The Super-Kamiokande collaboration has published results on a search for dinucleon decay to pions. Dinucleon decay - the simultaneous decay of two bound nucleons into leptons or mesons - is a process that violates baryon number by two units. The specific modes under study, 16O(pp)→14Cπ+π+, 16O(pn)→14Nπ+π0, and 16O(nn)→14Oπ0π0,
have never been searched for in a water Cherenkov detector. The large exposure of bound nucleons in Super-Kamiokande makes it well-suited for this study.
The primary motivation of searching for dinucleon decay processes is that they are a test of baryon number conservation. Baryon number is not strictly conserved in the Standard Model, and so it is natural to ask whether baryon number-violating processes exist. In addition to purely theoretical concerns, it is known that processes violating baryon number must exist in order to explain the Baryon Asymmetry of the Universe - the observed excess of baryonic matter to antimatter. Discovery of such a process would be a spectacular confirmation of new physics. Dinucleon decay is complimentary to other baryon number-violating processes sought at Super-Kamiokande, such as single proton decay in modes such as p→e+π0 and p → ν K+.
The experimental signature of dinucleon decay to pions, in principle, is very clear. Two pions, with a large angle separating them, should emerge from the nucleus and create bright Cherenkov rings. This signal is heavily distorted from charged pion interactions, both in the nucleus and the water. Thus, for modes involving charged pions, a simple set of kinematical cuts does not provide an optimal sensitivity to the search. Instead, a multivariate method is employed, specifically a boosted decision tree, which combines different event variables into a single classifier that separates the dinucleon decay signal from atmospheric neutrino background. For the mode 16O(nn)→14Oπ0π0, the signal behavior is more predictable, and simple kinematical cuts suffice to provide an excellent sensitivity.
No excess was observed in 282.1 kiloton-years of Super-Kamiokande data in any of these searches. Thus, lower limits were set on the lifetimes of the processes. These limits are τpp→π+π+ > 7.3x1031 years, τpn→π+π0 >1.70x1032 years, and τnn→π0π0 > 4.04x1032 years. These limits are about two orders of magnitude more stringent than previous results obtained from the Frejus detector, which used iron nuclei. The results have been published in Physical Review D, and have been chosen for an ``Editor's Selection.''