Higgs-Singlet Scalar Portals

A real gauge singlet communicates renormalizably with the Standard Model through $H^\dagger H$, but a portal coupling alone does not define a viable model. The scalar potential must select a declared vacuum, its quartic form must be bounded below, the physical Hessian must be positive, mixing must be diagonalized consistently, and any decay or decoupling claim must include thresholds, widths, and the origin of the heavy mass.

Required background. Consistency Checklist for Standard Model Extensions supplies the structural and evidence gates.

Helpful background. What an Interacting Lagrangian Does and Does Not Specify separates a formal potential from a constructed interacting theory. Nondecoupling Effects and Matching Validation supplies the heavy-mass tests.

The minimal real-singlet potential

For a real scalar $S$ with an exact $S\mapsto-S$ symmetry, a convenient normalization is

$$V(H,S)= -\mu_h^2H^\dagger H+\lambda_h(H^\dagger H)^2 -\frac{\mu_s^2}{2}S^2+\frac{\lambda_s}{4}S^4 +\frac{\lambda_{hs}}{2}H^\dagger H\,S^2.$$

This is a bounded model choice, not the most general singlet potential: without the $\mathbb Z_2$ symmetry, linear and cubic singlet terms and $H^\dagger H,S$ are also allowed. An omitted term is technically meaningful only when a symmetry or renormalization condition explains its absence. The special role of $H^\dagger H$ as a low-dimension singlet portal was emphasized by Patt and Wilczek 2006, pp. 1–3.

Write

$$H=\frac{1}{\sqrt2}\begin{pmatrix}0\\v+h\end{pmatrix}, \qquad S=x+s.$$

For the branch $v\ne0$, $x\ne0$, stationarity gives

$$\mu_h^2=\lambda_hv^2+\frac{\lambda_{hs}}{2}x^2, \qquad \mu_s^2=\lambda_sx^2+\frac{\lambda_{hs}}{2}v^2.$$

The CP-even mass matrix in the $(h,s)$ basis is

$$\mathcal M^2= \begin{pmatrix} 2\lambda_hv^2&\lambda_{hs}vx\\ \lambda_{hs}vx&2\lambda_sx^2 \end{pmatrix}.$$

Define $(h,s)^{\mathsf T}=R(\theta)(h_1,h_2)^{\mathsf T}$ with $R(\theta)=\bigl(\begin{smallmatrix}\cos\theta&-\sin\theta\\ \sin\theta&\cos\theta\end{smallmatrix}\bigr)$ and order $m_1^2\le m_2^2$. Then

$$m_{1,2}^2=\frac{\mathcal M_{hh}^2+\mathcal M_{ss}^2}{2} \mp\frac12\sqrt{(\mathcal M_{hh}^2-\mathcal M_{ss}^2)^2+4(\mathcal M_{hs}^2)^2},$$ $$\tan2\theta=\frac{2\mathcal M_{hs}^2}{\mathcal M_{hh}^2-\mathcal M_{ss}^2},$$

up to the declared eigenvector signs. Because only $h$ carries Standard Model gauge and Yukawa couplings, $h_1$ and $h_2$ inherit those couplings with factors $\cos\theta$ and $\sin\theta$ in this convention.

Boundedness, extrema, and the selected vacuum

At large fields the quartic form is

$$V_4=\frac{\lambda_h}{4}h^4 +\frac{\lambda_s}{4}s^4 +\frac{\lambda_{hs}}{4}h^2s^2.$$

The exact copositivity conditions are

$$\lambda_h\ge0, \qquad \lambda_s\ge0, \qquad \lambda_{hs}\ge-2\sqrt{\lambda_h\lambda_s}.$$

If either self-coupling vanishes, the last condition requires $\lambda_{hs}\ge0$; saturation may leave a flat direction. Strict inequalities put the quartic in its positive interior. These conditions control asymptotic directions only. One must separately enumerate the symmetric point, the $v$-only and $x$-only extrema, and the mixed extremum, test their physical Hessians, and compare their potential values. A positive Hessian establishes a local minimum, not the global vacuum.

