Catalog · Top / Higgs / EW · T020

T020 $\mathcal{B}(h\to e\mu)$

Lepton-flavor-violating Higgs decay h $ \to $ e $\mu$

Status FACT-CHECK-FIXED VERIFIED High Code: NO Priority Low

PDG / equivalent values

Observable	Value	Year	Experiment / source	Provenance
Higgs boson mass hypothesis for h $ \to $ e $\mu$ direct searches	125 GeV	2025	PDG2025:T020:higgs_mass_hypothesis	source ↑
B(h $ \to $ e $\mu$), CMS full Run-2 direct search	< 4.4 x 10^-5 branching fraction	2025	PDG2025:T020:cms_run2	source ↑
B(h $ \to $ e $\mu$), ATLAS full Run-2 direct search	< 6.2 x 10^-5 branching fraction	2025	PDG2025:T020:atlas_run2	source ↑
B(H $ \to $ e $\mu$), CMS full Run-2 direct search	< 4.4 x 10^-5 branching fraction	2023	CMS2023:T020:emu_limit	source ↑
CMS dataset for h $ \to $ e $\mu$ direct search	138 fb^-1 at sqrt(s) = 13 TeV integrated luminosity and collision energy	2023	CMS2023:T020:dataset	source ↑
CMS X $ \to $ e $\mu$ mass scan range and largest excess	110-160 GeV scan; largest excess near 146 GeV mass range and excess significance	2023	CMS2023:T020:mass_scan	source ↑
B(H $ \to $ e $\mu$), ATLAS full Run-2 direct search	< 6.2 x 10^-5 branching fraction	2020	ATLAS2020:T020:emu_limit	source ↑
ATLAS dataset for h $ \to $ e $\mu$ direct search	139 fb^-1 at sqrt(s) = 13 TeV integrated luminosity and collision energy	2020	ATLAS2020:T020:dataset	source ↑

Why this constrains the RS scan

Warped or partially composite lepton sectors can generate off-diagonal physical Higgs Yukawa couplings after the charged-lepton mass matrix is diagonalized. The $h\to e\mu$ search directly probes the $e$-$\mu$ entries of that Higgs Yukawa matrix and is therefore complementary to the repo's existing $\mu\to e\gamma$ dipole filter. It is not a constraint on the current quark $\Delta F=2$ scan by itself; it belongs to a future lepton-sector extension.

What's changed since the original paper

The requested arXiv:0804.1954 baseline predates the LHC Higgs LFV program (CsakiFalkowskiWeiler2008WarpedFlavor). Harnik, Kopp, and Zupan then framed flavor-violating Higgs decays as direct discovery channels and showed that LFV Higgs branching fractions can be sizable in generic EFT settings (HarnikKoppZupan2012FlavorViolatingHiggs). ATLAS produced a full Run-2 $h\to e\mu$ limit with $139~\mathrm{fb}^{-1}$ (ATLAS2020HiggsEMu). CMS later used $138~\mathrm{fb}^{-1}$, improved the direct Higgs limit to $4.4\times10^{-5}$, and extended the same $e\mu$ mass scan to $110$--$160~\mathrm{GeV}$. The largest excess was near $146~\mathrm{GeV}$, with local/global significance $3.8/2.8\sigma$; CMS did not interpret it as evidence (CMS2023HiggsLFVEMu; CMS2023:T020:mass\_scan).

Validity and model dependence

Experimentally this is a clean direct upper limit on a Standard-Model-forbidden Higgs decay. A literal branching-ratio cut assumes the SM-like Higgs production mixture and total width used by ATLAS and CMS. Models with altered Higgs production, exotic Higgs width, or correlated low-energy LFV signals need a recast. PDG notes that indirect $\mu\to e\gamma$-based constraints are stronger for this channel, so this entry should be used as the direct collider observable rather than the global $e$-$\mu$ LFV bound.

Code coverage in this repo

Coverage is NO. The required greps and a targeted search for $h \to e \mu$ , $H \to e \mu$ , Y\_e\_mu, and Y\_mu\_e across quarkConstraints/, qcd/, flavorConstraints/, neutrinos/, yukawa/, warpConfig/, solvers/, scanParams/, and tests/ found no observable implementation. The existing lepton LFV backend is specifically $\mu\to e\gamma$ (flavorConstraints/muToEGamma.py:1; flavorConstraints/muToEGamma.py:75), and the scan applies that filter at scanParams/scan.py:523--scanParams/scan.py:524. The charged-lepton Yukawa helper is diagonal (yukawa/charged\_lepton.py:1; yukawa/charged\_lepton.py:36). See T020.yaml:code\_coverage.

