The Telescope in the Ice

by Mark Bowen

  Dear Collaborators,

  We are pleased to announce the first flight of IC86 running as part of the full Data Acquisition System.

  On-Ice lead Greg Sullivan launched the first run.…

  These first IC86 runs are very exciting to us ‘Cubers at Pole, several of whom have been involved with the project since its earliest days or before (AMANDA).

  Greg, a professor at the University of Maryland, had joined at the beginning of IceCube. John, of course, was one of the real old-timers.

  There were still bugs to work out, equipment to install, and commissioning to be performed, so it wasn’t until May that the winterovers switched IceCube into full data-taking mode. The dream of a kilometer-scale neutrino telescope, first held by the DUMAND pioneers some thirty-five years earlier, was now a reality.

  Francis had been waiting for this moment for twenty-three of those years. He broke his usual silence on the collaboration-wide e-mail list and declared, “This is a great occasion. We will finally have a stable detector that we can fine-tune to death and get 0.1 degree angular resolution and 15% energy resolution!

  “Thanks. You know who you are (even I know).”

  * * *

  The interesting thing is that this epochal moment occurred close enough to the one-hundred-year anniversaries of both the balloon flight on which Victor Hess discovered cosmic rays and Roald Amundsen’s first footprints at the South Pole.

  22. Crossing the Threshold

  At its outer boundaries, experimental particle physics is largely a matter of groping in the dark. Since its practitioners are searching for a true signal of an unknown nature in the presence of a noisy background—or, as Fred Reines put it, “listening for a gnat’s whisper in a hurricane”—there is always the danger of convincing themselves that they’re seeing something that isn’t really there. To deal with this problem (and recognizing what Nobel Laureate Eugene Wigner once called “the unreasonable effectiveness of mathematics” in describing the natural world), they have come up with a mathematical definition of discovery, based on probability and statistics. Assuming that you’ve designed your experiment correctly and your analysis makes sense, a true discovery would be an event that is highly unlikely to have occurred as a result of a random fluctuation in the background. And the bar is set rather high. The agreed-upon threshold for a claim of discovery is that the odds that it was produced by random chance are less than one in about three million.

  Because of the way they do the math, however, physicists talk about it in another way. They speak in terms of the characteristic range of the noise or fluctuations in an experiment, which statisticians call their standard deviation. If you were trying to measure the height of a boat on a windblown sea, for example, the standard deviation might be proportional to the average height of the waves. If the boat was a small dingy in a heavy storm, it would be hard to measure its height—it might even be hard to see the thing. If it was a tall ship on a calm day, it would be much easier. The Greek letter sigma, σ, is the mathematical symbol for standard deviation. In particle physics parlance, saying that your result stands “five-sigma” proud of the noise in your experiment (or, loosely speaking, that the ship is five times taller than the average wave) is equivalent to saying that the odds that it was a chance fluctuation are less than one in three million—or that you have made a discovery.

  In April 2011, the University of Wisconsin held a splendid gala for the commissioning of IceCube. Eminent cosmic ray physicists and cosmologists, pioneers in neutrino astronomy, representatives of the funding agencies, and other friends were flown in from around the world. They gathered in a Frank Lloyd Wright–inspired conference center at the edge of Madison’s Lake Monona and were treated to a day of commemorative events, culminating in an elegant banquet in the city’s classy new arts center. Two symposia were held in conjunction with the gala, one on particle astrophysics and the other on science in Antarctica. John Learned was there. Grigorii Domogatsky was invited but had to graciously decline: he couldn’t get a visa in time, because he had just returned from his annual spring field trip to Lake Baikal. Congressman David Obey, he of the kosher pork, gave a speech; and one of the more memorable lines of the day was delivered by Anders Karlqvist, former director of the Swedish Polar Research Secretariat, which has been funding IceCube and AMANDA since the early nineties. “Now it is time for the scientists to deliver,” he said. “Let me remind you that my organization is located in the same building as the Royal Swedish Academy of Sciences. I look forward to welcoming three of you (for it cannot be more than three) as you walk in the door. Welcome to Stockholm!”

