^3}} $ and$ r_s = 24.42 $ kpc for local DM density$ \rho_{\odot} = 0.3\; {\rm{GeV/cm^3}} $ at$ r_{\odot} = 8.5 $ kpc. In our model, the DM particles can self-annihilate into charged leptonic final states and generate electron/positron flux. The propagation of the$ e^{\pm} $ number density energy spectrum$f\equiv {\rm d}N/{\rm d}E$ in the Milky Way can be described by the diffusion-loss equation,$ \begin{equation} \frac{\partial f}{\partial t} - \nabla({\cal{K}}(E, \vec{x}) \nabla f) - \frac{\partial}{\partial E}(b(E, \vec{x})f) = Q(E, \vec{x})\,, \end{equation} $
(40) where
$ {\cal{K}}(E, \vec{x}) $ is the diffusion coefficient function, which describes the propagation through the turbulent magnetic fields. Assuming the turbulent magnetic fields to be homogeneous and isotropic, the diffusion coefficient function can be parameterized as$ {\cal{K}}(E, \vec{x}) = {\cal{K}}_0 (E/ {\rm{GeV}})^{\delta} $ .$ b(E, \vec{x}) $ is the energy loss coefficient function, which describes energy loss due to inverse-Compton scattering of cosmic microwave background and starlight photons as well as synchrotron radiation. It can be parameterized as$ b(E, \vec{x}) = b_0 (E/{\rm{GeV}})^2 $ [47]. The electron/positron source term$ Q(E, \vec{x}) $ is given as$ \begin{equation} Q(E, \vec{x}) = \frac{1}{2}\left(\frac{\rho(\vec{x})}{M_{{\rm{DM}}}}\right)^2 \langle \sigma v \rangle \frac{{\rm d}N}{{\rm d}E}\,, \end{equation} $
(41) where
$ \rho(\vec{x}) $ is the DM density distribution,$ M_{{\rm{DM}}} $ is the mass of DM, and$ \langle \sigma v \rangle $ is the thermal averaged cross section. Considering a steady-case,$ \partial f/ \partial t = 0 $ , and introducing Green's function formalism, the number density energy spectrum of a given point in the Galaxy can be written as$ \begin{equation} f(E, \vec{x}) = \int {\rm d} \vec{x}_s \int {\rm d} E_s \, G(E, \vec{x}; E_s, \vec{x}_s) Q(E, \vec{x})\,, \end{equation} $
(42) where Green’s function
$ G(E, \vec{x}; E_s, \vec{x}_s) $ stands for the probability of injected electron or positron at$ x_s $ with the energy$ E_s $ to reach a location x with the degraded energy E. Green's function satisfies the following equation,$ \begin{aligned}[b]& {\cal{K}}(E) \nabla^2 G(E, \vec{x}; E_s, \vec{x}_s) + \frac{\partial}{\partial E}(b(E) G(E, \vec{x}; E_s, \vec{x}_s))\\=& \delta(E-E_s) \, \delta(\vec{x} - \vec{x}_s)\,. \end{aligned} $
(43) The equation has a general solution without boundary conditions [48],
$ \begin{equation} G(E, \vec{x}; E_s, \vec{x}_s) = \frac{1}{b(E)(4\pi \lambda^2)^{3/2}}\, {\rm{exp}} \left(-\frac{(\vec{x}-\vec{x}_s)^2}{\lambda^2}\right)\,, \end{equation} $
(44) where the diffusion length λ is given by
$ \begin{equation} \lambda = \theta(E_s - E) \sqrt{\frac{4{\cal{K}}_0(E^{\delta-1}-E_s^{\delta-1})}{b_0(1-\delta)}}\,. \end{equation} $
(45) We solve Eq. (43) numerically in the shape of a flat cylinder with height
$ 2L $ in the z direction and radius$ R = 20\,{\rm{kpc}} $ in the r direction, while considering the number density energy spectrum f vanished on the surface of the cylinder. The comparison of solving with and without boundary conditions is described in Ref. [49]. With the number density energy spectrum$ f(E,\vec{x}) $ , the electron and positron flux is given as$ \begin{equation} \Phi(E,\vec{x}) = \frac{v(E)}{4 \pi}f(E,\vec{x})\,, \end{equation} $
(46) where
$ v(E) $ is the velocity of electron and positron with energy E and is set as c because of$ E \gg m_e $ . -
According to the latest analysis about cosmic ray positrons collected by the Alpha Magnetic Spectrometer (AMS-02) [11] on the International Space Station, a significant excess starting from
$ 25\, {\rm{GeV}} $ to$ 1\,{\rm{TeV}} $ is reported, which possibly originates from DM particles annihilation. We choose the same background formula as the latest report by the AMS Collaboration [11],$ \begin{equation} \Phi_{e^+}^{{\rm{bkg}}}(E) = \frac{E^2}{\hat{E}^2}\left[ C_d\left( \frac{\hat{E}}{E_1} \right)^{\gamma_d}+ C_s \left( \frac{\hat{E}}{E_2} \right)^{\gamma_s} {\rm{exp}}\left(-\frac{\hat{E}}{E_s} \right) \right]\,, \end{equation} $
