In general, a particle, or system of particles will be stable if it is
either in the lowest possible energy state, or if there is some
mechanism which keeps it from reaching a lower energy state. For
example, and electron can not decay to a neutrino nor a photon, even
though both would be less massive, because this interaction would not
conserve charge. This it must do because there is an underlining
symmetry (local gauge invariance of a scalar field) which the
particles follow. This symmetry dictates charge conservation.
But consider the proton, which is a collection of three quarks not a single particle, one might expect it to decay to, say, a charged lepton and a neutral meson or a neutral lepton and a charged meson by having one of the quarks turn in to a lepton. This decay (which is actually a scattering when one looks at the level of quarks) would conserve charge. However, in changing a quark to a lepton a quantity called Baryon number is changed overall (as is Lepton number).
So, how is Baryon (or Lepton) number different than charge? There is no underlining symmetry which dictates the conservation of Baryon (or Lepton) number. In the Standard Model it is only assumed that these quantities are conserved because there has yet been no observation of nonconservation. But absence of observation does not mean observation of an absence. Furthermore, to make up for failings of the Standard Model even the assumption of Baryon number conservation must be thrown out for most theories that go beyond the Standard Model need protons to decay. The tricky part for most of these models is to actually get the proton to decay at a slow enough rate to agree with experiment.
To search for proton decay we search for the decay products. Depending on the mode of decay we will look for different products. The mode P->e+pi0 is favored by dimension 6 theories while dimension 5 theories drive the lifetime of this mode too high and instead prefer P->mu+K0 and P->nu K+.
The figure above shows an ideal decay of a proton at rest to a
positron and a neutral pion. The two products exit the decay back to
back, each carrying half the proton's energy. The pion will almost
immediately decay into two photons (98% branching ratio). These
photons will pair produce an electron/positron pair. These two pairs,
along with the first positron, will form an electromagnetic shower.
Since all of these charged particles are moving at relativistic speeds
they will emit Cherenkov radiation and
the three resulting cones, two of which may overlap, will form the
characteristic three rings on the detector wall. Such an event might
look like that in the event display to the left. You can see the
positron's ring on the right half of the detector and the two
overlapping gamma ray rings from the decay of the pion on the left
half. Of course this event is from a sample of Monte Carlo simulated
proton decay data.
This event display looks much different from the ideal cartoon above. This is due to several reasons. First, the neutral pion may decay asymmetrically with on gamma getting more energy than the other. The lesser energetic gamma can even become undetectable which would give only a two ring event. Second, the 80% of the protons in water are inside the oxygen nucleous and the pions from the decay of these protons must risk passage though the oxygen nucleous. In the nucleous the pion can be absorbed, scattered, create more pions or simply exit unscathed. If the pion is absorbed, only a single ring from the positron would be observed, while if more pions are created there would be more than three rings. If the pion undergoes scattering then the back to back direction of the positron and pion will be spoiled. Third, the proton in oxygen is not at rest, but has Fermi momentum. Besides boosting the total energy of the decay products this will also give a nonzero total momentum. When placing cuts to perform the search these things must be considered.
Some of the other modes that can (and will be studied) at
Super-Kamiokande are shown below. This shows the estimated 90%
confidence lower lifetime limit that SuperK can obtain after 5 years
assuming no candidates are found compared against current and past
best limits. The projected SuperK limits are in green hashes.
The first proton decay paper (as submitted to PRL and in a longer form, hep-ph/9806014) reports a partial lifetime limit of 1.6x1033 years (90% CL) using 414 days of data for the mode p->e+pi0. No candidates were found which is consistent for almost zero background (0.1 events). This limit is almost twice what the current world average is (excluding SK, of course).
The second proton decay paper will report on the search for
p->K+nu. Presently with 535 days of data, this partial
lifetime limit is about 6x1032 years (90% CL), just more
than all the worlds average.
With lifetimes in the 1032 to 1034 years if we
expect to observe any proton decay events it is imperative that we
little or no background events to confuse the signal. The primary
background will be any other contained event. By far the most
numerous is when atmospheric
neutrinos interact via charged current with the nucleons in the
water and some number pions are created along with the outgoing
lepton. It is these such reactions which can mimic the products from
decay the decay of a nuleon.