Indications for new scalar resonances at the LHC and a possible interpretation

Over the last few years, the CMS and ATLAS collaborations at the Large Hadron Collider (LHC) have reported excesses that could hint at several new scalar resonances. Although none of them has touched the discovery level, at least two of them, at about 95 GeV and 650 GeV, have been indicated by more than one experiments, and have reached statistical significance worthy of a serious investigation. Conservatively using only the numbers given by the experimental collaborations, we find combined global significances around 3$σ$ and 4$σ$ respectively for the 95GeV and 650GeV putative resonances. There are some more, like the one at 320 GeV, which have also been hinted at. We show that the data on only the 650 GeV resonance, assuming they stand the test of time, predict the existence of a doubly-charged scalar, and make the more common extensions of the scalar sector like those by gauge singlet scalars, the 2-Higgs doublet models or the Georgi-Machacek model, highly disfavored. We provide the readers with a minimalistic model that may possibly explain all the indications. Such a model can also accommodate the hints of a singly charged scalar at about 375 GeV, and a doubly charged scalar at about 450 GeV, as found by both the major LHC Collaborations, the combined global significance for each of them being above $2.5σ$. We show that even the scant data, with large error bars, have the potential to strongly constrain our model containing four scalar multiplets, which makes the model easily testable and falsifiable. Our analysis comes with the obvious caveat that the allowed parameter space that we find depends on the available data on all the new resonances, and may change in future. One may also note that this is an exploratory exercise that illustrates the difficulties when it comes to fitting several resonances simultaneously, even for next-to-minimal extensions of the SM.

💡 Research Summary

The paper undertakes a systematic review of several modest excesses reported by the CMS and ATLAS experiments at the Large Hadron Collider that could point to new scalar resonances beyond the Standard Model (SM). The authors focus on excesses that have reached a combined global significance of at least 3 σ, namely a narrow structure around 95 GeV, a broader bump near 650 GeV, and additional hints at roughly 320 GeV, 400 GeV, and 151 GeV, together with tentative signals of a singly‑charged scalar at about 375 GeV and a doubly‑charged scalar at about 450 GeV. By conservatively using only the published signal strengths (μ) and global significances (σ) from the collaborations, they combine the individual p‑values through a product‑method that accounts for the look‑elsewhere effect, obtaining combined global significances of ≈3 σ for the 95 GeV state and ≈4 σ for the 650 GeV state.

The authors then discuss three fundamental theoretical constraints that any viable extension of the scalar sector must satisfy: (i) custodial symmetry, encoded in the ρ‑parameter, which experimentally is extremely close to unity; (ii) the measured couplings of the 125 GeV Higgs boson (h₁₂₅) to gauge bosons and fermions, which are consistent with the SM within ≈10 %; and (iii) unitarity sum rules for longitudinal gauge‑boson scattering, which require that if a neutral scalar couples to WW or ZZ with SM‑like strength, there must also exist at least one doubly‑charged scalar to preserve the high‑energy cancellation.

Applying these constraints, the authors demonstrate that the simplest extensions—pure gauge singlets, models with only additional SU(2) doublets (including the various types of Two‑Higgs‑Doublet Models), and the original Georgi‑Machacek (GM) model—cannot simultaneously accommodate all observed excesses. A singlet cannot generate the sizable WW/ZZ rates seen for the 650 GeV resonance. Doublet‑only models either violate the unitarity sum rule (by lacking a doubly‑charged scalar) or require unacceptably large custodial‑symmetry breaking VEVs, which would shift the ρ‑parameter beyond experimental limits. The canonical GM model does provide doubly‑charged states, but its restricted scalar spectrum and fixed custodial multiplet structure cannot reproduce the pattern of masses (95 GeV, 320 GeV, 375 GeV, 450 GeV, 650 GeV) and decay branching ratios observed.

To resolve these tensions, the authors propose a minimal yet richer framework they call the “2‑Higgs‑Doublet extended Georgi‑Machacek” (2HDeGM) model. This construction augments the generalized GM model (which already contains a complex triplet and a real triplet) with an additional SU(2) doublet. The resulting field content comprises four scalar multiplets: two complex doublets, one complex triplet, and one real triplet. The model preserves custodial symmetry at tree level, keeping ρ ≈ 1, while the extra doublet supplies the necessary freedom to adjust the mixing angles and vacuum‑expectation‑value (VEV) ratios (v₁, v₂, u) so that the observed h₁₂₅ couplings remain SM‑like. Simultaneously, the triplet sector introduces both singly‑ and doubly‑charged scalars (H⁺, H⁺⁺) that naturally explain the unitarity sum rule for the 650 GeV resonance and predict observable signatures at the LHC.

Using only the limited experimental inputs (signal strengths in γγ, ττ, bb̄, WW, ZZ, and h h channels), the authors perform a parameter scan. They treat the global significances as independent probabilities, multiply them to obtain an overall likelihood, and then explore the allowed region in the space of mixing angles (α₁, α₂, …) and VEV ratios. The scan reveals a tightly constrained parameter space: the doubly‑charged scalar mass is forced to lie near 450 GeV, the singly‑charged scalar near 375 GeV, and the neutral CP‑even states align with the observed 95 GeV, 320 GeV, and 650 GeV peaks. Moreover, the model predicts specific branching‑ratio patterns, such as H⁺⁺ → W⁺W⁺ dominating over leptonic modes, and H⁺ → W⁺Z being sizable. These predictions are already close to current experimental limits, implying that the upcoming Run 3 data and the High‑Luminosity LHC will be able to confirm or exclude the scenario.

The paper also discusses indirect constraints, including electroweak precision observables (S, T parameters) and flavor‑changing neutral currents, showing that the chosen VEV hierarchy (with the triplet VEV u ≲ 10 GeV) comfortably satisfies them. The authors stress that their analysis deliberately avoids detailed assumptions about the scalar potential (stability, perturbativity, and higher‑order unitarity), focusing instead on the phenomenological implications of the observed excesses.

In the final sections, the authors outline experimental strategies to test the model: targeted searches for H⁺⁺ → W⁺W⁺ and H⁺ → W⁺Z in both leptonic and hadronic final states, precision measurements of h₁₂₅ couplings at future e⁺e⁻ colliders (ILC, CLIC), and dedicated analyses of the 95 GeV resonance in γγ, ττ, and bb̄ channels. They also note that a confirmation of the 650 GeV state with a sizable WW/ZZ rate would be a smoking‑gun for the presence of a doubly‑charged scalar, as required by the unitarity sum rule.

In summary, the work presents a coherent narrative that (1) compiles all statistically significant scalar excesses reported so far, (2) demonstrates the incompatibility of these excesses with the most common scalar‑sector extensions, and (3) introduces a minimally extended Georgi‑Machacek framework that can accommodate the data while making sharp, testable predictions. The authors emphasize that the model is highly predictive and therefore readily falsifiable with the next round of LHC data, making it a valuable benchmark for future searches for new scalar dynamics.