One of the most ambitious and extreme experiments on Earth opened at the South Pole in 2010. IceCube, a giant particle detector buried in the polar ice, captures elusive, high-energy species of neutrinos—fundamental particles that fly straight through almost everything they touch. The project, for which I am the principal investigator, aims to use neutrinos to study distant cosmic phenomena—particularly the mysterious, violent processes thought to produce the charged particles known as cosmic rays, which continually bombard Earth.

We expect IceCube to catch superenergetic neutrinos only rarely. The particles have almost no mass and no electric charge (which is why they seldom react with other particles), and they travel at nearly the speed of light. Once they arrive on Earth, most neutrinos do not stop to linger. They zip right through thousands of kilometers of rock—even through the planet's solid iron core—and continue on their way.

Because of these difficulties, we were not surprised that data taken during the first few years, while the detector was still under construction, turned up nothing extraordinary. But in 2012 that changed.

One day, during a routine conference call for team members, our screen lit up with patterns we had never seen before. The signals reflected two neutrinos carrying more than 1,000 times the energy of the most energetic neutrino ever produced by an accelerator on Earth. These wild neutrinos had almost a billion times the energy of the neutrinos that the sun regularly spits out. Clearly, they had come from some spectacularly energetic process occurring far from our planet.

Exhilaration spread through the room as we realized we were looking at something game changing. Capturing the whimsy of the moment, one of our graduate students nicknamed the two particles “Bert” and “Ernie,” after the Sesame Street characters (fun names are easier to keep straight than the long strings of numbers we usually assign to neutrino events).

It took us another year and a totally redesigned analysis of the same data to satisfy ourselves that these were indeed what they seemed: the first pixels of the first pictures of the distant neutrino universe. Since then, we have found 54 high-energy neutrinos in total—many of them given Muppet names, including one dubbed “Big Bird” that had an energy twice that of Ernie or Bert.

We are now trying to identify where in the sky these high-energy neutrinos came from and how they originated. Their suspected sources are extreme cosmic events such as supernovae and other stellar explosions called gamma-ray bursts—two phenomena rumored to give rise to cosmic rays. If we can definitively trace the high-energy neutrinos to these likely sources of cosmic rays, we will open a new frontier in our understanding of the physics behind the extraordinarily dramatic events that are thought to produce them.

Powerful particles
The Earth is bathed in cosmic rays arriving from outer space—IceCube detects 275 million of them every day. A cosmic ray is a charged particle, most often a proton, that has extremely high velocity and thus carries a lot of energy. More than a century after their discovery, the processes that create cosmic rays are still largely unknown. When they arrive at Earth, we cannot simply trace their trajectory backward to deduce where they came from, because they swerve about during their journey, deflected by galactic and intergalactic magnetic fields.

Luckily, however, theory suggests that cosmic rays also interact at their birthplaces with photons to produce neutrinos. And neutrinos do point back to where they started—they shun other matter so thoroughly that almost nothing can divert them from their path. So although we cannot pinpoint the origins of cosmic rays from the rays themselves, we can infer their birthplaces by analyzing the highly energetic neutrinos they presumably produced in their youth.

Of course, astronomers have some ideas about how cosmic rays are born, but we need data to help us confirm or discard those possibilities. It is thought likely, for example, that massive stars give off cosmic rays in their death throes. At the end of a large star's life, when its nuclear core can no longer support its mass, it collapses into a dense object called a neutron star or into an even denser black hole, from which nothing escapes. The collapse set off an incredibly bright explosion: a supernova. But it also converts large amounts of gravitational potential energy into kinetic energy—thrust for the acceleration of particles, presumably through shock waves.

Supernova remnants were proposed as a likely source of cosmic rays as early as 1934 by astronomers Walter Baade and Fritz Zwicky. Yet, 80 years later, astrophysicists still cannot agree whether the hypothesis is correct. It does strike many as plausible. About three stars in the Milky Way go supernova every century. And each supernova converts a reasonable fraction of the star's mass into fuel for particle acceleration. But we need more evidence to be certain that supernovae alone account for the steady flow of cosmic rays seen in the galaxy.

Extragalactic cosmic rays, which originate from beyond the Milky Way galaxy, generally pack even more energy than do cosmic rays coming from nearby. There must be some source more energetic than supernovae that creates them.

Gamma-ray bursts are a prime suspect. Even brighter than regular supernovae, gamma-ray bursts are somewhat mysterious but probably occur when stars of very high mass collapse under certain extreme conditions.

But there are competing ideas about the engines of extragalactic cosmic rays. It is possible they originate near active galactic nuclei—supermassive black holes that sit at the centers of their galaxies and consume matter voraciously. As clouds of dust or gas get pulled in toward such a black hole, some particles could be deflected outward at ultrahigh speeds and eventually reach us as cosmic rays.

To catch a neutrino
Spotting the neutrinos that point to the origins of cosmic rays is a game of long odds. To have any hope of seeing dozens or hundreds of these elusive particles within a few years, IceCube had to be gigantic. The experiment is now monitoring a cube of 100,000-year-old Antarctic ice that is 1.5 kilometers below the surface of the South Pole. The cube measures a full kilometer on each side.

Deep, old Antarctic ice is a perfect natural neutrino detector because it is pure, ultratransparent, and shielded from sunlight. When a neutrino does occasionally interact with atoms in the ice, a shower of charged particles radiate blue light, called Cherenkov radiation, that can travel hundreds of meters through the ice. IceCube's 5,160 optical sensors, encased in glass spheres and spaced evenly throughout the cube, are able to pick up these faint flashes.

