---
title: Anomalous Quartic Gauge Couplings (aQGCs)
url: https://www.emergentmind.com/topics/anomalous-quartic-gauge-couplings-aqgcs
type: topic
---

# Anomalous Quartic Gauge Couplings (aQGCs)

Searching arXiv for recent and foundational papers on anomalous quartic gauge couplings, SMEFT/HEFT bases, positivity, and collider probes.
Anomalous quartic gauge couplings (aQGCs) are deviations of four-gauge-boson self-interactions from the Standard Model prediction. In the Standard Model, quartic gauge couplings are fixed by the non-Abelian \(SU(2)_L\times U(1)_Y\) gauge structure and the Higgs mechanism; in a linearly realized SMEFT, genuine quartic interactions with no new cubic gauge interactions first arise from dimension-eight operators, whereas in HEFT genuine quartic gauge operators already appear at \({\cal O}(p^4)\) [2411.02483] [2311.09300]. Because aQGC contributions grow rapidly with energy and directly affect vector-boson scattering and related multi-boson processes, they are a central probe of electroweak symmetry breaking, heavy new physics, and the consistency constraints of quantum field theory.

## 1. Gauge-theory origin and physical meaning

In the Standard Model electroweak sector, the gauge group is
\[
SU(2)_L \times U(1)_Y.
\]
Because this structure is non-Abelian, the gauge fields self-interact. This produces triple gauge couplings such as \(WWZ\) and \(WW\gamma\), and quartic gauge couplings such as \(WWWW\), \(WWZZ\), \(WWZ\gamma\), \(WW\gamma\gamma\), \(ZZZZ\), \(ZZ\gamma\gamma\), \(Z\gamma\gamma\gamma\), and \(\gamma\gamma\gamma\gamma\) after electroweak symmetry breaking [2605.03116] [2107.01123]. In vector boson scattering, these quartic couplings appear already at leading order and are crucial to preserve perturbative unitarity at high energies; in the Standard Model, cancellations among quartic vertices, trilinear gauge couplings, and Higgs exchange tame the high-energy behavior of the amplitudes [2605.03116].

An aQGC is a deviation of these quartic interactions from the Standard Model value. The term “anomalous” is not used in the sense of gauge anomalies; it denotes non-standard interactions induced by heavy new physics and encoded in EFT Wilson coefficients [2107.01123]. Experimentally and phenomenologically, aQGCs are therefore interpreted as a model-independent description of new electroweak dynamics in processes such as VBS, triboson production, exclusive \(\gamma\gamma\to W^+W^-\), and neutral multi-photon channels.

## 2. EFT descriptions and operator bases

In a linearly realized SMEFT, the effective Lagrangian is organized as
\[
\mathcal{L}_{\text{SMEFT}} = \mathcal{L}_{\text{SM}} + \sum_i \frac{c_i^{(6)}}{\Lambda^2}\,\mathcal{O}_i^{(6)} + \sum_j \frac{c_j^{(8)}}{\Lambda^4}\,\mathcal{O}_j^{(8)} + \dots
\]
and the genuine aQGC sector is conventionally parameterized by dimension-eight operators of \(S\)-, \(M\)-, and \(T\)-type [2411.02483] [1902.08977]. In this language, \(S\)-type operators are built from Higgs covariant derivatives only, \(M\)-type operators mix Higgs derivatives with gauge field strengths, and \(T\)-type operators are quartic in gauge field strengths. The standard notation writes the quartic sector as
\[
\mathcal{L}_{\rm QGC} = \sum_i \frac{f_i}{\Lambda^4} O_i \,,
\]
or, more explicitly,
\[
\mathcal{L}_{\rm SMEFT}= \mathcal{L}_{\rm SM}
 \;+\;\sum_i \frac{f_{S,i}}{\Lambda^4}\,\mathcal{O}_{S,i}
 \;+\;\sum_j \frac{f_{M,j}}{\Lambda^4}\,\mathcal{O}_{M,j}
 \;+\; \sum_k \frac{f_{T,k}}{\Lambda^4}\,\mathcal{O}_{T,k}.
\]

| Class | Generic structure | Typical role |
|---|---|---|
| \(S\)-type | \((DH)^4\) | Longitudinal sector |
| \(M\)-type | \((DH)^2 F^2\) | Mixed longitudinal/transverse sector |
| \(T\)-type | \(F^4\) | Transverse sector, including neutral photonic vertices |

