Résumé

Neutrinos cosmiques, et autres : les trois premières années d’IceCube

Depuis quatre ans, l’Observatoire de neutrinos IceCube, situé au pôle Sud, récolte des données issues des collisions les plus violentes de l’Univers. Répondant aux attentes, le détecteur a observé des neutrinos provenant d’au-delà du Système solaire à des énergies dépassant 60 TeV, avec les 5 σ magiques permettant de valider un résultat. Ces neutrinos ne sont qu’un aspect du vaste programme de physique d’IceCube, qui inclut la détection de neutrinos astrophysiques, de ceux issus de la matière noire, l’étude des oscillations de neutrinos, la physique des rayons cosmiques et la recherche de supernovas. Toutes ces études reposent sur un détecteur d’exception situé à un endroit d’exception : le pôle Sud.

For the past four years, the IceCube Neutrino Observatory, located at the South Pole, has been collecting data on some of the most violent collisions in the universe. Fulfilling its pre-construction aspirations, the detector has observed astrophysical neutrinos with energies above 60 TeV, at the "magic" 5σ significance. The most energetic neutrino observed had an energy of about 2 PeV (2 × 1015 eV) – 250 times higher than the beam energy of the LHC.

These neutrinos are just one highlight of IceCube’s broad physics programme, which encompasses searches for astrophysical neutrinos, searches for neutrinos from dark matter, studies of neutrino oscillations, cosmic-ray physics, and searches for supernovae and a variety of exotica. All of these studies take advantage of a unique detector at a unique location: the South Pole.

IceCube observes the Cherenkov light emitted by charged particles produced in neutrino interactions in 1 km3 of transparent Antarctic ice. The detector is the ice itself, and is read out by 5160 optical sensors. Figure 1 shows how the optical sensors are distributed throughout the 1 km3 of ice, 1.5 km beneath the geographic South Pole. They are deployed 17 m apart, on 86 vertical cables or "strings". Seventy-eight of the strings are spaced horizontally, 125 m apart in a grid of equilateral triangles forming a hexagonal array across an area of a square kilometre. The remaining eight strings form a more densely instrumented sub-array called DeepCore. In DeepCore, most of the sensors are concentrated in the lower 350 m of the detector.

Each sensor, or digital optical module (DOM), is like a miniature satellite made up of a 10 inch (25 cm) photomultiplier tube together with data-acquisition and control electronics. These include a custom 300 megasample/s waveform digitizer with 14 bits of dynamic range, plus light sources for calibrations, all consuming a power of less than 5 W. The hardware is protected by a centimetre-thick pressure vessel.

The ice in IceCube formed from compacted snow that fell on Antarctica 100,000 years ago. Its properties vary with depth, with layers reflecting the atmospheric conditions when the snow first fell. Measuring the optical properties of this ice has been one of the major challenges of IceCube, involving custom "dust loggers", studies with LED "flashers" and cosmic-ray muons. During the past decade, the collaboration has found that the ice is layered, that the layers are not perfectly flat and, most recently, that the light scattering is somewhat anisotropic. Each insight has led to a better understanding of the detector and to smaller systematic uncertainties. Fortunately, advances in computing technology have allowed IceCube’s simulations to keep up, more or less, with the increasingly complex models of light propagation in the ice.

The distributed sensors give IceCube strong pattern-recognition capabilities. The three neutrino flavours – νe, νμ and ντ – each leave different signatures in the detector. Charged-current νμ produce high-energy muons, which leave long tracks. All νe interactions, and all neutral-current interactions, produce hadronic or electromagnetic showers. High-energy ντ produce a characteristic "double-bang" signature – one shower when the ντ interacts and a second when the τ decays. More complex topologies have also been studied, including tracks that start in the detector as well as pairs of parallel tracks.

Despite past doubts, IceCube works and works well. More than 98% of the sensors are fully operational, and another 1% are usable – most of the failures occurred during deployment. The post-deployment attrition rate is a few DOMs per year, so IceCube will be able to operate for as long as required. The "live" times are also impressive – in the range of 99%.

IceCube has excellent reconstruction capabilities. For kilometre-long muon tracks, the angular resolution is better than 0.4°, verified by studying the shadow of the Moon cast by cosmic rays. For high-energy contained events, the angular resolution can reach 15°, and at high energies the visible energy can be determined to better than 15%.

Cosmic neutrinos

The detector’s dynamic range covers from 10 GeV to infinity. The higher energy the neutrino, the easier it is to detect. Every six minutes, IceCube records an atmospheric neutrino, from the decay of pions, kaons and heavier particles produced in cosmic-ray air showers. These 100,000 neutrinos collected every year are interesting in their own right, but they are also the background to any search for cosmic neutrinos. On top of this, the detector records about 3000 atmospheric muons every second. This is a painful background for neutrino searches, but a gold mine for cosmic-ray physics.

Although IceCube has an extremely rich physics programme, the centrepiece is clearly the search for cosmic neutrinos. Many signatures have been proposed for these neutrinos: point source searches, a high-energy diffuse flux, identified ντ, and others. IceCube has looked for all of these.

