Novel Search for Heavy $\nu$ Mixing from the ${\beta }^{+}$ Decay of ${}^{\mathrm{38}\mathrm{m}}\mathrm{K}$ Confined in an Atom Trap

A new technique, full neutrino momentum reconstruction, is used to set limits on the admixture of heavy neutrinos into the electron neutrino. We measure coincidences between nuclear recoils and positrons from the beta decay of trapped radioactive atoms and deduce the neutrino momentum. A search for peaks in the reconstructed recoil time-of-flight spectrum as a function of positron energy is performed. The admixture upper limits range from $4\times {10}^{-3}$ to $2\times {10}^{-2}$ and are the best direct limits for neutrinos (as opposed to antineutrinos) for the mass region of 0.7 to 3.5 MeV.

Figure 1

Top view of the second MOT apparatus with the recoil (MCP) and $\beta$ (DSSSD and BC408 scintillator) detectors.

Figure 2

Bottom: 2D scatter plot of detected ${\beta }^{+}$ energy versus recoil TOF for $\approx 60\hspace{0.17em}000$ ${\mathrm{A}\mathrm{r}}^0$ events. Thin solid line: kinematic loci for “fast” branch events and pointlike back-to-back detectors. Thin dashed line: “slow” branch. The thick line shows fast and slow branches for a simulated 2 MeV neutrino. About twice as many events produce positive Ar ions, which are accelerated to TOF shorter than $1.0\mu \mathrm{s}$. Ion reconstruction is kinematically more complex and is not used here. Top: Using the events contained within the box on the 2D plot, the relative MCP efficiency $\eta (v)$ is determined.

Figure 3

The data points result from the kinematic reconstruction of events for the $T_{\beta }=1.7–1.9\mathrm{M}\mathrm{e}\mathrm{V}$ bin in terms of $\Delta \mathrm{T}\mathrm{O}\mathrm{F}$ assuming $m_{\nu }=0$ (see text). Also shown are the results of the ML fit to the data including the possible mixing of a 1 MeV ${\nu }_{\mathrm{x}}$. The dashed line shows the 90% confidence level upper limit of $|U_{e\mathrm{x}}{|}^2=6\times {10}^{-3}$ for the ${\nu }_{\mathrm{x}}$ peak as constrained by fits to all $T_{\beta }$ bins in the $\log $ plot (inset), and $\times 50$ for display purposes in the linear plot. The solid line shows the sum of the MC simulations (both ${\nu }_0$ and ${\nu }_{\mathrm{x}}$) and the accidental background (dotted line).

Figure 4

Top: Deduced admixture of ${\nu }_{\mathrm{x}}$ into ${\nu }_e$. Almost the same data set is fit for each fixed $m_{{\nu }_{\mathrm{x}}}$, so the errors are highly correlated. Bottom: 90% confidence level upper limits on the admixture of ${\nu }_{\mathrm{x}}$ into ${\nu }_e$. Thick solid line, our limits on ${\nu }_{\mathrm{x}}$ from ${}^{\mathrm{38}\mathrm{m}}\mathrm{K}$ ${\beta }^{+}$-recoil coincidences; dotted line, ${}^{\mathrm{37}}\mathrm{A}\mathrm{r}$ EC, Ref. 6; dash-dotted line, ${}^{\mathrm{20}}\mathrm{F}$ ${\beta }^{-}$ shape (${\overline{\nu }}_{\mathrm{x}}$), Ref. 4; dashed line, ${\overline{\nu }}_{\mathrm{x}}\to {\overline{\nu }}_e{e}^{+}{e}^{-}$, Ref. 21; thin solid line, $\pi \to e^{+}\nu$, Ref. 22.