The sensors chart, in exquisite detail, the light pool produced by the nuclear debris created when a single neutrino hits. This pattern reveals the neutrino's type (or “flavor”), energy and arrival direction. The energies of Ernie and Bert and the others that we have seen so far are about a peta–electron volt (PeV), or 1015 eV; Ernie and Bert were 1.07 PeV and 1.24 PeV, respectively. For comparison, the particle beams at the Large Hadron Collider at CERN near Geneva, the world's most powerful particle accelerator, are in the tera–electron volt (TeV), or 1012 eV, range, about a thousandth as energetic. Big Bird, Ernie and Bert are the most energetic neutrinos ever seen, by a wide margin. Each of them cast a light pool of roughly 100,000 photons spread over about six city blocks.

The PeV energies of these two neutrinos tell us something important: they must be part of some cosmic signal. Their energies are just too large to have been produced nearby.

Local neutrinos, in contrast, are a dime a dozen. Every six minutes, on average, IceCube detects a neutrino made in Earth's atmosphere when cosmic rays smack into hydrogen and oxygen nuclei. But neutrinos made in our own backyard tell us nothing about the nature of cosmic rays or other astrophysical phenomena.

Our analysis of the data generated by IceCube thus starts by screening out these distractions. From past experience, we know what kinds of light patterns garden-variety neutrinos make. We ignore those.

Any flashes that remain and correspond to PeV-energy neutrinos must come from the distant cosmos. They very well could have reached us from the same sources as cosmic rays. But there are other plausible, even more exotic explanations for these particles.

Some have suggested that they may be signatures of dark matter—the invisible material that seems to make up more than 80 percent of all matter in the universe [see “Mystery of the Hidden Cosmos,” by Bogdan A. Dobrescu and Don Lincoln]. If dark matter consists of very heavy particles having an average lifetime longer than the current age of the universe, then the occasional decay of a dark matter particle might produce the PeV-energy neutrinos that we observe.

Counting neutrinos
Before IceCube discovered PeV neutrinos, the search for cosmic neutrinos had focused almost exclusively on muon neutrinos and paid little regard for the other two flavors, known as electron and tau neutrinos. That's not because muon neutrinos are thought to be the most common variety of cosmic neutrino reaching us—that flavor just happens to be easier to spot in our detector. When muon neutrinos slam into atoms, they launch kilometer-long light trails.

We originally optimized IceCube to pick up muon-neutrino trails that extend into the detector volume, even if the trails originate outside the cube. In effect, this technique allows us to expand our observation volume. But it comes with a trade-off. The approach also increases the risk that particles other than cosmic neutrinos will contaminate the result. Screening out background noise thus becomes harder.

So we also ran a second analysis that homed in on just the inner half of the IceCube detector, a strategy that leaves less room for contamination to get in. The great advantage of confining the search in this way is that the detector can then measure the full energy that each neutrino deposits in the ice to within 10 to 15 percent accuracy. That is a big improvement on the measurements we can make of neutrinos that interact outside the detector.

This second, more tightly constrained search hunted specifically for a class of extremely high energy neutrinos called Greisen-Zatsepin-Kuzmin (GZK) neutrinos. Theorists predict that when cosmic rays interact with photons from the cosmic microwave background, which was left over from the big bang, neutrinos could emerge having energies as high as exa–electron volts (EeV)—roughly 1018 eV.

Neither we nor anyone else have yet found GZK neutrinos. But IceCube's search for them has turned up plenty of cosmic neutrinos in all three flavors.

Since the discovery of Ernie and Bert, both our search strategies have succeeded in detecting cosmic neutrinos. Our first two years' worth of data revealed 28 neutrinos with energies between 30 and 1,200 TeV, including Ernie and Bert. This number is more than four standard deviations above what we would expect purely from the atmospheric background. Looked at another way, the probability is greater than 99.9999 percent that these particles truly came from deep space.

When we later added an additional two years of data, we brought the tally to 54 cosmic neutrinos. The statistical significance of the signal then climbed to well over five standard deviations, the conventional threshold for a “discovery.”

Exactly where in the universe do all these neutrinos point back to? We have not yet collected a large enough sample of events to answer that question conclusively. The origins do not seem to be restricted to our galaxy—the sky map indicating their arrival directions shows only marginal evidence for an overlap with the plane of the Milky Way. Most of the cosmic neutrinos are almost certainly of extragalactic origin.

There appears, however, to be a somewhat higher than average number of neutrinos coming from the center of the Milky Way. Bert is part of that cluster; it points back to within one degree of the galactic center. We cannot say for sure why this region is spewing out relatively high numbers of neutrinos, but we know the galactic center is packed with supernova remnants, as well as a giant black hole. Any of these are likely candidates for the neutrinos' source.

Our goal is to continually refine the map of where cosmic neutrinos originate as we steadily collect more muon-flavored neutrinos reaching us through Earth. The kilometer-length light trails they shed allow us to reconstruct their trajectories with better than 0.5-degree resolution. Over time, the accumulating data will reveal a highly detailed map of the sky in high-energy neutrinos—and therefore cosmic rays. Astronomers will be analyzing the map to find overlaps with known celestial objects that could be sources, such as gamma-ray bursts or bright galaxies that host supermassive black holes and active galactic nuclei.

IceCube is just beginning to scratch the surface of what it can discover. The experiment is built to run for 20 years—maybe more. In the meantime, we are looking toward its sequel. Our team is proposing to eventually build an expanded detector using roughly 10 cubic kilometers of ice—about 10 times the volume of IceCube. This larger instrument should collect thousands more cosmic neutrinos, enough to determine once and for all what distant powerhouses are creating them and their cosmic-ray cousins.