  For the scientists in the audience, it is probably safe to say that the highlight of the day was when Francis Halzen went out on a limb and predicted that IceCube would have a five-sigma result within five years. He has nothing if not nerve.

  * * *

  While most of the sound and fury had swirled around the building of the instrument for the previous decade, the physicists had been making significant progress on the scientific front as well. Plenty of graduate students and post-docs had helped put AMANDA’s data to good use, and they had all hopped on IceCube in 2005, when it began producing data. Thanks to the digital optical module, they could detect up-going muons with just their first hard-won string, and they built upon that small beginning as the instrument grew.

  To give an idea of how much more sensitive IceCube is than AMANDA at detecting astrophysical or extraterrestrial neutrinos, the ones they’re really looking for, five years of data from the larger instrument is equivalent to about a million years from the smaller one. And IceCube detects about 100,000 atmospheric neutrinos a year, whereas AMANDA-B10, the ten-stringed instrument they used to prove feasibility, saw only about 700. As IceCube grew, therefore, the collaboration could study the atmospheric neutrino beam with greater and greater precision. With the early, smaller incarnations of the instrument, they also searched (a loaded word) for point sources and for the cascade and double-bang signatures of electron and tau neutrinos.

  Their instrument is also a cosmic ray detector. Between IceCube and IceTop, they observe four or five billion cosmic ray muons raining down from the southern sky every year—and it so happened that this brought their first small surprise. It turns out that cosmic rays don’t stream in uniformly across the sky; there’s a waviness in their arrival directions.

  The Swedes systematically expanded and enhanced their search for dark matter WIMPs or neutralinos. In 2006, employing AMANDA, the collaboration published a search for neutralino annihilations in the Sun, which happens to be the best place to look, for various reasons. In 2009, with the 22-string edition of IceCube, they produced a second study of the Sun that set stringent enough limits on the existence of WIMPs to begin threatening the viability of a certain class of supersymmetry theories. Not only did this add to the creeping unease about supersymmetry in general, it narrowed the possibility that neutralinos might comprise any significant fraction of the cold dark matter. Even before its completion, IceCube was making waves.

  As always, the holy grail was a point source or neutrino star, and since the mid-1990s, the most promising candidate for that has been the gamma ray burst. These enormously bright and distant explosions occur about once a day somewhere in the visible universe, and the explosion in infrastructure that has taken place in particle astrophysics over the past several decades has produced two marvelous telescopes for looking at them. Both are satellite-based and operated by NASA. The Fermi Large Area Telescope produces a map of the entire gamma ray sky every three hours, and Swift is specifically designed to detect gamma ray bursts. It has a special “burst alert telescope,” which detects and determines the position of about one hundred bursts a year. After a detection it “swiftly” reorients itself to observe the afterglow in the X-ray, ultraviolet, and visible bands. (Unfortunately, the satellite isn’t that swift, as it takes almost two minutes to spin into position. There’s an entire class of GRB that lasts less than two sec
onds, while the longest last several hours.) Both of these telescopes produce catalogs of the whens and wheres of the bursts they’ve detected, which the IceCube scientists can use to search for neutrinos. The great advantage here is that since gamma ray bursts are localized both in space and in time, there is essentially no background. Even one high-energy neutrino coming from the direction and within some short time window of a GRB would be statistically significant all by itself, and two might very well constitute a five-sigma discovery.

  The big question about these violent creatures is what they actually are. The short bursts are thought to result from the merger of a binary system, a neutron star or black hole absorbing some companion revolving around it, and the longer ones from an unusually powerful supernova or “hypernova,” the collapse of an extremely massive star. The end result of either of these events would be a rapidly spinning black hole, and the bursts of gamma rays are probably emitted during the brief time of the collapse. What separates GRBs from the usual supernova is that they emit their light in laser-like beams or “jets,” shooting out in opposite directions along the axis of rotation of the collapsing system—from opposite poles, you might say. These flashes of light can be extremely distant. Some come from so-called cosmological distances, which means that they occurred billions of years ago, when the universe was comparatively young. The very fact that we can see such distant events means that they must be incredibly bright, probably even brighter than supernovae.

