Search for invisible modes of nucleon decay in water with the SNO+ detector

This paper reports results from a search for nucleon decay through invisible modes, where no visible energy is directly deposited during the decay itself, during the initial water phase of $\mathrm{SNO}+$. However, such decays within the oxygen nucleus would produce an excited daughter that would subsequently deexcite, often emitting detectable gamma rays. A search for such gamma rays yields limits of $2.5\times {10}^{29}\mathrm{y}$ at 90% Bayesian credibility level (with a prior uniform in rate) for the partial lifetime of the neutron, and $3.6\times {10}^{29}\mathrm{y}$ for the partial lifetime of the proton, the latter a 70% improvement on the previous limit from SNO. We also present partial lifetime limits for invisible dinucleon modes of $1.3\times {10}^{28}\mathrm{y}$ for $nn$, $2.6\times {10}^{28}\mathrm{y}$ for $pn$ and $4.7\times {10}^{28}\mathrm{y}$ for $pp$, an improvement over existing limits by close to 3 orders of magnitude for the latter two.

Figure 1

Spectrum of deexcitation gamma rays emitted after the invisible decay of ${}^{16}\mathrm{O}$ for single nucleon decay [14] (left) and dinucleon decay [16] (right) as a function of gamma ray energy $E_{\gamma }$ between 6 and 10 MeV.

Figure 2

Comparison of the x-component of reconstructed position between ${}^{16}\mathrm{N}$ data acquired at a central source position and Monte Carlo simulation.

Figure 3

Comparison of reconstructed energy between ${}^{16}\mathrm{N}$ data acquired at a central source position and Monte Carlo simulation.

Figure 4

Data and Monte Carlo distributions of events in ${\beta }_{14}$ for ${}^{214}\mathrm{Bi}$ and ${}^{208}\mathrm{Tl}$ decays within the AV water analysis region for data set 3. The normalizations of the Monte Carlo spectra were fit to the observed data.

Figure 5

Fitted energy spectrum across all data sets for neutron decay and backgrounds. Backgrounds were fit for each data set and the spectrum shown is the sum over all of the time-bins for the individual components. The contributions of Bi and Tl were fit independently for internal backgrounds but were merged in this plot. The signal is shown at its maximum likelihood, not at 90% C.I. The errors around the full fit include the MC statistical uncertainties summed in quadrature with the individual systematic uncertainties. These errors are bin-by-bin correlated and are dominated by the energy resolution and the energy scale systematic uncertainties.

Figure 6

The likelihood ratio (the maximum likelihood as a function of the signal divided by the likelihood at the best fit value) for the neutron decay mode of the spectral analysis versus the limit of the number of signal decays per day. The associated cumulative distribution indicates the point at which the limit on the signal at 90% C.I. is drawn.