(47) where
$ \hat{E} = E+\varphi_{e^+} $ , with the effective solar potential$ \varphi_{e^+} $ accounting for the solar effects. We implement the model in FeynRules [50] to generate the model files and put it in micrOMEGAs 5.0.8 [51] to generate the positron flux from DM annihilation. For the propagation of flux, we set the diffusion coefficient$ {\cal{K}}_0 = 0.0112\, {\rm{kpc}}^2/{\rm{Myr}} $ ,$ \delta = 0.7 $ , and the energy loss coefficient$ b_0 = 10^{-16}\, {\rm{s}}^{-1} $ .With the DM mass (
$ m_{{\rm{DM}}} $ ) and triplet Higgs mass ($ m_{\Delta} $ ) fixed, the DM thermal relic density only depends on the coupling$ \lambda_{ij} $ in Eq. (38). In other words, the couplings$ \lambda_{ij} $ can be fixed by the observed DM thermal relic of$ \Omega h^2 = 0.12 $ [33], and then, the thermal averaged cross sections in Eq. (41) are also fixed. However, the thermal averaged cross sections obtained from observed DM thermal relic$ \langle\sigma v \rangle_0 $ are usually much smaller than the required value$ \langle\sigma v \rangle $ to account for the excess. In order to fit the AMS-02 data, we need enhancement to the DM annihilation cross sections. The enhancement is usually parameterized by a so-called boost factor (BF)$\kappa_{{\rm{BF}}} = $ $ \langle\sigma v\rangle / \langle\sigma v \rangle_0$ . The boost factor may originate from small-scale inhomogeneities of the DM density distribution [27,52] or Sommerfeld/Breit-Wigner enhancement of DM annihilation [53–59].To describe the goodness of the fit, we introduce the chi-square function
$ \chi^2 $ , which defined as$ \begin{equation} \chi^2 = \sum\limits_i \frac{\left(f^{{\rm{th}}}_i - f^{{\rm{exp}}}_i \right)^2}{\delta f_i^2}\,, \end{equation} $
(48) where
$ f^{{\rm{th}}}_i $ are the theoretical predictions,$ f^{{\rm{exp}}}_i $ are the central values of the experimental data, and$ \delta f_i $ are combined errors of theoretical and experimental data. For the fit to AMS-02 data, we have only considered the experimental errors and used all$ E > 1 {\rm{GeV}} $ AMS-02 data (71 data points). In our model, a pair of DM can annihilate to a pair of singly charged triplet Higgs$ H^{\pm} $ or doubly charged triplet Higgs$ H^{\pm\pm} $ , and these particles will further decay into two or four leptonic final states. The Yukawa coupling matrices are fixed by the neutrino mass and oscillation data, and the matrix elements are given by Eqs. (20)–(22) for different neutrino mass hierarchies. As shown in Table 2, the doubly charged Higgs$ H^{\pm\pm} $ dominantly decays into μ and τ leptons for the normal neutrino mass hierarchy, while for the inverted hierarchy, it dominantly decays to electrons, and for the quasi-degenerate case, the three flavors are almost democratic. When the difference value of DM mass and triplets mass is small ($ m_{{\rm{DM}}} \simeq m_{\Delta} $ ), the leptons generated from triplet Higgs decays have a roughly monochromatic energy spectrum. When$ m_{{\rm{DM}}} \gg m_{\Delta} $ , since the triplet Higgses are highly boosted, the leptons will have a continuous energy spectrum. The generated muons and taus will further decay into positrons and contribute to the cosmic ray positron flux. The energy spectrum of the positron directly produced by the triplet Higgs is sharp and drop around energy$ E = m_{{\rm{DM}}}/2 $ , while for the positrons generated from the muon- and tau-cascade decays, the energy spectrum will be softer, and the peak is lower.Hierarchy $H^{\pm\pm}$ decays$H^{\pm}$ decays$ee$ $e\mu$ $e\tau$ $\mu \mu$ $\mu \tau$ $\tau \tau$ $e\nu$ $\mu \nu$ $\tau \nu$ NH 0.49% 0.67% 4.36% 37.37% 34.90% 22.21% 3.00% 55.14% 41.86% IH 47.53% 1.32% 0.84% 9.84% 23.74% 16.73% 49.27% 50.72% 0% DH 25.26% 0.32% 0.28% 37.83% 1.88% 34.43% 24.79% 25.51% 49.70% Table 2. Branching ratios of charged Higgs leptonic decays for different neutrino mass hierarchies.
