Search for neutron-antineutron oscillations at the Sudbury Neutrino Observatory

Tests on $B-L$ symmetry breaking models are important probes to search for new physics. One proposed model with $\mathrm{\Delta }(B-L)=2$ involves the oscillations of a neutron to an antineutron. In this paper, a new limit on this process is derived for the data acquired from all three operational phases of the Sudbury Neutrino Observatory experiment. The search concentrated on oscillations occurring within the deuteron, and 23 events were observed against a background expectation of 30.5 events. These translated to a lower limit on the nuclear lifetime of $1.48\times {10}^{31}\mathrm{yr}$ at 90% C.L. when no restriction was placed on the signal likelihood space (unbounded). Alternatively, a lower limit on the nuclear lifetime was found to be $1.18\times {10}^{31}\mathrm{yr}$ at 90% C.L. when the signal was forced into a positive likelihood space (bounded). Values for the free oscillation time derived from various models are also provided in this article. This is the first search for neutron-antineutron oscillation with the deuteron as a target.

Figure 1

Simulated SNO detector response in the two neutron-antineutron momentum regimes as described in Sec. 3b. Here, the number of photoelectrons (p.e.) is proportional to the visible energy of the event. Using a conversion factor of 9 p.e./MeV, $n\text{−}\overline{n}$ events have a visible energy signature in the range of 200 MeV to 2 GeV.

Figure 2

Comparison between the data and the simulated SNO detector response to contained atmospheric neutrinos using the corrected Bartol fluxes. The channel composition is defined in Eq. (5). Since $n\text{−}\overline{n}$ events are constrained in an energy window between 2,000 and 18,000 photoelectrons (Fig. 1), only atmospheric contained events in this energy range are selected. A selection criterion requiring that signal is present in at least 2,000 photomultiplier tubes in an event is applied.

Figure 3

Simulated Cherenkov electron ring (top) demonstrating the showering effect and muon ring (bottom) demonstrating the absence of the showering effect. Both particles were generated with 600 MeV of kinetic energy and placed at the center of the detector; the resulting PMT hit pattern is projected in ($\cos {\theta }_{\mathrm{PMT}}$, ${\varphi }_{\mathrm{PMT}}$) space. The colors shown represent the time the PMT was hit, where green points represent PMTs that were hit first and blue points are PMTs that were hit at a later time.

Figure 4

The distribution of the opening angle subtended at the fitted vertex by the arc between the fired PMT and the ring center coordinate ($\cos {\theta }_{mp}^{†}$, ${\varphi }_{mp}^{†}$) for both a showering ($e$-like) and nonshowering ($\mu$-like) particle.

Figure 5

An example of the dodecahedron construct implemented for multiple-ring detection in the ring-parameter space. Each black star represents the ($\cos {\theta }_{mp}$, ${\varphi }_{mp}$) coordinates of a local high density maxima of midpoints, which is denoted as ($\cos {\theta }_{mp}^{†}$, ${\varphi }_{mp}^{†}$). The dodecahedrons structure is rotated and centered around the point in the ring-parameter space with the highest density, in this case the black star in the lighter shade region. In this example, this event is reconstructed as a three-ring event, the primary ring in the lighter shade section and two secondary rings in the darker shade sections. The other sections are regions where no high density of midpoints is found.

Figure 6

Reconstruction of ${\pi }^{+}$ simulated at the origin of the detector using a nonshowering (top) and showering (bottom) expectation, highlighting the complexity of single pion reconstruction. $x_{\text{rec}}^{\prime }{\widehat{u}}_{\text{rec}}$ is the reconstructed vertex position, and ${\theta }_{\text{rec}}$ is the angle of the reconstructed track with respect to the original true track. Visual cross-verification of these events shows good ring recognition except for rings with $x_{\text{rec}}^{\prime }>350\mathrm{cm}$. This is used as a ring selection criterion.

Figure 7

Visual display of the small-ring pathology of a simulated $e^{-}$ event. The blue ring is the primary ring, ring with the best reconstruction [i.e. minimum of $-2\ln {\lambda }_{\alpha }$ in Eq. (7)], while the red is the secondary ring. This secondary ring is a MRF flaw in handling low statistics of triggered PMTs if the local vertex ($x_{\text{rec}}^{\prime }{\widehat{u}}_{\text{rec}}$) is near the acrylic vessel. This pathology appears when the reconstruction of the local vertex from the origin is $x_{\text{rec}}^{\prime }>350\mathrm{cm}$ along the track.

Figure 8

Comparison of $N_{\text{rings}}$ distribution for SNO’s three-phase data to simulated atmospheric neutrino and $n\text{−}\overline{n}$ oscillation events. The error bars signify all systematic uncertainties that are described in Sec. 6b.

Figure 9

Scatter plot of the $\mathrm{\Lambda }$ parameter [Eq. (8)] and the number of photoelectrons for the background and signal expectation of multiple-ring events. The red line denotes the isotropy cut applied to the data set to isolate signal events from background events (detailed in Sec. 6).