Search for $B_s^0\to {\mu }^{+}{\mu }^{-}$ decays at D0

We report results from a search for the decay $B_s^0\to {\mu }^{+}{\mu }^{-}$ using $1.3{\mathrm{fb}}^{-1}$ of $p\overline{p}$ collisions at $\sqrt{s}=1.96\mathrm{TeV}$ collected by the D0 experiment at the Fermilab Tevatron Collider. We find two candidate events, consistent with the expected background of $1.24\pm 0.99$, and set an upper limit on the branching fraction of $\mathcal{B}(B_s^0\to {\mu }^{+}{\mu }^{-})<1.2\times {10}^{-7}$ at the 95% C.L.

Figure 1

Signal and background distributions for four of the variables used in the likelihood ratio test. The smooth curves are parametrizations of the sideband data (points), representing the background. The sideband distribution in (a) is parametrized as a Gaussian function. In (b), (c), and (d), the sideband distributions are parametrized as the sum of two exponential functions. The signal distributions (histograms) are from MC.

Figure 2

The distribution of discriminant $D_5$ for signal (solid histogram) and background (dashed histogram). The background distribution is derived from events in the sidebands, folded over possible values of $m_{\mu \mu }$ in the signal region. The signal distribution is from MC. The normalization of the MC is arbitrary.

Figure 3

The dimuon mass distribution at different stages in a sequence of cuts: (a) after $L_T>12{\sigma }_{L_T}$, (b) after $\min (\delta /{\sigma }_{\delta })>2.8$ and $P({\chi }^2)>0.01$, (c) after $D_4>0.5$, and (d) after $D_4>0.949$. In (a) and (b) the histograms are fit to the sum of two exponential functions, while in (c) and (d) they are fit to the sum of an exponential and a constant. The signal region (shaded) was excluded in the fits. In (d), three entries are included in the signal region: the entry near the upper bound of the signal region has a value of $D_4$ close to the threshold and fails the $D_5$ cut; the other two entries are the final candidates.

Figure 4

Mass distribution of $B^{+}$ candidates. The background distribution is parametrized as a parabola and the signal distribution as a Gaussian function.