2015 Balzan Prize for Astroparticle Physics Including Neutrino and Gamma-ray Observation
Icebound Neutrinos: Bern, 12.11.2015 Forum (video + text)
IceCube and the Discovery of Cosmic Neutrinos
I met Fred Reines, Nobel laureate for the discovery of the neutrino in 1956, on visits to the University of California-Irvine, home institution of my earliest companions on the long march toward IceCube, the largest neutrino detector ever built. With Clyde Cowan, he had detected neutrinos produced in the Savannah River nuclear reactor some 25 years after Wolfgang Pauli’s bold proposal, which introduced the new particle to rescue energy conservation in the beta decay of nuclei. Reines told me that as soon as they had demonstrated that neutrinos were real particles, literally everybody came up with the idea that neutrinos represented the ideal astronomical messengers. Having essentially no mass and no electric charge, the neutrino is much like a photon; it differs in one important attribute—its interactions with matter are extremely feeble. Neutrinos go through walls; light does not. In fact, neutrinos can escape unscathed from the inner neighborhoods of black holes and from the accelerators where cosmic rays are born.
In 1960, Kenneth Greisen, Moisey Markov and Reines himself published papers in which they developed the concept of neutrino astronomy in more detail. Its feasibility has since been demonstrated: neutrino detectors have “seen” the Sun and have detected a supernova in the Large Magellanic Cloud in 1987. Both observations were of tremendous importance; the former
showed that neutrinos have a tiny mass, opening the first chink in the armor of the Standard
Model of particle physics, and the latter confirmed the basic nuclear physics of the death of stars.
Because of the neutrino’s weak interaction with ordinary matter, neutrino detectors must be immense in order to collect statistically significant numbers. Already by the 1970s, it had been understood that a detector would need to be of kilometer-scale to observe Greisen-Zatsepin- Kuzmin (GZK) neutrinos. These are produced throughout the universe in interactions of cosmic rays with background microwave photons. With the completion of IceCube in 2010, we transformed a cubic kilometer of the Antarctic ice shelf into such a neutrino detector and thus created the first opportunity to observe GZK neutrinos, with an anticipated frequency of one to two events per year .
Detecting GZK neutrinos is far from guaranteed. Uncertainties associated with our understanding of the cosmic-ray flux complicate the matter—a century after the discovery of cosmic rays, their origin is a mystery. On the other hand, a handful of GZK neutrinos might be all it takes to finally reveal the sources of cosmic rays. We will show that neutrinos must be produced in association with cosmic rays and, unlike the electrically charged cosmic rays themselves, whose directions are scrambled by Galactic and intergalactic magnetic fields, neutrinos point back to the sources where they originated.
Of course, IceCube does not detect neutrinos directly, since neutrinos have no electric charge. A typical GZK neutrino will travel right through the Earth, and through IceCube, leaving no trace. The unlucky one—about one in a hundred thousand—will crash head-on into an atomic nucleus in the ice, producing a shower of charged particles that IceCube will spot from the blue light they radiate. This radiation, dubbed Cherenkov radiation, has been extensively studied since the mid-1930s and has been widely used as a detection technique since then. It is the same eerie blue
glow produced by radiation in the water shielding nuclear reactors. The Cherenkov light travels hundreds of meters through the pure ultratransparent ice, compacted snow that fell on Antarctica some hundred thousand years ago. It is detected by IceCube’s 5,160 optical sensors embedded within a cubic kilometer of ice between 1.5 and 2.5 kilometers below the surface; see Figure 1. The sensors chart in exquisite detail the light pool produced by the nuclear debris of a single neutrino interaction. This pattern reveals the neutrino’s type (or “flavor,” as it is called), energy, and arrival direction. For instance, the enormous energy deposited in the detector by an EeV-energy GZK neutrino interaction1 results in a flash of light that fills a good fraction of IceCube; it is hard to miss. A scan of the first two years of data taken with the completed instrument revealed two particularly interesting events, their detector event displays are shown in Figure 2; for a discussion see reference .
