Higgs Self-Interactions and the Scalar Potential

For the renormalizable one-doublet potential, the tree-level Higgs mass fixes both self-interactions:

$$\lambda_3^{(0)}=\frac{3m_h^2}{v}, \qquad \lambda_4^{(0)}=\frac{3m_h^2}{v^2},$$

when $\mathcal L_{\rm int}=-\lambda_3h^3/3!-\lambda_4h^4/4!$. Beyond tree level these symbols are not observables by themselves: their values depend on the renormalization, tadpole, momentum, and field-normalization prescriptions, while complete pole residues and scattering amplitudes are physical.

Required background. The Higgs doublet and electroweak symmetry breaking supplies the scalar potential and vacuum orbit. Renormalization conditions, schemes, and finite parts supplies the distinction between a renormalized parameter and an observable.

Helpful background. Unstable-particle observables and resonance approximations supplies pole and line-shape definitions used for a physical Higgs interface.

Tree-level potential and vertex normalization

Take

$$V(H)=-\mu^2H^\dagger H+\lambda(H^\dagger H)^2, \qquad H\sim(\mathbf1,\mathbf2)_{1/2}, \qquad Q=T_3+Y,$$

with $\mu^2>0$, $\lambda>0$, and $D_\mu H=(\partial_\mu-igT^aW_\mu^a-ig'B_\mu/2)H$. Along the radial direction $H=(0,(v+h)/\sqrt2)^{\mathsf T}$, stationarity gives $\mu^2=\lambda v^2$. Expanding around that point yields

$$V(h)=V(v) +\frac12(2\lambda v^2)h^2 +\lambda vh^3+\frac{\lambda}{4}h^4.$$

Thus

$$m_h^2=2\lambda v^2, \qquad \mathcal L_{\rm int} =-\frac{\lambda_3^{(0)}}{3!}h^3 -\frac{\lambda_4^{(0)}}{4!}h^4,$$

with

$$\lambda_3^{(0)}=6\lambda v=\frac{3m_h^2}{v}, \qquad \lambda_4^{(0)}=6\lambda=\frac{3m_h^2}{v^2}.$$

The corresponding all-incoming Feynman rules are $-i\lambda_3^{(0)}$ and $-i\lambda_4^{(0)}$. The factorials are part of the definition; omitting them changes the numerical symbol attached to the same vertex. The expansion and tree relations follow from Schwartz 2014, §29.1, pp. 584–588.

At this order, measuring $m_h$ and $v$ reconstructs the two coefficients only because the potential has exactly two renormalizable parameters and one scalar doublet. Higher-dimensional terms such as $(H^\dagger H)^3$, additional scalars, or mixing can change the cubic and quartic relations while preserving the same quadratic mass locally.

Renormalized self-couplings

A useful off-shell definition is the renormalized one-particle-irreducible vertex

$$\Gamma_{hhh}^{\rm ren}(p_1^2,p_2^2,p_3^2;\mu,\xi,\mathcal S),$$

where $\mu$ is a renormalization scale, $\xi$ denotes gauge-fixing parameters, and $\mathcal S$ records the input and tadpole scheme. This object is calculationally useful but is generally momentum-, gauge-, and scheme-dependent. It must not be presented as a process-independent measured number without specifying how it is embedded in an observable.

The vacuum condition also has to be renormalized. One consistent choice imposes a vanishing renormalized one-point function by a tadpole counterterm. Another keeps the bare vacuum expectation value at the minimum of the bare potential and includes tadpole graphs explicitly. Fleischer and Jegerlehner formulate the latter organization so that parameter definitions do not inherit an avoidable gauge dependence from a shifted vacuum prescription Fleischer and Jegerlehner 1981, §§II–III, pp. 2004–2011. Either organization can give the same physical amplitude when used consistently; mixing pieces from the two does not.

A physical definition therefore specifies, at minimum,

Item	Required statement
Parameters	input set, subtraction scheme, and scale
Vacuum	tadpole prescription and definition of $v$
Higgs state	complex pole and residue convention
Vertex	external momenta and wave-function factors
Observable	process, cuts, interference, and perturbative order

For example, a multi-Higgs amplitude contains self-coupling diagrams together with boxes, Yukawa interactions, gauge corrections, and continuum interference. Varying only a cubic vertex in one subset of diagrams can violate the gauge or renormalization consistency of the full amplitude.

Reconstructing a potential from observables

The direction from a Lagrangian to amplitudes is unique once the scheme and order are fixed. The inverse direction need not be. A rate or distribution can depend simultaneously on

• the cubic and quartic scalar interactions;
• top and other Yukawa couplings;
• gauge–Higgs interactions;
• higher-dimensional contact operators;
• total-width assumptions and unobserved channels;
• parton distributions, radiative corrections, cuts, and interference.

Consequently, a statement about a “self-coupling measurement” is always conditional on a model and likelihood. A defensible result names the parameterization, reports correlated directions, and distinguishes the renormalized Lagrangian coefficient from a pole residue or fiducial observable. This page establishes that durable interface; current numerical constraints require dated experimental and theory evidence and are not asserted here.

Checks and limiting cases

Derivative check. At the stationary point,

$$\left.\frac{d^2V}{dh^2}\right|_{0}=m_h^2,\quad \left.\frac{d^3V}{dh^3}\right|_{0}=\lambda_3^{(0)},\quad \left.\frac{d^4V}{dh^4}\right|_{0}=\lambda_4^{(0)}.$$

This catches missing factorials and sign errors.

Dimensional check. In four dimensions, $[\lambda_3]=1$ and $[\lambda_4]=0$. A proposed cubic proportional to $m_h^2/v^2$ has the wrong dimension.

Decoupling check. Adding heavy fields may reproduce the tree relation at leading order while leaving corrections suppressed by powers of the heavy scale. The statement requires an explicit matching limit; it is not automatic from a heavy mass alone.

Gauge check. Gauge dependence of an isolated off-shell $\Gamma_{hhh}$ is not a physical inconsistency. Gauge dependence remaining in a complete on-shell or pole-defined observable at a fixed perturbative order is.

Common pitfalls

Calling $\lambda$ the cubic coupling. The coefficient $\lambda$ multiplies $(H^\dagger H)^2$. After shifting the field, the conventionally normalized cubic is $6\lambda v$.

Equating a diagram subset with an observable. A multi-Higgs process usually contains contributions not proportional to $\lambda_3$. Keep the complete gauge-invariant amplitude and its interference.

Mixing tree and renormalized identities. Substituting measured pole quantities into $\lambda_3=3m_h^2/v$ defines a tree-level reference unless the counterterm and input conversion are also supplied.

Handoff

A loop-safe self-interaction result passes forward

$$\left\{\lambda_3,\lambda_4;\ \mathcal S,\mu,\text{tadpole prescription};\ \{p_i^2\},\text{pole/residue convention};\ \text{process definition and covariance}\right\}.$$

Use the scheme fields on electroweak renormalization and input schemes and the observable fields on Higgs production, decay, and pole observables.

References

• Fleischer, J., and F. Jegerlehner. “Radiative Corrections to Higgs Decays in the Extended Weinberg–Salam Model.” Physical Review D 23, no. 9 (1981): 2001–2026. DOI.
• Schwartz, Matthew D. Quantum Field Theory and the Standard Model. Cambridge University Press, 2014, §29.1, pp. 584–588. DOI.