Linked evidence (opens GitHub blob at flavor-catalog-website/2026q2):

Implementation difficulty

Production integration is HIGH. The direct limit is easy to catalog, but a scan constraint needs a new charged-lepton Higgs-Yukawa misalignment calculation, a convention for $Y_{e\mu}$ and $Y_{\mu e}$, and a Higgs-rate or total-width interpretation before accepting or rejecting points (T020.yaml:implementation\_difficulty\_rationale).

Reason: The direct experimental limit is simple to catalog, but production-grade RS integration needs a new charged-lepton Higgs-Yukawa misalignment calculation, a convention for Y_e_mu/Y_mu_e and their relation to B($h \to e \mu$), and a Higgs production/total-width interpretation. The current code only has diagonal charged-lepton Yukawas and a $\mu \to e \gamma$ dipole filter.

Key references

PDG2025HiggsLFVReview; CMS2023HiggsLFVEMu; ATLAS2020HiggsEMu; HarnikKoppZupan2012FlavorViolatingHiggs; CsakiFalkowskiWeiler2008WarpedFlavor; PerezRandall2008WarpedLeptonFlavor.

value 125

RESOLVED

Match snapshot line 25

L22:   B(h -> e mu) < 4.4 x 10^-5 (4.7 x 10^-5) from CMS [441].
L23: 
L24: The same paragraph notes that these numbers correspond to exotic decays of a
L25: 125 GeV Higgs boson. It also notes that the CMS e mu search was expanded to
L26: the mass range 110-160 GeV and reported a mild excess near 146 GeV with global
L27: significance 2.8 standard deviations, insufficient for a conclusion.
L28:

value < 4.4 x 10^-5

RESOLVED

Match snapshot line 22

L19: The full 13 TeV Run-2 ATLAS and CMS datasets set 95% CL upper limits on LFV
L20: Higgs branching ratios. For h -> e mu, the observed (expected) limits are:
L21:   B(h -> e mu) < 6.2 x 10^-5 (5.9 x 10^-5) from ATLAS [273].
L22:   B(h -> e mu) < 4.4 x 10^-5 (4.7 x 10^-5) from CMS [441].
L23: 
L24: The same paragraph notes that these numbers correspond to exotic decays of a
L25: 125 GeV Higgs boson. It also notes that the CMS e mu search was expanded to

value < 6.2 x 10^-5

RESOLVED

Match snapshot line 21

L18: 
L19: The full 13 TeV Run-2 ATLAS and CMS datasets set 95% CL upper limits on LFV
L20: Higgs branching ratios. For h -> e mu, the observed (expected) limits are:
L21:   B(h -> e mu) < 6.2 x 10^-5 (5.9 x 10^-5) from ATLAS [273].
L22:   B(h -> e mu) < 4.4 x 10^-5 (4.7 x 10^-5) from CMS [441].
L23: 
L24: The same paragraph notes that these numbers correspond to exotic decays of a

value 138 fb^-1 at sqrt(s) = 13 TeV

RESOLVED

Match snapshot line 19

L16: 
L17: Relevant public-page summary:
L18: 
L19: The CMS search uses 138 fb^-1 of pp collision data at sqrt(s) = 13 TeV and
L20: looks for the lepton-flavor-violating decay of the 125 GeV Higgs boson and for
L21: additional Higgs bosons X with masses in the 110-160 GeV range, in the e mu
L22: final state. No excess is observed for the 125 GeV Higgs boson.

value 110-160 GeV scan; largest excess near 146 GeV

RESOLVED

Match snapshot line 21

L18: 
L19: The CMS search uses 138 fb^-1 of pp collision data at sqrt(s) = 13 TeV and
L20: looks for the lepton-flavor-violating decay of the 125 GeV Higgs boson and for
L21: additional Higgs bosons X with masses in the 110-160 GeV range, in the e mu
L22: final state. No excess is observed for the 125 GeV Higgs boson.
L23: 
L24: The observed (expected) upper limit on the branching fraction is:

value 139 fb^-1 at sqrt(s) = 13 TeV

RESOLVED

Match snapshot line 18

L15: 
L16: Relevant arXiv abstract extract:
L17: 
L18: ATLAS used 139 fb^-1 of pp collision data collected at sqrt(s) = 13 TeV. For a
L19: Higgs boson mass of 125 GeV, no significant signal was observed.
L20: 
L21: The observed (expected) 95% CL upper limit on the branching fraction is: