Search for the proton decay mode $p\to \overline{\nu }{K}^{+}$ with KamLAND

We present a search for the proton decay mode $p\to \overline{\nu }{K}^{+}$ based on an exposure of 8.97 kton-years in the KamLAND experiment. The liquid scintillator detector is sensitive to successive signals from $p\to \overline{\nu }{K}^{+}$ with unique kinematics, which allow us to achieve a detection efficiency of 44%, higher than previous searches in water Cherenkov detectors. We find no evidence of proton decays for this mode. The expected background, which is dominated by atmospheric neutrinos, is $0.9\pm 0.2$ events. The nonbackground-subtracted limit on the partial proton lifetime is $\tau /B(p\to \overline{\nu }{K}^{+})>5.4\times {10}^{32}\mathrm{years}$ at 90% C.L.

Figure 1

The $K^{+}$ kinetic energy distribution from bound proton decays in ${}^{12}\mathrm{C}$ with (solid line) and without (dashed line) the intranuclear cascade. The monochromatic line indicates the $K^{+}$ kinetic energy for free proton decays.

Figure 2

Fractional difference of the average total detected p.e.’s of ${}^{60}\mathrm{Co}$ events relative to the detector center for data (squares) and MC (circles). The decrease of charge around the $\pm 6\mathrm{m}$ region is due to the absence of PMTs in the detector entrance at the top and bottom region and shadowing by suspension ropes.

Figure 3

Reconstructed energy spectrum of atmospheric neutrino candidates (black dots) within a 5-m-radius volume. The atmospheric neutrino MC (solid line) is normalized to the live time of data. The neutrino interaction uncertainties evaluated by the genie code are indicated by the shaded region. The expectations for the linear energy scaling with $\pm 1\sigma$ are also shown (dashed lines).

Figure 4

Typical waveform shape (shaded) expected from (a) and (b) $p\to \overline{\nu }{K}^{+}$ MC example events and (c) an atmospheric neutrino MC event, together with the best-fit curves from the multipulse fit (solid thick line) and from the single-pulse fit (solid thin line). In the upper panel, the residuals from the best fit are shown to identify multipulse events. The fitted energies and, for the multipulse fits, time differences between the first pulse from the $K^{+}$ (red dashed line) and the second pulse from the ${\mu }^{+}$ decay daughter of the $K^{+}$ (blue dashed line) are also shown. The shape parameters (see text) for each fit are (a) $\log (L_{\text{shape}})=0.06$ and (b) $\log (L_{\text{shape}})=0.02$.

Figure 5

Typical waveform shape (shaded) from (a) and (b) muon $+$ decay electron data and (c) atmospheric neutrino data, together with the best-fit curves from the multipulse fit (solid thick line) and from the single-pulse fit (solid thin line). In the upper panel, the residuals from the best fit are shown to identify multipulse events. The fitted energies and, for the multipulse fits, time differences between the first pulse from muons (red dashed line) and the second pulse from decay electrons (blue dashed line) are shown.

Figure 6

Vertex distribution of atmospheric neutrino candidates in the energy range of 30–1000 MeV for (a) pointlike events (${\chi }_T^2<1.2$) and (b) short-tracklike events ($1.2\le {\chi }_T^2<5.0$).

Figure 7

Distribution of best-fit energies of the two pulses $E_1$ and $E_2$ for data (left), atmospheric neutrino MC events (middle), and $p\to \overline{\nu }{K}^{+}$ MC events (right). $E_1$ provides an estimate of the $K^{+}$ kinetic energy, as illustrated in Fig. 1, and $E_2$ the kinetic energy sum for decay daughters of the $K^{+}$, summarized in Table 1. The box shows the selection cuts for $E_1$ and $E_2$. The detection efficiency and the expected number of background events were estimated from the MC event distributions for $p\to \overline{\nu }{K}^{+}$ (right) and atmospheric neutrinos (middle), respectively.

Figure 8

Shape parameter ($L_{\text{shape}}$) distributions of data (black dot), atmospheric neutrino MC (thick red line), and $p\to \overline{\nu }{K}^{+}$ MC (thin blue line) for (a) pointlike events (${\chi }_T^2<1.2$) and (b) short-tracklike events ($1.2\le {\chi }_T^2<5.0$). The contribution of the atmospheric neutrino background is normalized to the live time of data.

Figure 9

Detection efficiency as a function of $\mathrm{\Delta }T$ estimated from $p\to \overline{\nu }{K}^{+}$ MC in period I. The decrease for $\mathrm{\Delta }T<7\mathrm{ns}$, due to poor resolution of the multipulse fit, is the primary source of inefficiency for the $p\to \overline{\nu }{K}^{+}$ search.