Novel Heavy-Water Detector Unveils the Missing Solar Neutrinos

To address that crucial question, the Sudbury Neutrino Observatory (SNO), a detector with novel capabilities, began observing solar neutrinos at the end of 1999 from deep inside a Canadian nickel mine north of Lake Huron. SNO is an international collaboration headed by Art McDonald of Queen’s University in Kingston, Ontario. Now, at last, we have the much anticipated first report of results from SNO’s first year of observations. ¹  

Like the 50-kiloton Super Kamiokande detector that has been operating under a mountain in Japan since 1996, SNO is an imaging water Čerenkov detector whose thousands of photomultiplier tubes can measure the energies and directions of electrons scattered by neutrinos bombarding the water. But the kiloton of heavy water (D₂O) at the heart of the new 8-kiloton detector gives SNO a unique ability to determine what has actually become of the missing solar electron neutrinos. (See the article by McDonald, John Bahcall, Frank Calaprice, and Yoji Totsuka in Physics Today July 1996, page 30.)

In brief, the first SNO results tell us that about $2 / 3$ of the most energetic electron neutrinos produced in the solar core have indeed metamorphosed into v_µ or v_τ , and that the total flux of neutrinos of all three flavors is in excellent agreement with the v_e flux predicted by the SSM in the absence of any neutrino oscillation. ²  

Not only do these results confirm the conviction that the astrophysics of the power plant at the core of the Sun is well understood. They also leave very little room for “sterile neutrinos.” There had been serious speculation that solar neutrinos were metamorphosing into some putative sterile neutrino species that does not participate in the standard weak interactions. The apparent absence of sterile neutrinos, in turn, calls into question the controversial 1995 report of flavor oscillations over short distances in an accelerator-based neutrino beam at Los Alamos (see Physics Today January 2001, page 16).

What’s so special about D₂O, aside from the fact that Canada has lots of it left over from its heavy-water nuclear reactor program? Solar neutrino energies do not extend much above 14 MeV. At these modest energies, the only way a neutrino can signal its presence in a traditional H₂O Čerenkov detector is by elastic scattering off an electron. And, indeed, one only gets sufficient Čerenkov light above background if the recoil energy of the struck electron exceeds a threshold of about 5 MeV.

The cross section for such elastic scattering by v _e is notoriously small. But it’s even smaller, by about a factor of 7, for v_µ and v_τ. That’s because a v _e can scatter off an electron by two different weak-interaction mechanisms: “neutral current” exchange (figure 1(a)) and “charged current” exchange (figure 1(b)). The v_µ and v_τ , by contrast, can only scatter off electrons by neutral-current exchange. In any case, if one has only elastic scattering data, one cannot disentangle the incident v _e flux from that of the other neutrino flavors.

That’s where the deuteron, the heavy hydrogen nucleus, comes to the rescue. Because it contains a weakly bound neutron, the deuteron can respond to solar neutrinos by breaking up in one of two useful ways: A neutrino of any flavor can break up the deuteron quasi-elastically,

leaving a free neutron whose subsequent capture can be recorded by the release of a telltale γ. Alternatively, a v _e can turn a deuteron’s neutron into a second proton, while turning itself into an electron detectable by its Cerenkov radiation. This kind of inverse-β-decay reaction

is a charged-current interaction (figure 1(c)) unavailable to the other neutrino flavors. Because both of these deuteron-breakup reactions have much larger cross sections than neutrino-electron elastic scattering, SNO can make do with a much smaller fiducial volume of water than Super Kamiokande.

The most obvious way of using SNO to find out what becomes of the missing solar neutrinos would be to compare the surviving v _e flux implied by the observed rate of reaction 2 with the total neutrino flux implied by the observed rate for reaction 1. (All the reaction cross sections necessary for turning observed rates into incident neutrino fluxes can be calculated from standard particle theory.) Because quasi-elastic deuteron breakup is a neutral-current reaction to which all three neutrino flavors contribute equally, it becomes a direct measure of all the neutrinos arriving from the Sun, irrespective of any flavor oscillation en route—provided, of course, that none of the solar neutrinos have been rendered sterile.