The LHC EFT Working Group note gives a definitive dimension-eight basis for operators that generate quartic but no cubic electroweak gauge interactions, and distinguishes CP-even from CP-odd structures [2411.02483]. In that basis, the CP-even sector contains \(S\)-, \(M\)-, and \(T\)-type operators, while there are no \(S\)-type CP-odd operators; CP-odd aQGCs occur only in the \(M\)- and \(T\)-type sectors. The note also states that earlier AEG bases omitted two operators that are C-odd, P-odd, but CP-even, namely \(\mathcal{O}_8^M\) and \(\mathcal{O}_9^M\), and provides mappings to the experimental FS/FM/FT conventions and to UFO implementations [2411.02483].

After electroweak symmetry breaking, these operators map onto physical quartic vertices. For example, \(\mathcal{O}_{T,0,1,2}\) and \(\mathcal{O}_{T,5,6,7}\) contribute to \(WW\gamma\gamma\), \(ZZ\gamma\gamma\), \(Z\gamma\gamma\gamma\), and \(\gamma\gamma\gamma\gamma\), while \(\mathcal{O}_{T,8,9}\) affect only neutral quartic couplings such as \(ZZZZ\), \(ZZ\gamma\gamma\), \(Z\gamma\gamma\gamma\), and \(\gamma\gamma\gamma\gamma\) [2307.01326] [2109.12572]. This is why purely neutral final states are especially sensitive to \(f_{T8}/\Lambda^4\) and \(f_{T9}/\Lambda^4\).

HEFT provides a different, non-linear organization of the same physics. In the HEFT analysis of genuine QGCs, the leading custodial-preserving operators are \({\cal P}_6\) and \({\cal P}_{11}\), with custodial-violating companions \({\cal P}_{23}\), \({\cal P}_{24}\), and \({\cal P}_{26}\), and no photons appear at \({\cal O}(p^4)\) because the basic chiral building blocks generate only \(W^\pm\) and \(Z\) quartic vertices at that order [2311.09300]. This difference matters when comparing HEFT fits to the more common SMEFT \(f_{S,i}\), \(f_{M,j}\), \(f_{T,k}\) parameterization.

## 3. Vertices, high-energy growth, and theoretical consistency

The defining phenomenological property of aQGC operators is their rapid energy growth. In the \(\mu^+\mu^- \to \nu\bar\nu\gamma\gamma\) study, the dimension-eight contribution is described schematically by
\[
\mathcal{M}_{\text{aQGC}} \sim \frac{f_i}{\Lambda^4}\,s^2,
\]
with a corresponding pure-new-physics cross section scaling approximately as
\[
\sigma_{O_i} \propto \left(\frac{f_i}{\Lambda^4}\right)^2 s^4
\]
until unitarity is approached [2409.07010]. In the same-sign muon-collider VBS study the same point is expressed as
\[
\mathcal{M}_{\rm BSM} \sim \frac{f_i}{\Lambda^4} E^4,
\]
with the resulting strong improvement when the collider energy is raised from \(2\) to \(6\) TeV [2605.03116]. Operationally, this is why essentially every analysis emphasizes high-\(p_T\), high-invariant-mass, or large-separation tails.

This growth also makes EFT consistency a central issue. Partial-wave unitarity is commonly imposed through
\[
|T^J| \le 2,
\]
either at the subprocess level or through event-level invariant-mass cuts [2409.07010] [1912.10686]. In \(\gamma\gamma\to W^+W^-\), the leading unitarity bounds take the form
\[
\left|\alpha _0\right|\leq \frac{32\pi M_W^2}{\hat{s}^2},\quad
\left|\alpha _1\right|\leq \frac{128\pi M_W^2}{\hat{s}^2},\quad
\left|\alpha _2\right|\leq \frac{16\pi}{\hat{s}^2},\quad
\left|\alpha _3\right|\leq \frac{64\pi }{\hat{s}^2},\quad
\left|\alpha _4\right|\leq \frac{48\pi}{\hat{s}^2},
\]
and, using the 95% event criterion, the 13 TeV study quotes a characteristic \(\sqrt{\hat s}=0.78\) TeV with bounds such as \(|\alpha_0|<1.75\) TeV\(^{-2}\) and \(|\alpha_2|<136\) TeV\(^{-4}\) [1912.10686]. Other analyses implement unitarity-safe regions by clipping or cutting on \(m_{WW}\), \(M_{\gamma\gamma}\), or related subprocess energies [2606.06436] [2502.20248].