Figure 1: Sketch of the IceCube observatory and a digital optical module.
When first scrutinizing the experiment’s display of these events, two thoughts immediately came to mind: “I have never seen anything like this before,” closely followed by “This is not what we were looking for.” It turned out, upon further investigation, that the energies of these two neutrinos, rather than super-EeV, as expected for GZK neutrinos, were in the PeV range: 1,070 TeV and 1,240 TeV. Nonetheless, the energies of these neutrinos exceeded by an order of magnitude those of the highest energy neutrinos previously detected by IceCube. They were particle showers initiated by neutrinos that interacted inside the instrumented detector volume. Their light pool of roughly one hundred thousand photons extended over more than 500 meters, or about five city blocks. For context, the energies exceed those of neutrinos from the sun and from supernova 1987A by a factor of close to one billion; also, no photon with this energy has ever been detected.
If they are not GZK neutrinos, then what is special about PeV-energy neutrinos? They are too energetic to be produced in the Earth’s atmosphere but must have reached us from some accelerator beyond. Neutrinos produced in the atmosphere are a dime a dozen. Every six minutes, IceCube detects a neutrino that is produced in the interactions of cosmic rays with hydrogen and oxygen nuclei in the Earth’s atmosphere. As in conventional astronomy, we therefore have to look through the atmosphere to do astronomy. IceCube has an advantage, though—we have calibrated our detector by measuring the atmospheric neutrino flux with a statistic of hundreds of thousands of neutrino events by now. The key is that, as the neutrino energy increases, the atmospheric flux diminishes precipitously, leaving no more than a single event per year in the energy range above a few hundred TeV. This is why the PeV neutrinos stood out on the event display; while we had scrutinized in detail high-energy atmospheric neutrinos, these events clearly looked different. We had spotted our first hint of cosmic neutrinos.
Theorists have speculated widely, and wildly, on the origin of PeV-energy neutrinos, including that they may be signatures of dark matter. The educated guess at this point is that IceCube is observing neutrinos that, although not GZK neutrinos, originate in the same cosmic accelerators that produce cosmic rays. Although IceCube was designed as a discovery instrument, detecting neutrinos associated with the sources of high-energy cosmic rays has been its most conspicuous science mission. Cosmic accelerators produce particles with energies in excess of 100 EeV; we still do not know from where or how. The bulk of cosmic rays are Galactic in origin, but any association with our Galaxy presumably disappears at EeV energy when the gyroradius of a proton in the Galactic magnetic field exceeds its size. The detailed blueprint for a cosmic-ray accelerator must meet two challenges: the highest-energy particles in the beam must reach beyond 103 TeV for Galactic (108 TeV for extragalactic) sources, and their luminosities must be able to accommodate the observed flux. Both requirements represent severe constraints that have guided (and confined!) theoretical speculations.
Supernova remnants were proposed as a likely source of cosmic rays as early as 1934 by Walter Baade and Fritz Zwicky; after 80 years the issue is still debated. Three Galactic supernova explosions per century, converting a reasonable fraction of a solar mass into particle acceleration, can accommodate the steady flux of cosmic rays in the Galaxy.
Energetics has similarly guided speculations on the origin of extragalactic cosmic rays. A gamma-ray-burst fireball converts a fraction of a solar mass into the acceleration of electrons, seen by astronomers as synchrotron photons. The observed energy in extragalactic cosmic rays can be accommodated with the reasonable assumption that shocks in the expanding fireball convert a roughly equal amount of energy into the acceleration of cosmic rays. Problem solved? Not really: it turns out that the same outcome can be achieved assuming that active galactic nuclei convert similar amount of energies into photons and cosmic rays. In the end, whether gamma-ray bursts or active galaxies are the sources, the observation that cosmic-ray accelerators radiate similar amounts of energy in photons and cosmic rays may not be an accident.
