Diphotons from Tetraphotons in the Decay of a 125 GeV Higgs at the LHC

TL;DR

The paper investigates whether the Higgs diphoton excess at $m_h\simeq125$ GeV can arise from $h\to aa$ with $a\to\gamma\gamma$, where a very light pseudoscalar $a$ yields highly collimated photon pairs that can be misidentified as single photons, producing an apparent enhancement in $h\to\gamma\gamma$. It formalizes an effective Lagrangian with a pseudoscalar coupling to photons via $aF\tilde F$ and derives the rates for $h\to aa$ and $a\to\gamma\gamma$, along with the decay length and the required relation between the scales $M$ and $\Lambda$. The analysis shows that a misidentification probability $\epsilon$ together with $\mathcal{B}(h\to aa)$ can produce an elevated diphoton rate while suppressing other Higgs channels, and it maps the current constraints from low-energy, beam-dump, and collider experiments to regions in the $(m_a,M)$ plane. The work also outlines several UV-complete scenarios to generate the $aF\tilde F$ coupling, including couplings to $\tau$ leptons and $a-\pi^0$ mixing, and discusses direct tests at Primakoff experiments and through refined Higgs analyses in future data. Overall, the model provides a concrete, testable mechanism linking a light pseudoscalar to an apparent Higgs diphoton enhancement, with distinctive phenomenology across high- and low-energy experiments.

Abstract

Recently the ATLAS and CMS experiments have presented data hinting at the presence of a Higgs boson at $m_h\simeq125$ GeV. The best-fit $h\rightarrowγγ$ rate averaged over the two experiments is approximately $2.1\pm0.5$ times the Standard Model prediction. We study the possibility that the excess relative to the Standard Model is due to $h\rightarrow aa$ decays, where $a$ is a light pseudoscalar that decays predominantly into $γγ$. Although this process yields $4γ$ final states, if the pseudoscalar has a mass of the order tens of MeV, the two photons from each $a$ decay can be so highly collimated that they may be identified as a single photon. Some fraction of the events then contribute to an effective $h\rightarrowγγ$ signal. We study the constraints on the parameter space where the net $h\rightarrowγγ$ rate is enhanced over the Standard Model by this mechanism and describe some simple models that give rise to the pseudoscalar-photon interaction. Further tests and prospects for searches in the near future are discussed.

Figures (7)

• Figure 1: Fraction of events where both photon pairs are sufficiently collimated to pass ATLAS isolation cuts, as a function of the pseudoscalar mass.
• Figure 2: Regions of light pseudoscalar parameter space that are favored (blue/light) and disfavored (red/dark) by the current best-fit signal strengths in the $h\rightarrow\gamma\gamma,ZZ,WW,bb,\tau\tau$ channels. Contours are overlaid for the net diphoton (solid green) and $ZZ,WW,bb,\tau\tau$ rates (dashed yellow) expected at the LHC relative to the SM rates.
• Figure 3: A representative diagram of the leading contribution of the pseudoscalar, $a$, to $(g-2)_\mu$.
• Figure 4: Diagram that gives the leading contribution to $s\to d+a$ from an effective interaction between $a$ and the top quark.
• Figure 5: Regions of pseudoscalar mass and inverse $a\gamma\gamma$ coupling $M$ excluded by the constraints in Secs. \ref{['sec:g-2']}--\ref{['sec:further']}. The dashed curves are contours of $1~{\rm cm}$ and $50~{\rm cm}$ decay lengths for the pseudoscalar, assuming a boost $\gamma=m_h/2m_a$ with $m_h=125{~\rm GeV}$.