  Back in the mid-nineties, John Bahcall and his brilliant young Israeli protégé, Eli Waxman, proposed a “fireball model” for GRBs that gave estimates for the number of neutrinos that would be emitted along with the gammas. The notion was that protons produced during the gravitational collapse would be accelerated to very high energies by the enormous magnetic fields of the spinning system and would then interact with the gamma ray field, producing pions that would decay quickly into other particles, including neutrinos. Hence, the neutrinos would emerge in the jets along with the gammas—and at extremely high energies, even higher than that of the famous Oh-My-God particle, 1020 electron-volts. The two theorists even proposed at that early date that some future kilometer-scale neutrino telescope should be able to observe GRB neutrinos. Other fireball models, involving neutrons rather than protons, were also proposed.

  The models suggested that particles besides neutrinos might also be emitted at ultrahigh energies: protons, neutrons, and helium and other nuclei. And this held out the possibility of answering the oldest and most intriguing question in cosmic ray physics: where do ultrahigh-energy cosmic rays come from? Eli Waxman had proposed GRBs as the answer back in 1995, two years before he and his mentor filled in the details of their fireball model. But as the one-hundred-year anniversary of Victor Hess’s seminal balloon flight approached, this overarching question remained unanswered.

  John Bahcall was not only one of the foremost theoretical astrophysicists of his time, he was also admired for his big ideas, his integrity, his sense of fun, and his personal warmth. Neutrino astrophysics wasn’t the only field that lost a visionary when he died of a rare blood disorder in 2005, at the young age of seventy. Eli took up the mantle as touchstone theorist in neutrino astrophysics and as IceCube neared completion, began boldly predicting that his faith in GRBs would soon be confirmed.

  Francis had been on the trail for more than a decade at that point. Rellen Hardtke, the student whom he suspected of pulling his leg when she showed him the Peacock events that August morning in 1998, wrote her doctoral thesis on a GRB search with AMANDA. They had hoped to see GRB neutrinos even then, of course, though they knew it was a long shot because AMANDA wasn’t sensitive enough to detect the fluxes predicted by the models. The first incarnation of IceCube that was, forty-string IC40, was completed in early 2008, by which time the IceCube group at the University of Maryland had joined in. A grad student there by the name of Kevin Meagher, working under professor Kara Hoffman, conducted one study, and Francis’s student Nathan Whitehorn conducted another. In January 2011, about three weeks after IceCube’s last string was deployed, they uploaded a report on these studies to the arXiv preprint server at Cornell University, the preferred way for high-energy physicists to communicate their latest results to their colleagues. Their search was still a search: one year’s data from IC40 had yielded no evidence of neutrino emission by gamma ray bursts.

  At the same time, Francis and two fellow theorists, Maria Concepcion Gonzalez-Garcia and her post-doc Markus Ahlers, had been coming at the problem from a different angle. (Gonzalez-Garcia held appointments at the University of Barcelona and the State University of New York at Stony Brook. Ahlers worked with her at the latter.) In March, the three published a second paper on the arXiv, under the title “GRBs on Probation,” demonstrating that the simple fact that IC40 had detected no extraterrestrial neutrinos indicated that gamma ray bursts probably weren’t the source of ultrahigh-energy cosmic rays. If they were, the three surmised, there would be a diffuse flux of equally high-energy neutrinos, raining in from every direction, as the cosmic rays do, and the neutrino flux could be calculated from the cosmic ray flux using the fireball model. Carrying out the calculation, they demonstrated that IC40 was sensitive enough to detect the predicted neutrino flux. Since it hadn’t, they deduced, GRBs probably don’t produce ultrahigh-energy cosmic rays either.

  This is an excellent example of the way a null result, or new limit, can propel science forward. Everyone would have preferred to see GRB neutrinos, and it would have been spectacular if they had, but nature is indifferent to the preferences of humans. As the author of a perspective in Nature wrote, “Carl Sagan’s antimetabole ‘absence of evidence is not evidence of absence’ immediately comes to mind: not detecting an expected signal can be as powerful a diagnostic as the arguably more desirable discovery, and may send us back to re-evaluate how we arrived at our expectations in the first place.”