That straightforward comparison, however, will have to wait until later this year. Measuring reaction 1 requires a reliable way of detecting the capture of the neutrons freed in the reaction. To that end, highly purified NaCl has recently been added to the heavy water so that liberated neutrons will be captured by the chlorine nuclei, each such capture releasing an easily detected 8-MeV gamma.

But the data that have now been reported were taken before the salt was added, the philosophy being that one should start out as simply as possible. Any added ingredient is another potential source of background radioactivity. SNO is much more sensitive to radioactive backgrounds than Super Kamiokande. First of all, it only takes a 2.2-MeV γ to produce a spurious deuteron breakup. Furthermore, reaction 2 produces a fairly isotropic electron distribution not unlike the radioactive background. By contrast, the neutrino–electron elastic scattering that Super Kamiokande relies on produces a strongly peaked angular distribution of electrons. One can separate the elastic scattering events from radioactive backgrounds or other reactions simply by requiring that the electron be scattered within a few degrees of the direction from the Sun. Extraordinary measures are taken to keep SNO free of radioactive contamination. For example, everyone entering the cavernous underground laboratory must first shower and put on a clean-room garment.

So how is it that SNO can tell us so much about the missing solar neutrinos even before it acquires its full capacity to detect neutrons? Until there’s adequate data on reaction 1, the SNO collaboration can make do with Super Kamiokande’s enormous harvest of neutrino–electron elastic scattering events. ³ Because v_µ and v_τ can participate only via neutral-current interaction, the total rate for v–e elasic scattering is proportional to the flux φ _e of electron neutrinos plus about one-seventh of φ_µτ , the combined flux of v_µ and v_τ.

This admixture is not very informative about the flavor of the missing solar neutrinos until one compares it to the SNO rate for reaction 2, which, by itself, tells us only about φ_e. If the incident neutrino flux implied by the rate of elastic scattering at Super Kamiokande turns out to exceed that implied by the rate for reaction 2 observed at SNO, the excess can be attributed to the other neutrino flavors, presumably produced by flavor oscillation during the journey from the solar core to the detector.

Figure 2 shows just such an excess for recoil electron energies above the provisional SNO threshold of 6.75 MeV. Each observed reaction rate, plotted against electron energy, is shown as a fraction of the rate predicted by the SSM in the absence of any flavor oscillation. As measured by Super Kamiokande, the elastic scattering rate is about 47% of the solar-model prediction, with the shortfall showing no discernible energy dependence. By contrast, the SNO data for reaction 2 come to only about 35% of the SSM prediction, also without any obvious energy dependence (The electron energy approximates the incident neutrino energy after one corrects for the small deuteron binding energy and mass changes in reaction 2.)

If the missing solar neutrinos had somehow disappeared entirely or become sterile, the rates for both processes would fall short of the SSM prediction by the same fraction. But, in fact, the two shortfalls turn out to differ by 3.3 standard deviations.

A direct comparison of the SNO and Super Kamiokande data, quite independent of any SSM predictions, translates this discrepancy into the separate flavor fluxes φ _e and φ_µτ. In figure 3, the orange and red bands represent the SNO and Super Kamiokande results in terms of these neutrino fluxes. Where they cross, one has a consistent pair of integrated fluxes above the threshold energy. The best fit to the combined data yields a value of 68% for the fraction of electron neutrinos leaving the solar core that arrive at the detectors with their flavors changed either to v_µ or v_τ. The data do not distinguish between v_µ and v_τ.

The blue diagonal swath represents the total flux φ _e + φ_µτ one gets by combining the SNO and Super Kamiokande data. It overlaps, almost perfectly, the total v _e flux predicted by the SSM for neutrino energies above SNO threshold, vindicating the solar model that Bahcall and others have been refining and testing since the mid-1960s, when pioneer Ray Davis was planning to build the first radio-chemical solar neutrino detector.

Irrespective of solar models, the new SNO result is the first direct evidence for a large component of active non-electron neutrinos at the high-energy end of the solar neutrino spectrum.

Neutrino oscillation implies neutrino mass. If, as it now seems, only the three active neutrino flavors are involved in the oscillation of solar neutrinos and of atmospheric neutrinos produced by cosmic rays, then one can deduce a lower limit for the neutrino mass density of the universe. That lower limit, the SNO paper tells us, is about 2% of the cosmic density of ordinary baryonic matter.