Contact Person(s)

Jessie Shelton, Tao Liu More details on this mode may be found in Section 5 of Survey of Exotic Higgs Decays (arXiv:1312.4992).  

Theoretical Motivations

The signature h→ γ+MET can be usefully represented through the decay of the Higgs into two neutral fermions, h→χ₁χ₂, followed by the decay χ₂→γχ₁. This signature can be realized in several ways: First, in low energy GMSB [1], where the Higgs decays into a gravitino and a bino, h→~G ~B followed by ~B → γ~G [2]. For masses in the range m_h/2 < m_~B < m_h, the Higgs decays as above. For lower ~B mass the decay h→ ~B~B will dominate, leading to a h→ 2γ+MET signature instead. Another realization of this signature is in the NMSSM. Here the singlino ~s plays the role of the light fermion, while the heavier one is mostly a bino. The mass splitting between the two fermions is now much more free. However, a light singlino is always accompanied by a light scalar s, and for the loop-induced branching fraction Br(χ₂→χ₁γ) to be sizable, the tree level decays Br(χ₂→ s^(*)χ₁→ f―f χ₁) must be phase-space suppressed. Therefore, in order to have a sizable rate into h→γ+MET, the mass splitting between the two neutralino species should be at most 10-20 GeV. More generally, any SM extension which includes two Majorana fermions, χ₂ and χ₁, with a dipole moment coupling

(1)

may potentially lead to a γ+MET signal, along with Z+MET, if allowed by phase space. However, in many UV completions of this scenario, the mass-splitting between the fermions is smaller than m_Z since it arises from some symmetry breaking, leading to a suppressed Z mode. The simplified model is then characterized by two parameters m₁ and m₂, where m₁ < m₂ and m₁+m₂ < m_h.  

Existing Collider Studies

  An LHC study was carried out at parton level in [2]. This study targets Higgs bosons produced in gluon fusion and estimates that 20 fb⁻¹ of 8 TeV data would allow 95% CL sensitivity to branching fractions ranging between Br(h→γ+MET) < 0.002 for m_χ2 = 120 GeV, and Br(h→γ+MET) < 0.010 for m_χ2 = 60 GeV. These results are based on selection criteria that are not obviously compatible with current LHC triggers, however, as the selection of Ref. [2] requires

(2)

and no other triggerable objects. Current monophoton triggers require p_T,γ > 80 GeV, although trigger cuts for CMS parked data are more relaxed, p_T,γ > 30 GeV and MET > 25 GeV for central photons, and therefore could be relevant for this decay channel. Replacing the cut on the photon's p_T with a cut on m_T (γ,MET) proves efficient in separating signal from background. Trigger thresholds ensure that the dominant contribution to the reach comes from the high-p_T tail of the Higgs production spectrum, where the Higgs recoils against one or more hard ISR jets. Depending on the mass difference between χ₁ and χ₂ and the analysis threshold achieved in parked monophoton +MET triggers, the best signal acceptance may be achieved in monojet+MET-triggered events rather than monophoton+MET-triggered events.  

Existing Experimental Searches and Limits

Very few existing collider searches place any limits on Br(h→γ+MET). The vast majority of existing DM and SUSY searches at both the LHC and the Tevatron use p_T cuts that are too hard to be effective for the typical photon p_T and  MET in this signal.

CMS' supersymmetry search in the γ+MET+jets final state with 4.04 fb⁻¹ of 8 TeV data [4] comes closer to being constraining, although it requires all events to have H_T > 450 GeV, implying major reduction in the signal efficiency. The reinterpretation of this search to h→γ+MET is difficult due to signal contamination in a region MET < 100 GeV used to model the dominant QCD background (the light 125 GeV Higgs contributes proportionately more to the control region than does the BSM signal used in the search). This makes the recast likely overconservative to an extent that is difficult to estimate. In Fig. 1 we show the result of performing this recast.  With perfect photon efficiency, the 95% CL limits obtained on Br(h→γ+MET) is approximately unity in a large range of parameter space, suggesting that an analysis more tailored to the signal kinematics could place meaningful limits on the branching fraction for this channel.   As with all semi-invisible signals, collider reach could be extended by forming the transverse mass of the visible decay product(s), here the photon, with the missing transverse momentum vector, and requiring this to be bounded from above as consistent with production from an initial resonance. Much better sensitivity could be achieved if the prohibitively hard cut on H_T could be relaxed. This H_T cut is necessitated by the γ+H_T trigger used to select the data in the current analysis, and is not suited well to the study of the relatively low-p_T Higgs events. Somewhat better signal acceptance is realized for the monophoton+MET triggers in current use for dark matter searches, though the degree of improvement depends on the spectrum; again, monojet+MET triggers may provide better sensitivity.

References