  GRBs were only on probation, however; they were not yet expelled. There was enough wiggle room in both the models and the results, factors of two here and there, to keep the fireball alive.

  Eli Waxman was invited to speak at the particle astrophysics symposium that accompanied IceCube’s inaugural gala, and since all this news was hot off the press at the time, his talk was much anticipated. Eli’s a bit of a prizefighter up there. His simplest statements come across as challenges somehow. While he was willing to admit that IceCube was “getting dangerous,” he stuck to his guns. The instrument wasn’t sensitive enough yet, he told us. The Cubists had missed a factor of two in their analysis. The fireball would survive.

  Such cocksureness on the part of theorists gets under the skin of the working experimentalist. There was noticeable grumbling and rolling of the eyes as Eli spoke. Some said he wasn’t particularly concerned with experimental evidence. John Learned, who has his own rather well-informed ideas about how GRBs might work, became especially argumentative.

  The other shoe dropped about a year later, when the collaboration added another year’s data to their GRB search, from the next iteration of their instrument, IC59. Nathan Whitehorn again took the lead in Madison, and grad student Peter Redl, advised by professor Greg Sullivan, took the lead in Maryland. Again they failed to find neutrinos, now associated with two years of GRBs from the catalogs. Their revised upper limit on the flux was at least a factor of four below the fireball predictions and possibly a factor of ten or even thirty. (Nathan explains it more simply: they would have expected to see about twenty neutrinos and they saw none.) “This implies either that GRBs are not the only sources of cosmic rays with energies exceeding 1018 electron-volts or that the efficiency of neutrino production is much lower than has been predicted,” the collaboration wrote in Nature.

  In the thick of it all, in November 2011, as he and Peter Redl were putting the finishing touches on the Nature manuscript, Nathan admitted that this was a bit disappointing. “It would have been really nice to see something; I’m not going to say that’s not true. A
nd IceCube will eventually see some astrophysical neutrinos, and that will be very exciting.… But one of the nice things about this particular thing is that you had actually a null result that was really meaningful in this case, right? You have a very distinct prediction and you can say something about it, you know, just as powerfully as if you had seen something.… And IceCube is pushing into this region in general on a lot of fronts, not just gamma ray bursts, where if we continue to see nothing then that has some really interesting meaning.”

  Four years later, with that much more sensitivity and a larger and more capable detector to boot, IceCube still has not seen GRB neutrinos. The biggest mystery in cosmic ray physics endures, and the theorists have gone back to the drawing board. Eli’s latest guess is a type of galaxy called a starburst, which has an unusually high rate of star formation.

  * * *

  Life became easier when construction ended. As Francis had said in his e-mail, they now had a stable instrument that they could fine-tune to death. But they did much more than simply polish the lens of their new telescope. When they were released from the enormous burden of adding to it every year, they channeled at least that much energy—and probably more—into scientific creativity. It was like a bunch of kids being let out onto the playground. There was a buzz in the air at the fall collaboration meeting in Uppsala, less than six months after the inaugural. Francis told me, “IceCube is so incredibly multidimensional in ways that we discover every week.” They were pushing out the boundaries in both astrophysics and pure particle physics.

  They had just detected their first cascades, which meant that after a decade of relying solely on muon neutrinos they were now opening up the possibility of detecting the electron and tau flavors. Per Olof’s vision, DeepCore, was looking as though it would be capable of studying the details of neutrino oscillation as well as the accelerators could, and be the only game in town at high energies. (They would open many eyes—and roil the accelerator community—at a neutrino conference in Kyoto the following spring by presenting five-sigma evidence for the highest-energy oscillations ever detected.) And in connection with oscillation, the prospects looked good for solving the longstanding mystery of the sterile neutrino, a potential fourth flavor of the particle, whose existence had been hinted at by several experiments over the previous two decades. Francis had been focusing on that prospect for about a year, and was now realizing that “if you have to design an experiment to see sterile neutrinos, you design IceCube.”