For the single triplet Higgs portal model, we scan over the mass range from 500 GeV to 2 TeV for both DM and triplet Higgs and assume the triplet Higgs mass splitting is negligible (
$ m_{\Delta} \simeq m_{H} \simeq m_{A}\simeq m_{H^{\pm}} \simeq m_{H^{\pm\pm}} $ ). As shown in Fig. 1(a), we present the favored region for$ \chi^2/{\rm{d.o.f.}} $ in the plane of the DM mass$ m_{{\rm{DM}}} $ versus the triplet Higgs mass$ m_{\Delta} $ . In the right panel of Fig. 1(b), we plot the mass difference$ (m_{{\rm{DM}}}-m_{\Delta}) $ against the triplet Higgs mass for the NH scenario for different$ \chi^2/{\rm{d.o.f.}} $ values. The black data points in both Fig. 1(a) and (b) are corresponding to the best fit data points obtained from$ \chi^2/{\rm{d.o.f.}} $ analysis for the NH case. Similar plots are generated for the DH case and depicted in Fig. 1(c-d) along with the best fit data points. A comparison between the NH and DH scenarios reveals that the NH case has a larger range of DM mass of approximately (1–1.8) TeV; for the chosen range of$ \chi^2/{\rm{d.o.f.}}\in (1.0-2.5) $ , the corresponding range is restricted into$ (1-1.4) $ TeV for the DH case. Figs. 1(b) and 1(d) show that the mass splitting between DM mass and the triplet Higgs in the case of NH is larger than that in the DH case. Various parameters for the best fit to AMS-02 data with single triplet Higgs portal DM model are listed in Table 3 along with the contribution ratios from different DM annihilation channels. It is to be noted for the IH case that$ \chi^2/{\rm{d.o.f.}} $ is always found to be larger than 5 for the chosen range of DM and triplet Higgs mass. Therefore, we do not present the results for the IH scenario.Figure 1. (color online) Favored region of
$ \chi^2/{\rm{d.o.f.}} $ for the single triplet portal case. The upper and lower panel are dedicated to the normal and degenerate neutrino mass hierarchies, respectively.Best fit $m_{\text{DM}}$ /GeV$m_{\Delta}$ /GeV$\rm{Br}(H_1 H_1/A_1 A_1)$ $\rm{Br}(H^+_1 H^-_1/H^{++}_1 H^{--}_1)$ $\lambda_{\Delta}$ BF NH 1340 660 $16\,$ %$34\, $ %0.18 1758 DH 980 600 $17\,$ %$33\, $ %0.12 608 Table 3. Best fit data points to AMS-02 positron excess in the single triplet DM model.