Beyond unitarity, positivity imposes stringent UV-consistency requirements. The positivity analysis derives 19 linear inequalities, 3 quadratic inequalities, and 1 quartic inequality for the 18-dimensional dimension-eight aQGC parameter space, and finds that they reduce the allowed solid-angle volume to about \(2.1\%\) of the naïve space [1902.08977]. In one-operator benchmarks, some coefficients must be strictly positive, some strictly negative, and some are forbidden when switched on alone. This directly qualifies the widespread one-operator-at-a-time practice: it is a useful experimental projection, but it does not always correspond to a UV-completable direction in EFT parameter space [1902.08977].

## 4. Collider channels and the structure of sensitivity

Different processes isolate different quartic vertices. At high-energy muon colliders, \(\mu^+\mu^- \to \nu\bar\nu\gamma\gamma\) is used as a direct probe of \(WW\gamma\gamma\) through VBS-type diagrams \(\mu^+\mu^- \to \nu\bar\nu W^+W^- \to \nu\bar\nu\gamma\gamma\), with triboson topologies more important at lower energies and VBS dominance at higher energies [2409.07010]. A same-sign muon collider sharpens this logic further: because \(\mu^+\mu^+\) or \(\mu^-\mu^-\) cannot annihilate through a neutral \(s\)-channel, VBS becomes the dominant production mechanism, and the analysis can be organized into signal regions such as \(2V2\nu\), \(V\gamma\ell\nu\), \(2V\ell\nu\), \(2\gamma2\ell\), and \(2V2\ell\), probing charged and neutral QGC subclasses in a single setup [2605.03116].

Lepton and lepton-photon colliders are particularly effective for neutral quartic couplings. At CLIC stage 3, \(e^- \gamma \to e^- \gamma\gamma\) probes \(\gamma\gamma\gamma\gamma\) and \(Z\gamma\gamma\gamma\) vertices using the Weizsäcker–Williams approximation, and the analysis is restricted to \(f_{T,j}/\Lambda^4\) with \(j=0,1,2,5,6,7,8,9\) because the process is sensitive to neutral-photon-rich vertices [2307.01326]. The companion channel \(e^+e^- \to Z\gamma\gamma \to \ell^+\ell^-\gamma\gamma\) at \(\sqrt{s}=3\) TeV targets the same neutral quartic structures, with \(f_{T,8}/\Lambda^4\) and \(f_{T,9}/\Lambda^4\) especially sensitive to the \(Z\gamma\gamma\) final state [2112.03948].

Hadron-collider studies distribute sensitivity across a wide process set. \(pp\to Z\gamma\gamma\) constrains \(ZZ\gamma\gamma\) and \(Z\gamma\gamma\gamma\) at the HL-LHC, HE-LHC, and FCC-hh [2109.12572]. \(pp\to Z\gamma jj\) and \(pp\to W\gamma jj\) isolate mixed and tensor operators in VBS topologies, with polarization effects especially useful for \(O_{T_i}\) in \(Z\gamma jj\) and for \(O_{M_{2,3,4,5}}\), \(O_{T_{5,6,7}}\) in \(W\gamma jj\) [2107.01123] [2002.03326]. Electroweak \(pp\to \gamma\gamma jj\) is tailored to \(f_{T8}/\Lambda^4\) and \(f_{T9}/\Lambda^4\), while same-sign \(W^\pm W^\pm\) VBS at the LHC directly probes \(WWWW\) using leptonic angular information [2502.20248] [2606.06436]. Exclusive \(\gamma\gamma\to W^+W^-\) in pp collisions remains the cleanest channel for the \(\gamma\gamma WW\) quartic vertex itself [1912.10686].

