
TUM-HEP 1422/22
IFT-UAM/CSIC-22-124
Indirect upper limits on `i→`jγγ from `i→`jγ
Fabiola Fortuna,1Alejandro Ibarra,2Xabier Marcano,3Marcela Mar´ın,1and Pablo Roig1
1Departamento de F´ısica, Centro de Investigaci´on y de Estudios Avanzados del Instituto Polit´ecnico Nacional,
Apdo. Postal 14-740, 07000 Ciudad de M´exico, M´exico.
2Physik-Department, Technische Universit¨at M¨unchen, James-Franck-Straße, 85748 Garching, Germany
3Departamento de F´ısica Te´orica and Instituto de F´ısica Te´orica UAM/CSIC,
Universidad Aut´onoma de Madrid, Cantoblanco, 28049 Madrid, Spain
We perform an effective field theory analysis to correlate the charged lepton flavor violating
processes `i→`jγγ and `i→`jγ. Using the current upper bounds on the rate for `i→`jγ, we
derive model-independent upper limits on the rates for `i→`jγγ. Our indirect limits are about
three orders of magnitude stronger than the direct bounds from current searches for µ→eγγ, and
four orders of magnitude better than current bounds for τ→`γγ. We also stress the relevance of
Belle II or a Super Tau Charm Facility to discover the rare decay τ→`γγ.
I. INTRODUCTION
The experimental observation of charged lepton fla-
vor violation (cLFV) would undoubtedly imply the ex-
istence of new physics beyond neutrino oscillations [1].
This has motivated a strong experimental program over
the last 75 years searching for different cLFV processes,
see Fig. 1, each observable providing complementary in-
formation about possible beyond the Standard Model
(BSM) scenarios.
Here we consider the cLFV decays of leptons to two
photons, `i→`jγγ [2–4], which have been explored in
less detail than other cLFV processes such as the single
photon process, `i→`jγ, specially for the case of τ→
`γγ [5–9].
Experimentally, µ→eγγ was searched for by several
experiments aiming also for µ→eγ. The latest of these
experiments was the Crystal Box detector, whose result
still provides the strongest bound for µ→eγγ [10].
This limit is however two orders of magnitude weaker
than present µ→eγ bounds, see Table I, since the
MEG experiment was optimized for back-to-back topolo-
gies and no new dedicated experiment for µ→eγγ has
been carried out since Crystal Box. On the other hand,
τ→`γγ has rarely been searched for. To the best
of our knowledge, the only existing direct experimental
search was performed by ATLAS, setting an upper limit
of BR(τ→µγγ)<1.5×10−4after the LHC run-I [11].
No direct experimental search exists for τ→eγγ.
An alternative for exploring the `i→`jγγ channels is
to recast the searches for `i→`jγ, as some of the events
of the former would fall into the signal region defined
for the latter [3]. This idea has been recently applied to
recast the BABAR search for τ→`γ [12], finding that at
90%CL BR(τ→µγγ)<5.8×10−4and BR(τ→eγγ)<
2.5×10−4[9]. These limits are however several orders
of magnitude weaker than the associated ones on τ→`γ
due to the low acceptance of these searches for τ→`γγ
events.
In this letter, we consider the theoretical correlation
●
●
●
●
●
●
●
●
●
●
●
●
●
○○
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▽▽
▲
▲
▲
▲
▲
▲
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
○○
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▽▽
▲▲ ▲▲
●μ → ���
▼μ → �γ
▲μ → �γγ
●τ → μ μ μ
▼τ → μ γ
▲τ → μ γγ
1940 1950 1960 1970 1980 1990 2000 2010 2020 2030
10-16
10-15
10-14
10-13
10-12
10-11
10-10
10-9
10-8
10-7
10-6
10-5
10-4
10-3
10-2
10-1
1
%
FIG. 1. Historical evolution for the 90%CL upper lim-
its on several cLFV leptonic decays. Solid markers corre-
spond to published direct experimental searches, while empty
ones indicate future expected sensitivities at MEG II [13],
Mu3e [14] and Belle II [15] (arbitrary year). Lighter mark-
ers for `i→`jγγ were obtained by recasting the available
searches for `i→`jγ[3,9] or µe →γγ [2]. The evolution for
τ→esector is similar to the τ→µone, with the exception
that no direct search for τ→eγγ exists.
between the `i→`jγγ and the `i→`jγdecays.
Clearly, any scenario generating `i→`jγwould auto-
matically generate a (model-independent) contribution
to `i→`jγγ, from the radiation of an additional pho-
ton in the final state. Further, any scenario generating
`i→`jγγ will generate a (model-dependent) contribu-
tion to `i→`jγat the quantum level. Barring cancella-
arXiv:2210.05703v1 [hep-ph] 11 Oct 2022