New constraint on the existence of the mu+-> e+ gamma decay

TL;DR

The paper reports an updated search for the lepton-flavor-violating decay mu+ -> e+ gamma using a combined dataset of 3.6 × 10^14 stopped muons collected by the MEG experiment at PSI. No excess events over background are observed, leading to a new upper limit on the branching ratio BR(mu+ -> e+ gamma) < 5.7 × 10^-13 at 90% CL, a fourfold improvement over the previous best limit. The analysis blends 2009–2010 data with 2011 data, employing improved positron and photon reconstruction, per-event PDFs, a Kalman-filter–based tracking, and a blind likelihood approach to separate signal from radiative muon decay (RMD) and accidental (ACC) backgrounds. The result significantly constrains beyond-Standard-Model scenarios predicting LFV, with future data and upgrades anticipated to push sensitivity further.

Abstract

The analysis of a combined data set, totaling 3.6 \times 10^14 stopped muons on target, in the search for the lepton flavour violating decay mu^+ -> e^+ gamma is presented. The data collected by the MEG experiment at the Paul Scherrer Institut show no excess of events compared to background expectations and yield a new upper limit on the branching ratio of this decay of 5.7 \times 10^-13 (90% confidence level). This represents a four times more stringent limit than the previous world best limit set by MEG.

Figures (3)

• Figure 1: (Top) The Michel positron spectrum in the same dataset in the time sidebands with the old (black dashed line) and the new (red solid line) track reconstruction code. (Bottom) The photon background spectra from the time sidebands of the muon data in 2010 with different pile-up elimination algorithms. Black dots: no pile-up elimination; blue dot-dashed: previous algorithm; red solid: new algorithm.
• Figure 2: Event distributions for the combined 2009--2011 dataset in the ($E_\mathrm{e}, E_{\mathrm{\gamma}}$)- and ($\cos\Theta_{\mathrm{e \gamma}}, t_{\mathrm{e \gamma}}$)-planes. In the top (bottom) panel, a selection of $|t_{\mathrm{e \gamma}}|<0.244\,{\rm ns}$ and $\cos\Theta_{\mathrm{e \gamma}}<-0.9996$ with 90% efficiency for each variable ($52.4<E_\mathrm{e}<55\,{\rm MeV }$ and $51<E_{\mathrm{\gamma}}<55.5\,{\rm MeV}$ with 90% and 74% efficiencies for $E_\mathrm{e}$ and $E_{\mathrm{\gamma}}$, respectively) is applied. The signal PDF contours (1, 1.64 and 2 $\sigma$) are also shown.
• Figure 3: Observed profile likelihood ratios ($\lambda_\mathrm{p}$) as a function of the branching ratio for the 2009--2010 combined data, the $2011$ data alone and the combined 2009--2011 data sample.