The Telescope in the Ice

by Mark Bowen

  At that time, no one in the world of glaciology understood why there was such a long transition zone between the depth at which the bubbles in a glacier began to disappear, in other words, when the air inside them began to migrate into the ice to form clathrate crystals, and the depth at which they disappeared altogether. In some ice cores this zone is as long as 800 meters. The theories of the time held that the bubbles should disappear instantly at a depth that reached the right combination of temperature and pressure.

  After returning from Venice, Francis gave Buford his first lead by telling him about the paper Domogatsky had pointed out to him. (It was fortuitous that Buford read Russian.) The paper summarized a painstaking study by a man name Vladimir Lipenkov, who had spent months in a cold room with an ice core recovered at Vostok in 1980. Inspecting the core manually with a microscope and reticule, Lipenkov had catalogued the size and concentration of the air bubbles as a function of depth, down to 1,400 meters.

  Looking further in the literature, Buford found laboratory studies of clathrate formation, which showed that the reason the transition doesn’t occur instantly is that it takes awhile for the air molecules to diffuse into the ice. Thus the ice has to reach a certain age in addition to the right temperature and pressure. Putting together empirical data from ice cores from Greenland and Antarctica, he used the venerable diffusion equation, first developed in the mid-1800s, to produce a theory of the rate of clathrate formation that could be applied to all ice cores. It accounted convincingly for the very different transitions in the Byrd and Vostok cores. His paper presenting the theory, which again appeared in Science, is now recognized as one of the classics in the journal’s archive.

  The main factors controlling the disappearance of the bubbles are the age and temperature of the ice, as a function of depth. Bruce Koci had developed a model for the temperature profile at Pole, but since no deep core had ever been obtained there, the age profile was unknown. So Buford went ahead and developed one of his own.

  In the course of all this work, he realized that every core that had ever been obtained in the polar regions showed evidence of a series of dust layers, which had been laid down all over the world during cold intervals and ice ages in the distant past, when the atmosphere had been drier and dustier than it is today. These layers would both scatter and absorb light and thus hinder muon tracking significantly.

  In both his papers, Buford made specific suggestions about how deep the collaboration needed to drill the next time around. His clathrate theory predicted that the bubbles should cease to be an issue at about 1,400 meters, and his theoretical age profile predicted that the two most recent, that is, highest, dust layers would be found in the next thousand meters or so. These, he suggested, were depths to avoid.

  This impressive tour de force had obvious practical value, and it was accomplished in time to be useful as well. He submitted his papers several months in advance of the next possible deployment season, 1994–95.

  * * *

  The third contribution came from Serap Tilav, who was the only pure cosmic ray physicist in the group at the time. You could say that Serap was “greedy on” the data. She was looking for muons from the minute she landed on the Ice in the same year that the instrument landed in the bubbles.

  She had “trained her eye” beforehand so that she would know what to look for when the data began coming in. She grilled Steve Barwick on the electronics in order to understand exactly which so-called observables the instrument would produce. She developed a Monte Carlo for the instrument and sent down-going muons through it, using a Monte Carlo for air showers that she had brought to Madison from Tom Gaisser’s lab at Bartol. At first, since she had nothing to go on, she “dipped” her simulated instrument in perfect ice, with no absorption and no scattering—and, of course, no bubbles.

  Remember that in 1987, Gaisser, Alan Watson, and friends had installed the South Pole Air Shower Experiment, or SPASE, at Pole. This instrument was designed to detect down-going showers of secondary particles produced by cosmic rays hitting the atmosphere above the instrument. (As we’ve seen with the Oh-My-God particle, a single cosmic ray can produce billions of offspring, and many of them turn out to be muons.) SPASE was located on the surface of the Ice, not far from AMANDA, and it could also tell the direction of the showers, so it could be used to “tag” the ones that were headed in AMANDA’s direction.

  When the data began rolling in, Serap realized just as quickly as her friends looking at the flasher data that the ice was not perfect. She would see an overall rise in the amount of light hitting AMANDA when an air shower tagged by SPASE passed through, but the pattern of the light as it spread across the array in no way resembled the perfect air showers she had trained her eye to see. As with the laser pulses, the signals from the air showers lasted much too long a time. The Cherenkov light from the muon showers was bouncing around in the bubbles just as the laser light was.

  “So nobody slept,” she says. “We tried to think.”

  She couldn’t solve her problem as the others had with pencil and paper. It required a lot of “Monte Carlo-ing,” as she says, to begin to get a handle on it. When she returned to the real world, she kept track of the Swedes’ progress on the optics, and she incorporated their work on bubbles and ice properties into her Monte Carlo, but it still didn’t reproduce the patterns in the data.

  As it worked out, the collaboration did not drill during the 1994–95 season. At that point, in fact, they weren’t sure they’d be given the chance to drill ever again. They did some maintenance on the instrument, and Barwick installed a new flasher system that could send different colors of light through the ice. This proved key for Serap, since it showed that the shorter the wavelength of the light, that is, the more blue or violet it was, the more transparent the ice. At the shortest wavelength they tried, which was violet but still visible, the absorption length was an astounding 230 meters. This was fortuitous, because the Cherenkov light given off by muons is weighted heavily toward the ultraviolet, just beyond the range that the human eye can see.