## 5. Observables, polarization, and modern analysis strategies

Across channels, aQGC sensitivity is concentrated in high-energy corners of phase space. Typical discriminants are hard photons, large diboson or diphoton invariant masses, large transverse masses, and VBS-like jet configurations. Examples include \(p_T^\gamma\) and \(m_{\ell^+\ell^-\gamma\gamma}\) in \(Z\gamma\gamma\) [2109.12572], \(M_{Z\gamma}\) and the polarization-inspired variable
\[
r = \left(1 - |\cos\theta_\gamma|\right)^2 + \frac{\cos^2\theta'}{4}
\]
in \(Z\gamma jj\) [2107.01123], and the transverse-mass-like quantity
\[
M_{o1} = \sqrt{\left(|\mathbf{p}_T^{\ell^+}|+|\mathbf{p}_T^{\ell^-}|+|\slashed{\mathbf{p}_T}|\right)^2
  - \left|\mathbf{p}_T^{\ell^+}+\mathbf{p}_T^{\ell^-}+\slashed{\mathbf{p}_T}\right|^2}
\]
together with \(\cos(\theta_{\ell\ell})\) in \(\gamma\gamma\to W^+W^-\)-motivated pp analyses [1912.10686]. In same-sign \(WW\) VBS, the main traditional handle is
\[
m_T^{WW} = \sqrt{m_{\ell\ell}^2 + 2\big(E_T^{\ell\ell}E_T^{\rm miss} - \vec{p}_T^{\,\ell\ell}\cdot \vec{p}_T^{\rm miss}\big)},
\]
whose high-mass tail is strongly enhanced by dimension-eight operators [2606.06436].

Polarization and spin correlation have become a distinct methodological layer. In same-sign \(W^\pm W^\pm\) scattering, spin-correlation asymmetries provide sensitivity to anomalous \(WWWW\) interactions comparable to that obtained from the transverse-mass distribution of the \(WW\) system, and combining angular asymmetries with \(m_T^{WW}\) improves one-parameter limits by roughly \(13\)–\(20\%\) at \(3\) ab\(^{-1}\) [2606.06436]. In \(W\gamma jj\), the lepton-projection variable \(L_p\) and the two-dimensional \((L_p,\cos\theta')\) structure identify the distinctive transverse-helicity patterns of \(O_{T_{5,6,7}}\), while in \(Z\gamma jj\) the polarization effect is explicitly noted to highlight the signals of \(O_{T_i}\) operators [2002.03326] [2107.01123].

Machine learning has been used in both supervised and unsupervised forms. In electroweak \(pp\to\gamma\gamma jj\), the multivariate analysis is based on Boosted Decision Trees with 850 trees, maximum depth 3, AdaBoost, and 20–22 input variables built from the two leading photons and two leading jets [2502.20248]. Unsupervised anomaly detection has also been deployed: the isolation forest analysis of \(pp\to jj\ell^+\ell^-\nu\bar{\nu}\) achieved \(\hat{\mathcal S}_{\text{stat}}^{\max}\approx 0.630~\text{fb}^{1/2}\) for \(V_0\) and \(0.425~\text{fb}^{1/2}\) for \(V_3\), to be compared with \(0.691\) and \(0.341\) for the cut-based event-selection strategy [2103.03151]. Quantum-inspired anomaly detection has been explored in \(\mu^+\mu^-\to \nu\bar\nu\gamma\gamma\) through kernel \(k\)-means with three quantum kernels; in that setup the real-vector quantum kernel gives the best overall performance, outperforming the classical kernel for most of \(O_{M_2,M_3,M_4,T_5}\), with the classical kernel slightly better only for \(O_{T_1}\) [2409.07010].

## 6. Representative bounds and projected reach

Representative constraints span a very wide range because they depend on the vertex class, the channel, the collider energy, and whether the result is quoted in terms of \(\alpha_i\) or \(f_{S,M,T}/\Lambda^4\). The numbers below are taken directly from the cited studies and illustrate the present scale of the field [1912.10686] [2112.03948] [2307.01326] [2109.12572] [2502.20248] [2409.07010] [2605.03116].