The exact fixture

$$(\lambda_h,\lambda_s,\lambda_{hs})=(1,4,-2)$$

lies inside the stable domain because $-2>-4$, whereas $\lambda_{hs}=-5$ fails. Independently, the mass matrix

$$\mathcal M^2=\begin{pmatrix}2&1\\1&5\end{pmatrix}$$

has positive minors $2$ and $9$, eigenvalues $(7\mp\sqrt{13})/2$, and $|\tan2\theta|=2/3$. Sending the off-diagonal entry to zero continuously removes mixing. Exact matrix arithmetic reproduces these checks.

Zero-temperature boundedness does not decide a thermal history, nucleation rate, or cosmological abundance. Metastability additionally requires a tunneling calculation with a declared effective action and gauge treatment. Those questions are external to this minimal vacuum calculation.

Widths and observable interfaces

On the unbroken $\mathbb Z_2$ branch $x=0$, there is no scalar mixing and the lightest $S$ is stable under the declared symmetry. Its field-dependent mass is

$$m_S^2=-\mu_s^2+\frac{\lambda_{hs}}{2}v^2.$$

If $m_h>2m_S$, the cubic interaction is $\mathcal L\supset-(\lambda_{hs}v/2)hS^2$ and the tree-level partial width is

$$\Gamma(h\to SS)= \frac{\lambda_{hs}^2v^2}{32\pi m_h} \sqrt{1-\frac{4m_S^2}{m_h^2}}.$$

This formula assumes canonically normalized stable final states and an on-shell parent. A likelihood application must also specify the total-width scheme, competing channels, higher-order corrections, and whether the narrow-width approximation is justified.

On the broken branch, both mass eigenstates couple to visible currents. Production and decay amplitudes contain the appropriate mixing factors, but overlapping resonances require the coherent propagator matrix rather than two independently added rates. Current coupling fits, invisible-width bounds, and search exclusions require dated evidence and are not supplied by the timeless mixing formulas.

Decoupling and matching

When $\mathcal M_{ss}^2\equiv M_s^2\gg v^2$ with $\lambda_{hs}vx$ held perturbative,

$$\theta=\frac{\lambda_{hs}vx}{\mathcal M_{hh}^2-M_s^2} +O\!\left(\frac{v^3}{M_s^3}\right),$$

so visible-coupling shifts scale as $\theta^2$. Integrating out the heavy scalar also changes the low-energy Higgs potential and generates higher-dimensional operators. The matching must be performed before declaring decoupling: if $M_s$ is increased by taking $x$ or a quartic large, a threshold correction can remain finite and perturbative unitarity can fail. Structural singlet matching and vacuum conditions are reviewed in Robens and Stefaniak 2015, §§2–3.

Checks and failure modes

• Verify boundedness independently of the Hessian and compare every stationary branch.
• Check $\operatorname{tr}\mathcal M^2=m_1^2+m_2^2$ and $\det\mathcal M^2=m_1^2m_2^2$.
• Recover zero mixing at $\lambda_{hs}vx=0$ and the matched low-energy theory in the declared heavy-mass limit.
• Do not call $x$ or $v$ an observable without specifying gauge and renormalization conventions; pole masses and $S$-matrix elements are the physical outputs.
• Do not combine two resonant rates without their interference, or import a present-day exclusion without its likelihood and provenance.

Additional electroweak multiplets belong to Extended Higgs Sectors. Generic matching belongs to Integrating Out Heavy Fields, thermal histories to the thermal volume, and versioned phenomenology to Effective Field Theory and Tests of the Standard Model.

References

• Patt, Brian, and Frank Wilczek. “Higgs-Field Portal into Hidden Sectors.” arXiv:hep-ph/0605188 (2006). arXiv.
• Robens, Tania, and Tim Stefaniak. “Status of the Higgs Singlet Extension of the Standard Model after LHC Run 1.” European Physical Journal C 75 (2015): 104. DOI.