In Fig. 2, we plot positron flux spectrum using the best fit data points obtained for the NH and DH cases (Table 3). One can notice that the best fit data do not provide a good fit to the positron flux tail at the high energy end. The reason is that, in the single triplet Higgs portal DM model, the flavor structure of the leptonic final states generated by the DM annihilation is fixed by the neutrino oscillation data. The positron flux spectrum range from 10 GeV to 300 GeV indicates that the positrons are dominantly generated by the muon- and/or tau-cascade decays but not for the case when the energy is beyond 300 GeV. This is also the reason that there are more favored ranges for the NH case compared with the IH case. In the IH case, the charged triplet Higgses dominantly decay to electrons (as demonstrated in Table 2). In the following, we investigate the hybrid triplet Higgs portal DM, and we find that the fit gets much better, especially for the IH case.
Figure 2. (color online) The best-fit of the positron flux observed by the AMS-02 in the single triplet portal DM model.
We now fit the AMS-02 data by the hybrid triplet portal DM, with a light triplet Higgs and a heavy triplet Higgs final states as the DM main annihilation channels. We find that the positron flux can be best fit when the light triplet Higgs mainly decay to muons and taus, whereas the heavy one mainly decay to electrons. In order to generate the high energy region positrons by the heavier triplet Higgs decays, we open the two annihilation channels
$ \chi \chi \to \Delta_1 \Delta_1 $ and$ \chi \chi \to \Delta_1 \Delta_2 $ , while the channel$ \chi \chi \to \Delta_2 \Delta_2 $ is forbidden by setting$ m_{\Delta_1}+m_{\Delta_2} < 2m_{{\rm{DM}}} < 2m_{\Delta_2} $ . Without loss of generality, as a benchmark, we set$2m_{{\rm{DM}}}- $ $ (m_{\Delta_1}+m_{\Delta_2}) = 10$ GeV. We scan over the light triplet Higgs mass$ m_{\Delta_1} $ ranging from 500 GeV to 1.5 TeV while the heavier triplet Higgs mass is varied within the range from 1.5 TeV to 3.5 TeV. In Fig. 3 (a-c), we plot the allowed range of$ m_{\Delta_1} $ versus$ m_{\Delta_2} $ parameter space, which fits the AMS-02 data with$ 0.8\leq \chi^2/{\rm{d.o.f.}} \leq 1 $ for the NH, IH, and DH scenarios, respectively.Figure 3. (color online) Favored region with
$ \chi^2/{\rm{d.o.f.}} $ for the hybrid triplet portal DM, for the case of the mass difference taken as$ 2m_{{\rm{DM}}}-(m_{\Delta_1}+m_{\Delta_2}) = 10\ {\rm{GeV}} $ .The favored regions for the masses of DM and the triplet Higgses are shown in Fig. 3, and the best fit data sets are given in Table 4, with the corresponding positron flux results shown in Fig. 4. One intriguing feature of the hybrid triplet portal DM is that it can provide a pretty good fit to the AMS-02 data for the IH case, unlike the single triplet portal DM model discussed earlier. Apart from the above, we also find that the hybrid triplet portal DM can provide better fit for the excess at high energy region for the energy above 300 GeV. This unique feature of the hybrid triplet model is due to that the heavy triplet Higgses produce hard positrons, while the light triplet Higgses generate soft positrons. Finally, we make a direct comparison between the single triplet Higgs portal DM model and the hybrid triplet Higgs portal DM model. The
$ \chi^{2}/{\rm{d.o.f.}} $ of the best fit points in varied scenarios are listed in Table 5, which clearly indicates that the hybrid triplet Higgs portal model provides a much better fit to the AMS-02 data. Especially, the fit is significantly improved for the neutrino mass IH scenario.Best fit $m_{\Delta1}$ /GeV$m_{\Delta2}$ /GeV$\rm{Br}(H_1 H_1/A_1 A_1)$ $\rm{Br}(H^+_1 H^-_1/H^{++}_1 H^{--}_1)$ $\rm{Br}(H_1 H_2/A_1 A_2)$ $\rm{Br}(H^+_1 H^-_2/H^{++}_1 H^{--}_2)$ $\lambda_{11}$ $\lambda_{12}$ BF NH 780 2720 $12\,$ %$25\, $ %$4\, $ %$9\, $ %0.10 0.10 2100 IH 650 2370 $15\,$ %$30\, $ %$2\, $ %$3\, $ %0.14 0.09 1126 DH 590 2210 $13\,$ %$29\, $ %$3\, $ %$5\, $ %0.12 0.10 1267 Table 4. Best fit data points to AMS-02 positron excess for hybrid triplet Higgs portal DM model. The "Branch ratios" indicate the branch ratios of the DM annihilation channels. "BF" denotes the needed enhancement factors.