Point-source searches are the simplest strategy conceptually – just create a sky map showing the arrival directions of all of the detected neutrinos. Figure 2 shows the IceCube sky map containing 400,000 events gathered across four years (Aartsen et al. 2014c). In the southern hemisphere, the large background of downgoing muons is only partially counteracted by selecting high-energy muons, which are less likely to be of atmospheric origin. The 177,544 events in the northern-hemisphere sample are mostly from νμ. So far, there is no statistically significant evidence for any hot spots, even in searches for spatially extended sources. IceCube has also looked for variable sources, whether episodic or periodic, with similar results. These limits constrain theoretical models, especially those involving gamma-ray bursts.

If there are enough weak sources in the cosmos, they should be visible as an aggregate, diffuse flux. This diffuse flux is expected to have a harder energy spectrum than do atmospheric neutrinos. Calculations have indicated that IceCube would be more sensitive to this diffuse flux than to point sources, which is indeed the case. Several early searches, using the partially completed detector, turned up intriguing hints of an excess over the expected atmospheric neutrino flux. Then the search diverged from the anticipated script.

One of the first searches for diffuse neutrinos with the complete detector looked for ultra-high-energy cosmogenic neutrinos – neutrinos produced when ultra-high-energy cosmic-ray protons (E > 4 × 1019 eV) interact with photons of around 10–4 eV in the cosmic-microwave background, exciting them to a Δ+ resonance. The decay products of the pion produced in the Δ’s decay include a neutrino with a typical energy of 1018 eV (1 EeV). The search found two spectacular events, one of which is shown in figure 3. Both events were well contained within the detector – clearly neutrinos. Both had energies around 1 PeV – spectacular, but too low to be produced by cosmic rays interacting with CMB photons. Such events were completely unexpected.

Inspired by these events, the IceCube collaboration instigated a follow-up search that used two powerful techniques (Aartsen et al. 2013). The first was a filter to identify neutrino interactions that originate inside the detector, as distinct from events originating outside it. The filter divides the instrumented volume into an outer-veto shield and a 420 megatonne inner active volume. Figure 4 shows how this veto works: by rejecting events with significant in-time energy deposition in the veto region, neutrino interactions within the detector’s fiducial volume can be separated from backgrounds. For neutrinos that are contained within the instrumented volume of ice, the detector functions as a total absorption calorimeter, measuring energy with 15% resolution. It is flavour-blind, equally sensitive to hadronic or electromagnetic showers and to muon tracks. This veto analysis also used a "tagging" approach to estimate the atmospheric-muon background using the data, rather than relying on simulations. Because of the veto, the analysis could observe neutrinos from all directions in the sky.

The second innovation was to take advantage of the fact that downgoing atmospheric neutrinos should be accompanied by a cosmic-ray air shower depositing one or more muons inside IceCube. In contrast, cosmic neutrinos should be unaccompanied. A very high-energy, isolated downgoing neutrino is highly likely to be cosmic.

The follow-up search found 26 additional events. Although no new events had an energy near 1 PeV, the analysis produced evidence for cosmic neutrinos at the 4σ level. To clinch the case, the collaboration added a third year of data, pushing the significance above the "magic" 5σ level (Aartsen et al. 2014a). One of the new events had an energy above 2 PeV, making it the most energetic neutrino ever seen.

The observation of a flux of cosmic neutrinos was soon confirmed by the independent and more traditional analysis recording the diffuse flux of muon neutrinos penetrating the Earth. Both observations are consistent with a diffuse flux composed equally of the three neutrino flavours. No statistically significant hot spots were seen. The observed flux is consistent with that expected from cosmic accelerators producing equal energies in gamma rays, neutrinos and, possibly, cosmic rays.

Newer studies are shedding more light on these events, extending contained-event studies down to lower energies and adding flavour identification. At energies above 10 TeV, the astrophysical neutrino flux can be fit by a single power-law spectrum that is significantly harder than the background cosmic-ray muon spectrum: φν = 2.06+0.4–0.3 × 10–18 ( Ev / 100TeV ) –2.46±0.12 GeV–1 cm–2 sr–1 s (Aartsen et al. 2014d).

Within the limited statistics, the flux appears isotropic and consistent with the νeμτ ratio of 1:1:1 that is expected for cosmic neutrinos. The majority of the events appear to be extragalactic. Some might originate in the Galaxy, but there is no compelling statistical evidence for that at this point.

Many explanations have been proposed for the IceCube observations, ranging from the relativistic particle jets emitted by active galactic nuclei to gamma-ray bursts, to starburst galaxies to magnetars. IceCube’s dedicated searches do, however, disfavour gamma-ray bursts as the source. A spectral index of –2 (dNν/dE ~ E–2), predicted by Fermi shock-acceleration models, is also disfavoured, but many other scenarios are possible. Of course, the answer is clear: more data are needed.