Neutrinos must be produced at some level in association with the cosmic-ray beam. Cosmic rays accelerated in regions of high magnetic fields near black holes or neutron stars inevitably interact with radiation surrounding them. Thus, cosmic-ray accelerators are what particle physicists refer to as beam dumps. For instance, cosmic rays accelerated in supernova shocks interact with gas in the Galactic disk, producing equal numbers of pions of all three charges that decay into pionic photons and neutrinos. High-density molecular clouds, ubiquitous in the star-forming regions where supernovae are more likely to explode, represent a target for efficiently converting cosmic rays into neutrinos. For extragalactic sources, the neutrino-producing target may be electromagnetic, for instance photons radiated by the accretion disk of an active galaxy, or synchrotron photons that coexist with protons in the expanding fireball producing a gamma-ray burst. A more detailed description of the theoretical estimates can be found in reference 
In Figure 3 we compare estimates of astrophysical neutrino fluxes with IceCube measurements of the atmospheric neutrino flux, the background for observing a cosmic signal. The shaded band indicates the level of expectations for high-energy neutrinos of astrophysical origin. At this intensity, neutrinos from theorized cosmic-ray accelerators will cross the steeply falling atmospheric neutrino flux above an energy level of a few 100 TeV. PeV events are not atmospheric, thus the excitement surrounding the serendipitous discovery of the two PeV events. It is interesting to note that the generic astrophysical flux in Figure 3 predicts 10–100 neutrinos per year. This had been our benchmark when proposing the IceCube concept.
Traditionally, searches for cosmic neutrinos had almost exclusively focused on the observation of muon neutrinos that interacted primarily outside IceCube’s boundaries to produce kilometer- long muon tracks passing through its instrumented volume. Although creating the opportunity to observe neutrinos interacting outside the detector, it is then necessary to use the Earth as a filter to remove the huge background of muons produced by cosmic ray interactions in the atmosphere. This limits the neutrino view to a single flavor and half the sky. Inspired by the two PeV events, we instead designed a filter that exclusively identifies neutrinos interacting inside the detector [2,3]. It divides the instrumented volume of ice into an outer veto shield and a 420- megaton inner active volume. The great advantage of specializing to neutrinos interacting inside the instrumented volume of ice is that the detector functions as a total absorption calorimeter capable of measuring energy with a 10–15% resolution. Also, one can identify neutrinos from all directions in the sky, including both muon tracks produced in muon-flavored neutrino interactions as well as secondary showers produced by neutrinos of the other flavors, electron and tau.
Thanks to the excellent performance of the detector, finding additional events that start within the detector turned out to be relatively straightforward. This is illustrated in Figure 4, showing one year of data in a 3D display. The vertical axis shows the number of events as a function of the total number of photons in the event and the number present in the veto region. In the bottom right of the figure, nine events with reconstructed neutrino energies in excess of 100 TeV are clearly separated from the atmospheric neutrino background. At best, one to two of these are expected based on the well-measured atmospheric neutrino background; the straightforward conclusion is that most of these events are extraterrestrial in origin.
The actual analysis, which was done blind, defined a signal region selecting events with more than 6,000 photoelectrons, and with fewer than three of the first 250 in the veto region. The signal box is shown as the shaded box in Figure 4. The separation between veto and signal regions was optimized to reduce the background of atmospheric muons and neutrinos to about five events per year each, while keeping 98% of a possible cosmic signal. Applied to the same data sample used in the previously discussed GZK neutrino search, the analysis revealed 28 neutrino events with in-detector deposited energies between 30 and 1,200 TeV. Of these, 21 are showers with an energy reconstruction of better than 15%, but a poor angular resolution of about 10 to 15 degrees. The remaining seven are muon events, which allow for subdegree angular reconstruction; they are of course difficult to separate from the competing atmospheric background. The 28 events include the two PeV events previously identified. The signal represents an excess over background of more than 4 standard deviations, meaning a probability greater than 99.9999% that they do not represent atmospheric neutrinos.
