Search for lepton flavor-violating decay modes $B^0 \to K^{\ast 0}τ^\pm\ell^\mp$ ($\ell = e,μ$) with hadronic B-tagging at Belle and Belle II

TL;DR

The paper addresses charged-lepton-flavor-violating decays in the $B^0$ system, specifically $B^0 \rightarrow K^{*0}\tau\ell$ with $\ell\in\{e,\mu\}$. It uses a hadronic tagging strategy and a simultaneous unbinned maximum-likelihood fit to recoil mass $M_\tau$ in Belle and Belle II data, modeling the signal with a Johnson function and backgrounds with a polynomial. No signal is found, and 90% CL upper limits are set for four decay modes, providing the most stringent electron-mode constraints to date at a $B$-factory and contributing to LFV/BSM constraint space; the analysis also explores limits under left-handed and scalar Wilson-coefficient scenarios. The results advance experimental LFV searches in the $b$-quark sector and help tighten constraints on new physics explanations of LFU/LFV anomalies, with data and methods shared via HEPData.

Abstract

We present the results of a search for the charged-lepton-flavor violating decays $B^0 \rightarrow K^{*0}τ^\pm \ell^{\mp}$, where $\ell^{\mp}$ is either an electron or a muon. The results are based on 365 fb$^{-1}$ and 711 fb$^{-1}$ datasets collected with the Belle II and Belle detectors, respectively. We use an exclusive hadronic $B$-tagging technique, and search for a signal decay in the system recoiling against a fully reconstructed $B$ meson. We find no evidence for $B^0 \rightarrow K^{*0}τ^\pm \ell^{\mp}$ decays and set upper limits on the branching fractions in the range of $(2.9-6.4)\times10^{-5}$ at 90% confidence level.

Figures (4)

• Figure 1: Final signal efficiencies after all selection described in sec. \ref{['sec:selection']} as a function of the kinematic variables $M^2(K^{*0}\ell\xspace)$ and $M^2(K\xspace\xspace^{*0}\xspace t_{\tau})$ for Belle (left) and Belle II (right). From top to bottom: $OSe$, $SSe$, $OS\mu$, $SS\mu$.
• Figure 2: Fits to the recoil mass of the $B_{\mathrm{tag}}$ and $D_s^+$ system in the control channel $B^0\rightarrow D^- D_s^+$ for simulation (left) and data (right). The top plots correspond to Belle data, while the bottom ones are for Belle II data.
• Figure 3: The $M_\tau$ distribution and results of the simultaneous fits for Belle (left) and Belle II (right). The black dots with error bars show the data, the red dash-dotted curve shows the signal component, the blue dashed curve shows the background component, and the purple solid curve shows the global fit. From top to bottom: $OSe$, $SSe$, $OS\mu$, $SS\mu$.
• Figure 4: BDT distributions in generic MC simulations (stacked histograms), data sidebands $M_\tau \in [1.0, 1.65[\cup]1.9, 2.5]$${\mathrm{\,Ge V\!/}c^2}$ (black points) and signal MC (red histogram) for Belle (left) and Belle II (right) datasets. The generic MC are corrected for known data/MC mismodelling and scaled to the data luminosity, and signal MC is scaled to the same area as data. The dashed line indicates the value of the BDT cut applied. From top to bottom: $OSe$, $SSe$, $OS\mu$, $SS\mu$.