Other physics

The 100,000 neutrinos and 85 × 109 cosmic-ray events recorded each year provide ample opportunities to search for dark matter and to study cosmic rays as well as neutrinos themselves. IceCube has measured the cosmic-ray spectrum and composition and observed anisotropies in the spectrum at the 10–4 level that have thus far defied explanation. It has also studied atypical events, such as muon-free showers expected from photons with peta-electron-volt energies, produced in the Galaxy, and investigated isolated muons produced in air showers. The latter have separations that shift from an exponential decrease to a power-law separation spectrum, as predicted by perturbative QCD.

IceCube observes atmospheric neutrinos across an energy range from 10 GeV to 100 TeV – at higher energies, the atmospheric flux is swamped by the flux of cosmic neutrinos. As figure 5 shows, the flux is consistent with expectations across a large energy range. Lower-energy neutrinos are of particular interest because they are sensitive to neutrino oscillations. For neutrinos passing vertically through the Earth, the νμ flux develops a first minimum at 28 GeV.

Figure 6 shows the observed νμ flux, seen in one year of data, using well-reconstructed events contained within DeepCore. The change in flux with distance travelled/energy (L/E) is consistent with neutrino oscillations and inconsistent with a no-oscillation scenario. IceCube constraints on the mixing angle θ23 and |Δm232| are comparable to constraints from other experiments.

IceCube also searched for neutrinos from dark-matter annihilation. Dark matter can be gravitationally captured by the Earth, the Sun, or in the centre or halo of the Galaxy. It then accumulates and the dark-matter particles annihilate, producing neutrinos. IceCube has searched for signatures of this annihilation, and has set limits. The Sun is a particularly interesting option, producing a characteristic dark-matter signature that cannot be explained by any astrophysical scenario. It is also mostly protons, allowing IceCube to set the world’s best limits on the spin-dependent cross-section for the interaction of dark-matter particles with ordinary matter.

The collaboration has also looked for even more exotic signatures, such as magnetic monopoles and pairs of upgoing particles. One particularly spectacular and interesting signature could come from the next supernova in the Galaxy. These explosions produce a blast of neutrinos with 10–50 MeV energy. This energy level is far too low to trigger IceCube directly, but the neutrinos would be visible as a collective increase in the singles rate in the buried IceCube photomultipliers. Moreover, IceCube has a huge effective area, which will allow measurements of the time structure of the supernova-neutrino pulse with millisecond precision.

IceCube is still a novel instrument unlikely to have exhausted its discovery potential. However, at high energies, it might not be big enough. Doing neutrino astronomy could require samples of 1000 or more, high-energy neutrino events. In addition, some key physics questions require a detector with a lower energy threshold. These two considerations are driving two different upgrade projects.

DeepCore has demonstrated that IceCube is capable of making precise measurements of neutrino-oscillation parameters. If precision studies can be extended to neutrino energies below 10 GeV, it will be possible to determine the neutrino-mass hierarchy. Neutrinos passing through the Earth interact coherently with matter electrons, modifying the oscillation pattern in a way that differs for normal and inverted hierarchies. In addition to a threshold of a few giga-electron-volts, this measurement requires improved control of systematic uncertainties. An expanded collaboration has come together to pursue the construction of a high-density infill array called Precision In Ice Next-Generation Upgrade, or PINGU (Aartsen et al. 2014b). The present design consists of 40 additional high-sensitivity strings equipped with improved calibration devices. PINGU should be able to determine the mass hierarchy with 3σ significance within about three years, independent of the value of the CP-violation phase.

The IceCube high-energy extension (IceCube-gen2) aims for a detector with a 10-times-larger instrumented volume, albeit with a higher energy threshold. It will explore the observed cosmic neutrino flux and pin down its origin. With a sample of more than 100 cosmic neutrinos per year, it will be possible to observe multiple neutrinos from the same sources, and so do astronomy. The instrument will also have an improved sensitivity to study the ultra-high-energy neutrinos produced in the interactions of cosmic rays with microwave photons.

Of course, IceCube is not the only collaboration studying high-energy neutrinos. Projects on the cubic-kilometre scale are also being prepared in the Mediterranean Sea (KM3NeT) and in Lake Baikal (GVD), with a field of view complementary to that of IceCube. Within KM3NeT, ORCA, a proposed low-threshold detector, would pursue the same physics as PINGU. And the radio-detection experiments ANITA, ARA, GNO and ARIANNA are beginning to explore the neutrino sky at energies above 1017 eV.

After a decade of construction, the completed IceCube detector came on line in December 2010. It has achieved the outstanding goal of observing cosmic neutrinos and has produced important results in diverse areas: cosmic-ray physics, dark-matter searches and neutrino oscillations, not to mention its contributions to glaciology and solar physics. The observation of cosmic neutrinos at the peta-electron-volt energy scale has attracted enormous attention, with many suggestions about the location of the requisite cosmic accelerators.

Looking ahead, IceCube anticipates two important extensions: PINGU, which will determine the neutrino-mass hierarchy, and IceCube-gen2, which will expand a discovery instrument into an astronomical telescope.