Fitting the data to a combination of an extraterrestrial flux and an atmospheric background yields a cosmic flux for the sum of the 3 three neutrino flavors of
E 2 dNν ~ (3.6 ±1.2) ×10−8 GeVcm−2 s−1sr−1.
This is the magnitude and energy dependence of the neutrino flux anticipated from sources that potentially accelerate cosmic rays; see Figure 3. For historical context, the appearance of this flux follows a fifteen-year-long effort initiated by the first incarnations of the AMANDA detector, IceCube’s predecessor. The observed flux of cosmic neutrinos finally appeared at a
level three orders of magnitude below upper limits established by underground neutrino detectors commissioned inside traffic tunnels in Frejus, France, and Gran Sasso, Italy in the mid-nineties at the time of our first deployments at the South Pole.
So, finally, where do the cosmic neutrinos come from? The 28 events discussed above do not provide a conclusive answer. A sky map in Galactic coordinates indicating their arrival directions is shown in Figure 5. Not all of these events are Galactic in origin, with many reconstructing far off the plane of our Galaxy corresponding to the major axis of the ellipse. An apparent hot spot does appear at a right ascension of 281 degrees and a declination of 23 degrees, close to the galactic center. However, its post-trial probability after correcting for the look- elsewhere effect is only 8%.
In the meantime, two additional years of data were analyzed, doubling the statistics of the discovery publication. A purely atmospheric explanation can be excluded at 7 standard deviations . The four-year data set contains a total of 54 neutrino events with deposited energies ranging from 30 to 2000 TeV. In the fourth year, muon neutrinos were found that deposited ~500TeV energy inside the detector indicating PeV-energy parent muon neutrinos. One of them reconstructs through IceCube’s surface array with no evidence for an air shower. Combining the absence of an air shower with the large energy deposited by the neutrino results in a high significance for astrophysical origin from a single event!
In parallel, an independent analysis of high-energy muon neutrinos penetrating the Earth from the Northern Hemisphere also revealed an excess of cosmic over atmospheric background, confirming the signal observed in the analyses of events originating inside the detector .
Figure 6 shows the muon neutrino flux as a function of the energy deposited by the muons inside
the detector. Standard Model physics allows one to infer the energy spectrum of the parent neutrinos; for instance, the highest energy muon tracks in Figure 6 correspond, on average, to parent neutrinos with energy in excess of PeV. A best fit to the spectrum that includes a conventional atmospheric, charm, and astrophysical component with free normalizations yields the results shown in the figure. With one additional year of data, the statistical significance exceeds 4.3 standard deviations. In the 2014 sample, an event was found with energy 2.6 ± 0.3
PeV representing the highest energy neutrino ever recorded with estimated energy of ~ 9 PeV.
Figure 7 shows the arrival directions of the highest energy upgoing muon events together with events of the three-year starting-event sample in equatorial coordinates. No significant local excess is found when compared to randomized pseudo-experiments. The correlation of neutrino events with the Galactic plane is not significant. Letting the width of the plane float freely, the best fit returned a correlation for a value of 7.5 degrees with a post-trial chance probability of 3.3%. Neither probability decreased after doubling the data. We also searched for clustering of the events in time and investigated a possible correlation with the times of observed GRBs. No statistically significant correlation was found.
In summary, the cosmic neutrino flux observed is consistent with an isotropic distribution of arrival directions and equal contributions of all neutrino flavors, suggesting the observation of extragalactic sources whose flux has equilibrated in the three flavors after propagation over cosmic distances. Increasingly, a variety of analyses suggest that the cosmic neutrino flux dominates the atmospheric background above an energy that may be as low as 30 TeV with an energy spectrum that is not described as a single power.
Various astrophysical scenarios have been suggested that might be (partially) responsible for the observed flux of neutrinos. The absence of significant signs of anisotropy in the data is consistent with an extragalactic population of sources. Source candidates include galaxies with intense star formation, cores of active galactic nuclei (AGN), low-luminosity AGN, blazars, low-power GRBs, cannonball GRBs, intergalactic shocks, and galaxy clusters.
