h → 4Ɣ

Contact Person(s)
David McKeen
More details on this mode may be found in Sec. 9 of the Survey of Exotic Higgs Decays, arXiv:1312.4992.

Theoretical Motivation

The basic decay chain that we consider is h→ aa(′), a(′)→γγ. There are a number of theoretically well-motivated candidates for a(′), among them the lighter pseudoscalar of the NMSSM, any pseudoscalar that mixes with the CP-odd Higgses of the (N)MSSM, or a generic SM-singlet boson whose coupling to photons is mediated by a renormalizable coupling to heavy vector-like matter. In the first two cases, the coupling of a to light SM fermions can make the branching for a→γγ subdominant, but the low backgrounds in 4γ can nonetheless make it an interesting final state. On the other hand, if a couples at the renormalizable level only to the Higgs and to heavy vector-like uncolored matter, it may only be able to decay to γγ, rendering the 4γ final state extremely important. If, alternatively, the vector-like matter is colored and a→ gg is allowed, h→ ggγγ can also be important (see the γγjj page for details).

Existing Collider Studies

The h→ aa→4γ decay chain was studied in [1], focusing on the Tevatron. The results of this analysis imply that the full Tevatron dataset is sensitive to branchings of h→ aa at about the 0.5% level or larger, assuming Br(a→γγ)=1.
In [2], a detailed study was performed of the h→ aa→4γ decay at the LHC with √s=14 TeV. The experimental cuts made in this study were that the transverse momenta of the photons were all greater than 20 GeV, the distance between the photons was ∆R > 0.4, the photons had rapidity |η| < 2.5, and there were two separate pairs of photons that reconstructed the same invariant mass (the candidate a mass) to within 5 GeV. This work indicated that for a Higgs at 125 GeV, 300 fb−1 of data at the 14 TeV LHC would allow branchings Br(h→ aa) ≅ 5×10−5 to be discovered at the 5σ level for 10 GeV <~ma <~mh/2, assuming that the a's decay promptly to photons only. A naive rescaling by the decreased luminosity and Higgs production cross section of the 7 and 8 TeV datasets, assuming that the dominant backgrounds' cross sections do not change appreciably, implies that the current data is sensitive to Br(h→ aa) ∼  few×10−4. As emphasized in [2], the reach is extremely sensitive to the value of the photon pT cut, especially in the case of a relatively light Higgs with mh=125 GeV.
Closely-spaced pairs of photons in h→ aa→4γ at the LHC when ma << mh=125 GeV were studied recently in [3], motivated by early hints at √s=7, 8 TeV that the Higgs rate to diphotons could be larger than in the SM. However, photon pairs that fail mutual isolation criteria might or might not be detected as a single photon. For details, see Sec. 9 of the Survey of Exotic Higgs Decays, arXiv:1312.4992.
Additionally, the usefulness of jet-substructure-motivated detector variables in distinguishing closely-separated photons from single photons and their interplay in h→4γ faking h→γγ at the LHC was studied in detail in [4,5], dealing with both the case where the photons were merged enough to potentially fake a single photon and that in which they are less closely merged but do still fail isolation cuts, potentially looking like a jet. Examining h→ 4γ with a Higgs mass of 120 GeV, they determined that the use of such variables could decrease the rate of photon-jets faking single photons by a factor of over 10 while preserving at least 80% of the single photon signal.

We pause here to note that if a or a′ is light, it is quite natural to get a decay length that is detector-scale. For example, parametrizing the coupling of a pseudoscalar to photons as


one gets a decay length, if they are produced in the decay of h at rest, of


It is easy to see that for ma <~100 MeV and M >~1 TeV,1 a's decay length could be of the order of several meters. Long decay lengths are therefore a generic feature of light pseudoscalars decaying to photons and should be kept in mind when contemplating such signals.2

Existing Experimental Searches and Limits

A search for h→ aa→4γ in the case where ma << mh leading to very collimated pairs of photons was performed by ATLAS on 4.9 fb−1 of 7 TeV data [6]. The search was very similar to the standard one for h→γγ but shower shape variable cuts were relaxed to allow for increased acceptance of the 4γ signal. Results were presented for ma=100, 200, 400 MeV, limiting Br(h→ aa)Br(a→γγ)2 < 0.01 at mh=125 GeV. For larger a masses, there are no limits from collider searches.
Results from low energy experiments (see, e.g. Ref. [7]) are not constraining on this scenario for ma >~10 MeV so long as the a's decay promptly at the LHC [3].


[1]B. A. Dobrescu, G. L. Landsberg, and K. T. Matchev, Higgs boson decays to CP odd scalars at the Tevatron and beyondPhys.Rev. D63 (2001) 075003, [hep-ph/0005308].
[2]S. Chang, P. J. Fox, and N. Weiner, Visible Cascade Higgs Decays to Four Photons at Hadron CollidersPhys.Rev.Lett. 98 (2007) 111802, [hep-ph/0608310].
[3]P. Draper and D. McKeen, Diphotons from Tetraphotons in the Decay of a 125 GeV Higgs at the LHCPhys.Rev. D85 (2012) 115023, [arXiv:1204.1061].
[4]S. D. Ellis, T. S. Roy, and J. Scholtz, Jets and Photons, [arXiv:1210.1855].
[5]S. D. Ellis, T. S. Roy, and J. Scholtz, Phenomenology of Photon-Jets, Phys.Rev. D87 (2013) 014015, [arXiv:1210.3657].
[6]ATLAS Collaboration, Search for a Higgs boson decaying to four photons through light CP-odd scalar coupling using L of 7 TeV pp collision data taken with ATLAS detector at the LHC.
[7]J. Hewett, H. Weerts, R. Brock, J. Butler, B. Casey, et. al. Fundamental Physics at the Intensity Frontier, [arXiv:1205.2671].


1We would expect such a scale if a's coupling to photons came from integrating out charged matter above the electroweak scale.
2This conclusion can be modified slightly when other decay channels for a are present or if the operator a Fμν ~Fμν is generated below the electroweak scale. See [3] for details.
File translated from TEX by TTH, version 4.03. On 17 Dec 2013, 22:31.