Figure 4. (color online) The best fit to the positron flux observed by the AMS-02 in the hybrid triplet portal DM model.
Normal ordering Inverted ordering Degenerate ordering single triplet portal 1.01 8.12 1.07 hybrid triplet portal 0.82 0.84 0.82 Table 5. The
$ \chi^{2}/{\rm{d.o.f.}} $ of the best fit for different neutrino mass hierarchies in the single and hybrid triplet portal DM scenarios.Note that the DM annihilation also produces gamma-rays. We compared the gamma-rays produced by the DM annihilation with the isotropic diffuse gamma-ray background (IGRB) observed by the Fermi-LAT experiment [60], and we find that the predicted gamma-rays are slightly constrained by the Fermi-LAT observation in the energy region beyond hundreds of GeV. An earlier study about the single triplet Higgs portal model reached a similar conclusion [28]. A feasible way to release the inconsistency between the DM origin of cosmic positron excesses and the IGRB constraints is to introduce a dark matter disk instead of a simple dark matter halo as the dark matter density distribution [61]. More studies on the compatibility between the DM origin of cosmic positron excesses and gamma-ray limits are reported in Refs. [62–66].
-
In this section, we attempt to explain the cosmic ray electron and positron excesses observed by the DAMPE satellite. Different from AMS-02, DAMPE does not distinguish the charge of electron events and collect the sum of electron and positron events. The DAMPE data have been considered to be the electron and positron flux associated with a smoothly broken power law spectrum [12]. The signals have also been considered as an excess in the range (0.6–1.5) TeV from DM sources with single broken power-law background and double broken power-law background [21, 23, 67] or excess in the range of (0.1–1.5) TeV with single power law background [20]. Here, we assume that there exists a smooth excess in the energy range (0.1–1.5) TeV along with an excess peak at 1.5 TeV. We consider that the continuous excess of electron and positron flux is attributed to the DM annihilation within the Milky Way (MW), and the sharp peak at 1.5 TeV originates mainly from a nearby DM sub-halo. Meanwhile, we take a simple single power law background to fit the background for
$ E<100 $ GeV and$ E>1.5 $ TeV, and the background formula of the electron and positron flux is expressed as$ \begin{equation} \Phi_{e^{\pm}}^{{\rm{bkg}}} = C \left(\frac{E}{{\rm{GeV}}}\right)^{-\gamma}, \end{equation} $
(49) where we set the free parameter:
$C = 450\ {\rm (GeV\cdot m^2\cdot s\cdot $ $ \rm sr)^{-1} }$ and$ \gamma = 3.26 $ by fitting the first two points and the four points between 1.5 TeV and 2.5 TeV of the DAMPE data.To explain the peak around 1.5 TeV in the DAMPE data, the heavy triplet Higgs mass
$ m_{\Delta_2} $ is required to be set around 3 TeV, and the heavy charged triplet Higgs mainly decays to the electron flavor. From the results of fitting AMS-02 data, we learn that the heavy triplet Higgs can be chosen as high as 3 TeV only in the normal neutrino mass hierarchy scenario; therefore, we choose the NH case to fit the DAMPE electron and position excess in the hybrid triplet Higgs portal model. We scan over the light triplet Higgs mass$ m_{\Delta_1} $ ranging from 500 GeV to 1.5 TeV and the heavy triplet Higgs mass$ m_{\Delta_2} $ ranging from 1.5 TeV to 4 TeV. We obtain the allowed range of$ m_{\Delta_1} $ versus$ m_{\Delta_2} $ that fits both the AMS-02 data with$ 0.8 \leq \chi^2/ $ $ {\rm{d.o.f.}} \leq 1 $ and the DAMPE data with$ 0.2 \leq \chi^2/{\rm{d.o.f.}} \leq 0.3 $ for the NH case, as shown in Fig. 5. For the dark matter sub-halo profile, we adopt the NFW density profile with$ \rho_s = 100\ {\rm{GeV/cm^3}} $ ,$ r_s = 0.1 $ kpc, and$ \gamma = 0.5 $ , and the distance between the Sun and the center of the sub-halo is set as$ d_s = 0.2 $ kpc.Figure 5. (color online) (a): The allowed parameter space of
$ m_{\Delta_1} $ versus$ m_{\Delta_2} $ that fits both the AMS-02 data with$ 0.8 \leq \chi^2/{\rm{d.o.f.}} \leq 1 $ and the DAMPE data with$ 0.2 \leq \chi^2/{\rm{d.o.f.}} \leq 0.3 $ for the NH case. (b): The best fit to the DAMPE CRE spectrum with$ m_{\Delta_1} = 810\ {\rm{GeV}} $ ,$ m_{\Delta_2} = 3000\ {\rm{GeV}} $ , and$ \chi^2/{\rm{d.o.f.}} = 0.2 $ . The blue line indicates the contribution from DM annihilation in the Milky Way DM halo, and the green line includes the contribution of the nearby sub-halo. -
Here, we present the model parameter values corresponding to the best fit points (BFPs) in Table 3 and Table 4.