Galactic contributions are in general identifiable by anisotropies in the arrival direction of neutrinos. Neither data set shows evidence for this, but a Galactic component might be hidden by the limited event statistics and the poor angular resolution of cascade events. The most energetic neutrinos detected so far, with energies in the PeV range, should be produced by cosmic ray nucleons with energy of 10-100 PeV. This corresponds to a source population contributing to the spectrum between the knee and ankle; the latter has been traditionally associated with a transition between Galactic and extragalactic source populations. Consequently, speculations on Galactic sources have persisted. Possible contributions to super-TeV neutrinos are the diffuse neutrino emission of galactic cosmic rays, the joint emission of galactic PeV sources or microquasars, and extended galactic structures like the Fermi Bubbles or the galactic halo. A possible association with the sub-TeV diffuse galactic gamma-ray emission and constraints from the non-observation of diffuse galactic PeV gamma-rays, have also been investigated. More exotic scenarios have suggested a contribution of neutrino emission from decaying heavy dark matter.
Rather than speculate, it may be more useful to focus on the multimessenger connection of cosmic neutrinos to cosmic rays and gamma rays. The overall energy density of the observed neutrino flux matches the energy produced by the sources of ultrahigh energy extragalactic cosmic rays, whatever they may be. This may be a coincidence, but it can also indicate a multimessenger relation pointing at beam dumps with very efficient neutrino production in sources with optical thickness of order unity. These are often referred to as cosmic ray calorimeters or reservoirs such as starburst galaxies or galaxy clusters.
The production of PeV neutrinos is inevitably associated with the production of PeV gamma rays, hadronic accelerators produce fluxes of both neutral and charged pions that are the parents of gamma rays and neutrinos, respectively. The production rates Q of neutrinos and gamma rays are related by ……
Note that this relation does not depend on the pion production efficiency but only on the relative charged-to-neutral pion rate Kπ. On the other hand, the production rate described by the equation is not necessarily the emission rate observed. For instance, in hadronic sources that efficiently produce neutrinos via pp interactions, the target photon field can also efficiently reduce the pionic gamma rays via pair production. This is a calorimetric process that will, however, conserve the total energy of hadronic gamma rays.
It is straightforward to apply the multimessenger relation to the cosmic neutrino flux observed by IceCube. Figure 8 shows the associated gamma ray flux for two illustrative pp emission scenarios that accommodate the observed neutrino flux. The black and red lines show the neutrino and gamma-ray spectra after accounting for cosmic evolution and cascading in cosmic radiation backgrounds. The thick solid line shows the case of an emission following E-2.15 with an exponential cutoff around PeV. This scenario actually matches the extragalactic diffuse gamma-ray background (IGRB) measured by the Fermi gamma ray satellite. The alternative scenario illustrates that even for harder emission spectra the cascaded flux will still have a significant contribution to the Fermi IGRB. For illustration, we also show the effect of hadronic emission that produces a peaked neutrino spectrum in the 10 TeV to 1 PeV energy region (thin black line). This emission spectrum is not expected for a pp scenario. The observed gamma-ray spectrum (thin red line) is in this case dominated by secondary cascaded photons. The contribution to the Fermi IGRB between 100 GeV to 1 TeV is still at the level of 10% even though the input neutrino spectrum underestimates the neutrino flux substantially. Soft emission spectra are inconsistent with the Fermi observations and, if confirmed, might indicate obscured or hidden sources.
