Search for radiative decays of ${\rm Y}(1\mathrm{S})$ into $\eta$ and $\eta \prime$

We report on a search for the radiative decay of ${\rm Y}(1\mathrm{S})$ to the pseudoscalar mesons $\eta$ and ${\eta }^{\prime }$ in $(21.2\pm 0.2)\times {10}^6$ ${\rm Y}(1\mathrm{S})$ decays collected with the CLEO III detector at the Cornell Electron Storage Ring. The $\eta$ meson was reconstructed in the three modes $\eta \to \gamma \gamma$, $\eta \to {\pi }^{+}{\pi }^{-}{\pi }^0$, or $\eta \to {\pi }^0{\pi }^0{\pi }^0$. The ${\eta }^{\prime }$ meson was reconstructed in the mode ${\eta }^{\prime }\to {\pi }^{+}{\pi }^{-}\eta$ with $\eta$ decaying through any of the above three modes, and also ${\eta }^{\prime }\to \gamma {\rho }^0$, where ${\rho }^0\to {\pi }^{+}{\pi }^{-}$. Five out of the seven submodes are found to have very low backgrounds. In four of them we find no signal candidates and in one [${\rm Y}(1\mathrm{S})\to \gamma {\eta }^{\prime }$, ${\eta }^{\prime }\to {\pi }^{+}{\pi }^{-}\eta$, $\eta \to {\pi }^{+}{\pi }^{-}{\pi }^0$] there are two good signal candidates, which is insufficient evidence to claim a signal. The other two submodes ($\eta \to \gamma \gamma$ and ${\eta }^{\prime }\to \gamma \rho$) are background limited, and show no excess of events in their signal regions. We combine the results from different channels and obtain upper limits at the 90% C.L. which are $\mathcal{B}({\rm Y}(1\mathrm{S})\to \gamma \eta )<1.0\times {10}^{-6}$ and $\mathcal{B}({\rm Y}(1\mathrm{S})\to \gamma {\eta }^{\prime })<1.9\times {10}^{-6}$. Our limits are an order of magnitude tighter than the previous ones and below the predictions made by some theoretical models.

Figure 1

Candidate $\eta \to {\pi }^{+}{\pi }^{-}{\pi }^0$ (top) and $\eta \to {\pi }^0{\pi }^0{\pi }^0$  (bottom) invariant mass distributions from ${\rm Y}(1\mathrm{S})$ data. The large number of events near $780\mathrm{MeV}/c^2$ (top) is due to the abundant process $e^{+}{e}^{-}\to \gamma \omega$. No events are observed in our acceptance region, bounded by the arrows.

Figure 2

Invariant mass distribution of $\gamma \gamma$ candidates in ${\rm Y}(1\mathrm{S})$ data for the mode ${\rm Y}(1\mathrm{S})\to \gamma \eta ;\eta \to \gamma \gamma$, overlaid with fits using (a) floating area (solid red line) yielding $-2.3\pm 8.7$ events, and (b) area fixed to 14.5 events (dashed blue line), the upper limit corresponding to 90% C.L.

Figure 3

Invariant mass distributions of $\eta {\pi }^{+}{\pi }^{-}$ candidates from ${\rm Y}(1\mathrm{S})$ data. The $\eta$ candidate is constrained to the nominal $\eta$ meson mass. No events are observed in the signal box for $\eta \to \gamma \gamma$ (top) and $\eta \to {\pi }^0{\pi }^0{\pi }^0$ (bottom); two signal events are observed for $\eta \to {\pi }^{+}{\pi }^{-}{\pi }^0$ (middle).

Figure 4

Invariant mass distribution of $\gamma \rho$ candidates in ${\rm Y}(1\mathrm{S})$ data for the mode ${\rm Y}(1\mathrm{S})\to \gamma {\eta }^{\prime };{\eta }^{\prime }\to \gamma {\rho }^0$ overlaid with fits using (a) floating area (solid red line) yielding $-3.1\pm 5.3$ events, and (b) area fixed to 8.6 events (dashed blue line), corresponding to the upper limit at 90% C.L.

Figure 5

Likelihood distributions as a function of branching fraction for the decay mode ${\rm Y}(1\mathrm{S})\to \gamma \eta$ (left) and ${\rm Y}(1\mathrm{S})\to \gamma {\eta }^{\prime }$ (right). All distributions are smeared by respective systematic uncertainties and normalized to the same area. The solid black curve denotes the combined likelihood distribution.