BFP 1: Input parameters for the NH case in Table 3:
$ \begin{aligned}[b] &m_{H}, m_{A}, m_{H^{\pm}}, m_{H^{\pm\pm}} = 660\,{\rm{GeV}}, \quad m_{{\rm{DM}}} = 1340\,{\rm{GeV}},\\ &\lambda_{\Delta} = 0.18, \quad \sin \alpha = 0,\quad v_{\Delta} = 10^{-6}\, {\rm{GeV}}. \end{aligned} \tag{A1}$
BFP 2: Input parameters for the DH case in Table 3:
$ \begin{aligned}[b] &m_{H}, m_{A}, m_{H^{\pm}}, m_{H^{\pm\pm}} = 600\,{\rm{GeV}}, \quad m_{{\rm{DM}}} = 980\,{\rm{GeV}},\\ &\lambda_{\Delta} = 0.12, \quad \sin \alpha = 0,\quad v_{\Delta} = 10^{-6}\, {\rm{GeV}}. \end{aligned} \tag{A2}$
BFP 3: Input parameters for the NH case in Table 4:
$ \begin{aligned}[b] &m_{12} = -120\,{\rm{GeV}}, \quad {\boldsymbol{\rho}} = {\boldsymbol{\rho}}^{\prime} = {\boldsymbol{\sigma}} = \begin{pmatrix} 0.5 & 0.0001 \\ 0.0001 & 0.5 \end{pmatrix},\\& {\boldsymbol{\sigma}}^{\prime} = \begin{pmatrix} 0.5 & 0.0001 \\ 0.0001 & 0.5 \end{pmatrix},\quad\rho_{\Delta} = \rho_{\Delta}^{\prime} = 0.1, \\ & {\boldsymbol{\mu}} = \begin{pmatrix} -9.5 \times 10^{-6}\,{\rm{GeV}} \\ -1.2 \times 10^{-4}\,{\rm{GeV}} \end{pmatrix}, \quad {\boldsymbol{V}} = \begin{pmatrix} 10^{-6} \,{\rm{GeV}} \\ 10^{-6} \,{\rm{GeV}} \end{pmatrix}, \\ &m_{\chi} = 1770\,{\rm{GeV}}, \quad \lambda_{\Phi} = 0, \quad \lambda_{11} = 0.10, \quad \lambda_{12} = 0.10\,. \end{aligned}\tag{A3} $
With these parameter values, the particle masses are determined as
$ \begin{aligned}[b] &m_{H_1}, m_{A_1}, m_{H_1^{\pm}}, m_{H_1^{\pm\pm}} = 780\,{\rm{GeV}}, \\& m_{H_2}, m_{A_2}, m_{H_2^{\pm}}, m_{H_2^{\pm\pm}} = 2720\,{\rm{GeV}}, \\& m_{{\rm{DM}}} = 1760\,{\rm{GeV}}. \end{aligned} \tag{A4}$
BFP 4: Input parameters for the IH case in Table 4:
$ \begin{aligned}[b] &m_{12} = -120\,{\rm{GeV}}, \quad {\boldsymbol{\rho}} = {\boldsymbol{\rho}}^{\prime} = {\boldsymbol{\sigma}} = \begin{pmatrix} 0.5 & 0.0001 \\ 0.0001 & 0.5 \end{pmatrix}, \\& {\boldsymbol{\sigma}}^{\prime} = \begin{pmatrix} 0.5 & 0.0001 \\ 0.0001 & 0.5 \end{pmatrix},\\ &\rho_{\Delta} = \rho_{\Delta}^{\prime} = 0.1, \\& {\boldsymbol{\mu}} = \begin{pmatrix} -6.5 \times 10^{-6}\,{\rm{GeV}} \\ -9.2 \times 10^{-5}\,{\rm{GeV}} \end{pmatrix}, \quad {\boldsymbol{V}} = \begin{pmatrix} 10^{-6}\,{\rm{GeV}} \\ 10^{-6}\,{\rm{GeV}} \end{pmatrix}, \\ &m_{\chi} = 1520\,{\rm{GeV}}, \quad \lambda_{\Phi} = 0, \quad \lambda_{11} = 0.14, \quad \lambda_{12} = 0.09\,. \end{aligned}\tag{A5} $