| Channel | Scenario | Representative bound |
|---|---|---|
| \(\gamma\gamma\to W^+W^-\) in pp | 13 TeV, \(137.1~\text{fb}^{-1}\), \(SS\le2\) | \(\alpha_0\in[-0.0070,\,0.0076]\) TeV\(^{-2}\); \(\alpha_2\in[-0.45,\,0.39]\) TeV\(^{-4}\) |
| \(e^+e^- \to Z\gamma\gamma\) | CLIC 3 TeV, \(5~\text{ab}^{-1}\), unpolarized, \(\delta_{\rm sys}=0\%\) | \(f_{T8}/\Lambda^4 \in [-1.78,\,1.85]\times10^{-2}\); \(f_{T9}/\Lambda^4 \in [-3.83,\,3.48]\times10^{-2}\) TeV\(^{-4}\) |
| \(e^-\gamma \to e^-\gamma\gamma\) | CLIC 3 TeV, \(5~\text{ab}^{-1}\), unpolarized, \(\delta_{\rm sys}=0\%\) | \(f_{T8}/\Lambda^4 \in [-1.79,\,2.05]\times10^{-3}\); \(f_{T9}/\Lambda^4 \in [-0.50,\,0.32]\times10^{-2}\) TeV\(^{-4}\) |
| \(pp\to Z\gamma\gamma\) | FCC-hh 100 TeV, \(30~\text{ab}^{-1}\), \(\delta_{\rm sys}=0\%\) | \(f_{T8}/\Lambda^4 \in [-1.16,\,0.54]\times10^{-3}\); \(f_{T9}/\Lambda^4 \in [-1.26,\,1.09]\times10^{-3}\) TeV\(^{-4}\) |
| \(pp\to\gamma\gamma jj\) | FCC-hh 100 TeV, \(30~\text{ab}^{-1}\), 95% CL, \(\delta_{\rm sys}=0\%\) | \(f_{T8}/\Lambda^4 \in [-4.84,\,4.84]\times10^{-3}\); \(f_{T9}/\Lambda^4 \in [-2.46,\,2.46]\times10^{-2}\) TeV\(^{-4}\) |
| \(\mu^+\mu^- \to \nu\bar\nu\gamma\gamma\) | 14 TeV, \(20~\text{ab}^{-1}\), real-vector QKKM, \(\mathcal S_{\rm stat}=2\) | \(|f_{M_3}/\Lambda^4|<1.94\times10^{-4}\); \(|f_{T_1}/\Lambda^4|<2.39\times10^{-4}\) TeV\(^{-4}\) |
| same-sign \(\mu^+\mu^+\) VBS | 6 TeV, \(10~\text{ab}^{-1}\), 95% CL combined | \(f_{T,0}/\Lambda^4 \in [-1.5,\,1.5]\times10^{-4}\); \(f_{T,8}/\Lambda^4 \in [-3.4,\,3.3]\times10^{-5}\) TeV\(^{-4}\) |

Several broad trends are already clear in the literature. First, neutral photonic operators \(f_{T8}/\Lambda^4\) and \(f_{T9}/\Lambda^4\) are exceptionally well constrained in clean neutral channels: the \(e^-\gamma\) CLIC study quotes improvements by factors between \(2\) and \(200\) over experimental results, with particularly dramatic gains for \(f_{T8}\) and \(f_{T9}\) [2307.01326]. Second, future hadron colliders extend the reach substantially: in \(pp\to Z\gamma\gamma\), the FCC-hh projections improve current CMS limits by about one order of magnitude for \(f_{T0,1,2}\) and two orders of magnitude for \(f_{T8,9}\) [2109.12572]. Third, muon colliders shift the sensitivity frontier most strongly in VBS-dominated settings: the \(\mu^+\mu^- \to \nu\bar\nu\gamma\gamma\) analysis finds projected bounds \(2\)–\(3\) orders of magnitude stronger than current LHC results, while the same-sign \(\mu\)TRISTAN projections push many tensor couplings to the \(10^{-4}\)–\(10^{-5}\) TeV\(^{-4}\) level [2409.07010] [2605.03116].

Systematic effects and EFT consistency qualifications remain numerically important. In the CLIC \(Z\gamma\gamma\) study, the best limits with \(\delta_{\rm sys}=0\%\) are approximately improved up to about \(1.1\) times better than those obtained with \(\delta_{\rm sys}=5\%\), while initial electron beam polarization improves the sensitivity by almost a factor of \(1.2\) [2112.03948]. In same-sign \(WW\) scattering at the HL-LHC, imposing invariant-mass cut-offs on the EFT contribution yields explicitly unitarity-safe regions and substantially weakens some coefficients, notably \(f_{S1}/\Lambda^4\) [2606.06436]. A plausible implication is that future aQGC reporting will increasingly have to separate naïve EFT limits, unitarity-safe limits, and positivity-compatible regions rather than quoting a single interval for each coefficient.

Source: https://www.emergentmind.com/topics/anomalous-quartic-gauge-couplings-aqgcs