In general, the exercise shows that the diffuse gamma-ray contribution to the Fermi IGRB is large for pp cosmic beam dumps and suggests a common origin of some of the sources. This is intriguing because a recent analysis shows that blazars dominate the Fermi flux. Are blazars the final answer? The good news is that IceCube can eventually identify blazars by observing multiple neutrinos from the same source. On the other hand, an IceCube search for neutrinos from Fermi’s identified blazars has come up empty. However, the identified blazars only represent 50% of the total diffuse background allowing for the possibility that a partially different source population produces the neutrinos. Alternatively, a subclass of blazars may be responsible for the production of the neutrinos seen by IceCube. At this point, it is rather clear that a multiwavelength path to the neutrino sources looks very promising.
As exhilarating as these scientific results are, they should not overshadow the daunting challenge of constructing the marvelous instrument  that made the observations possible. Given the detector’s required size, early efforts concentrated on instrumenting large volumes of natural water with photomultipliers. After a two-decade-long effort, building the Deep Underwater
Muon and Neutrino Detector (DUMAND) in the sea off the main island of Hawaii unfortunately
failed. However, DUMAND pioneered many of the detector technologies in use today and inspired the deployment of a smaller instrument in Lake Baikal as well as a suite of efforts to commission neutrino telescopes in the Mediterranean. These in turn have paved the way toward the planned construction of KM3NeT, a multi-cubic-kilometer neutrino telescope.
The first telescope on the scale envisaged by the DUMAND collaboration was realized instead by transforming a large volume of deep Antarctic ice into a particle detector, the Antarctic Muon and Neutrino Detector Array (AMANDA). In operation from 2000 to 2009, it represented the proof of concept for the kilometer-scale neutrino observatory, IceCube.
The South Pole effort flirted with disaster on multiple occasions. Most notable among these: the original deployments in ice at too shallow a depth, with remnant bubbles that diffused the light patterns, and the shaky start of the IceCube drilling system. IceCube light sensors are deployed into 2.5-kilometer-deep holes in which a column of ice has been melted. A jet of hot water under high pressure emerges from a nozzle that supplies 200 gallons per minute at 1,000 psi and a temperature of 190 F to melt the ice; see Figure 9. After a heroic effort by everyone involved, the hot water drill eventually completed three holes per week, delivering the project on time.
Figure 9: Some subsystems of the hot water drilling complex showing the drill tower and the 2.4 km hose.
Equally laudable is the still ongoing effort to steadily improve our understanding of the optics of the layers of glacier ice that constitute the actual detector medium. The ice is the detector, and we do not have the luxury to place a sample in an accelerator test beam. More than 20,000 LEDs,
a few lasers, and even a TV camera have been deployed into the ice in order to study the propagation of light.
Last but not least, all IceCube science is performed in a background of atmospheric muons that are triggered at a rate of 3,000 per second, tens of billions per year. That’s a large haystack to sort through. In the end, even those turned out to be interesting, revealing enigmatic anisotropies in their arrival directions; but that is a different story.
 F. Halzen and T. K. Gaisser 2014, Annual Review of Nuclear and Particle Science, 64 (2014)
101 and references therein.
 M. G. Aartsen et al. (IceCube Collaboration) 2013, “Evidence for High-Energy Extraterrestrial Neutrinos at the IceCube Detector,” Science 342 1242856 [arXiv:1311.5238]  M. G. Aartsen et al. (IceCube Collaboration) 2014, “Observation of High-Energy Astrophysical Neutrinos in Three Years of IceCube Data,” Phys.Rev.Lett. 113 101101 [arXiv:1405.5303]
 M. G. Aartsen et al. (IceCube Collaboration) 2014, “Atmospheric and Astrophysical
Neutrinos above 1 TeV Interacting in IceCube,” Phys. Rev. D 91, 022001 [arXiv:1410.1749]  M. G. Aartsen et al. (IceCube Collaboration) 2015, “Evidence for Astrophysical Muon Neutrinos from the Northern Sky with IceCube,” Phys.Rev.Lett. 115 081102 [arXiv:1507.04005]  F. Halzen and S. R. Klein 2010, Rev. Sci. Instrum. 81:081101 [arXiv:1007.1247v2 [astro- ph.HE]] and references therein.