With these parameter values, the particle masses are determined as
$ \begin{aligned}[b] &m_{H_1}, m_{A_1}, m_{H_1^{\pm}}, m_{H_1^{\pm\pm}} = 650\,{\rm{GeV}}, \\& m_{H_2}, m_{A_2}, m_{H_2^{\pm}}, m_{H_2^{\pm\pm}} = 2370\,{\rm{GeV}}, \\& m_{{\rm{DM}}} = 1520\,{\rm{GeV}}. \end{aligned} \tag{A6}$
BFP 5: Input parameters for the DH case in Table 4:
$ \begin{aligned}[b] &m_{12} = -120\,{\rm{GeV}}, \quad {\boldsymbol{\rho}} = {\boldsymbol{\rho}}^{\prime} = {\boldsymbol{\sigma}} = \begin{pmatrix} 0.5 & 0.0001 \\ 0.0001 & 0.5 \end{pmatrix}, \\& {\boldsymbol{\sigma}}^{\prime} = \begin{pmatrix} 0.5 & 0.0001 \\ 0.0001 & 0.5 \end{pmatrix}, \quad \rho_{\Delta} = \rho_{\Delta}^{\prime} = 0.1, \end{aligned} $
$ \begin{aligned}[b] &{\boldsymbol{\mu}} = \begin{pmatrix} -5.3 \times 10^{-6}\,{\rm{GeV}} \\ -8.0 \times 10^{-5}\,{\rm{GeV}} \end{pmatrix}, \quad {\boldsymbol{V}} = \begin{pmatrix} 10^{-6}\,{\rm{GeV}} \\ 10^{-6}\,{\rm{GeV}} \end{pmatrix}, \\ &m_{\chi} = 1410\,{\rm{GeV}}, \quad \lambda_{\Phi} = 0, \quad \lambda_{11} = 0.12, \quad \lambda_{12} = 0.10\,. \end{aligned}\tag{A7} $
With these parameter values, the particle masses are determined as
$ \begin{aligned}[b] &m_{H_1}, m_{A_1}, m_{H_1^{\pm}}, m_{H_1^{\pm\pm}} = 590\,{\rm{GeV}}, \\& m_{H_2}, m_{A_2}, m_{H_2^{\pm}}, m_{H_2^{\pm\pm}} = 2210\,{\rm{GeV}}, \\& m_{{\rm{DM}}} = 1410\,{\rm{GeV}}. \end{aligned}\tag{A8} $
Confronting cosmic ray electron and positron excesses with hybrid triplet Higgs portal dark matter
- Received Date: 2021-12-01
- Available Online: 2022-06-15
Abstract: We perform a detailed study of scalar dark matter with triplet Higgs extensions of the Standard Model in order to explain the cosmic ray electron and positron excesses reported by AMS-02 and DAMPE. A detailed analysis of the AMS-02 positron excess reveals that for different orderings (normal, inverted, and quasi-degenerate) of neutrino mass, the hybrid triplet Higgs portal framework is more favored with respect to the single triplet Higgs portal for TeV scale dark matter. We also show that the resonant peak and continuous excess in DAMPE cosmic ray data can be well explained with the hybrid triplet Higgs portal dark matter when a dark matter sub-halo nearby is taken into account.