  Serap incorporated every new piece of knowledge into her Monte Carlo. She adjusted and tweaked every parameter she could think of, trying to reproduce the patterns in the data. At one point she even tried increasing the speed of light, which is a no-no, thanks to Einstein’s theory of relativity. This produced some good matches with the data, but Francis reined her in. Finally, she arrived at a more sensible conclusion: she could match the patterns in the experimental data by adjusting absorption lengths in the ultraviolet range of the spectrum—beyond the range they had probed with lasers—to more than three hundred meters! This was more reasonable than violating relativity, but it was still outrageous enough to evoke strong resistance within the collaboration. It took her colleagues almost a year to accept the idea. Antarctic ice turns out to be clearer than diamond. It is the clearest natural substance known.

  Serap’s discovery was serendipitous in two respects. Not only did it demonstrate that the ice is remarkably clear in just the part of the spectrum where most Cherenkov light is produced, it turns out that photomultipliers reach a peak in sensitivity in the ultraviolet, too. So both the hardware and the ice happen to work best at just the right wavelengths for neutrino astronomy.

  So, Buford made contributions to glaciology and climatology; Serap and the Swedes uncovered new knowledge about the fundamental properties of one of the most heavily studied substances there is, ice; and the collaboration as a whole had made steps toward building a better instrument. But this wasn’t astrophysics exactly; it was a bit down in the weeds. Francis provided the pièce de résistance by demonstrating that their small bubble-filled instrument was already capable of cutting-edge particle astrophysics.

  * * *

  Markov-type, plum pudding instruments like AMANDA were not supposed to be sensitive to the low-energy neutrinos given off by supernovae. During the good old days of collaboration with the Russians, the early pioneers had come up with UNDINE, a sh
ell-type, Greisen design, for that side of the problem. The two Cherenkov instruments that had detected Supernova 1987a, IMB and Kamiokande, were both Greisen designs.

  But back in 1993, before AMANDA got “stuck in the bubbles,” Francis, his student John Jacobsen, and his former post-doc Enrique Zas had shown that Markov-type instruments could detect supernovae.

  Jacobsen is unique even among the varied individuals who have collaborated on AMANDA and IceCube over the years. At that point, he was enrolled in both the art and physics departments at Madison. He’d been what he calls an “art nerd” in high school and then majored in physics at Madison, where he was introduced to research by none other than Ugo Camerini.

  One day Camerini walked in to John’s office at the campus computer center, where John had a student job, and asked him to drop by his office sometime; he had something he wanted to talk to him about. These were the early days of personal computers, and IBM was handing out its landmark PC to professors free of charge in order to promote its use in educational environments. These were new-fangled gadgets to Camerini, who was decidedly old school.

  “So I showed up in his office,” John recalls. “And he says, ‘I have this data, and I want to shove it up the ass of this computer.’ Hah, hah! He’s like—you could smoke in your office back then—like chain smoking, and [this] grimy keyboard and gravelly voice.… And so I wound up working for him.”

  John’s first project was to help analyze data from an early gamma ray satellite named COS-B. He then got involved in the Haleakalā telescope in Hawaii and made the acquaintance of Bob Morse.

  After graduation, he worked at CERN for a year, programming again, and then returned to Madison to go to art school. Needing a way to pay the rent, he approached Morse, and Bob offered him a job doing Monte Carlos for AMANDA. This was August 1991, just before the first deployment season when they stuck Bucky-1.

  After about two years, Francis got on John’s case and said, “You know, you’re doing a Ph.D. project, why don’t you just get a Ph.D.? This is stupid. All you have to do is take a couple classes.”

  John considers this a good example of what he calls Francis’s “criminal optimism”: it ended up taking him about four years to earn his doctorate.

  But they were rewarding years. Francis showed him that there are different ways of being a physicist. “The standard idea is that … you’re going to live in this very formally-specified field where there are definitely wrong ways to do things and it’s all about dotting your i’s and crossing your t’s,” John observes. “Any sloppy thinking and basically you’re out; you’re going to fail, and nobody’s going take you seriously.

  “But there are different approaches, right? Some people take a particular corner of the room and they’ll start just digging;… they’re going to escape from jail by digging one spoonful at a time. And there are other people who are just going to think about things and write a few expressions down on a napkin or have a drink with a friend and, like, be creative! You might not be the person to fully realize whether this idea is … true or not, but someone’s got to come up with those ideas. Someone has to be sort of the artist of physics and do that. And I realized from Francis that you could do that, that he does that, that that’s actually why he does it, that this very creative kind of thinking was a deep source of pleasure and he could actually live that way.”

  The other collaborator on the supernova work, Enrique Zas, was brilliant mathematically. After a post-doc with Francis in Madison, he had returned home to Spain to become a professor. The way the three worked together, according to John, was that Francis “would go off—he would have an idea. He and Enrique would sit down, and they would say, ‘Oh, what if we did this?’ And then they would toss it over to me, and we’d do some Monte Carlo together, and then we’d write a paper. And then, you know, hundreds of people would cite it for the next decade, or three decades, or whatever.” Such was the case with their supernova work.

  * * *

  Although supernovae produce all six varieties of neutrino and antineutrino in roughly equal measure, a Cherenkov detector sees only the electron antineutrinos, because at supernova energies the other types don’t interact appreciably with water or ice. The typical energy of an electron antineutrino emitted by a supernova is about twenty million electron-volts. It will interact with a free proton to produce a positron, which produces a Cherenkov track that is only about twenty centimeters, or eight inches, long. This distance is so much shorter than the distance between the modules in AMANDA that the instrument can’t possibly reconstruct such a track.

  What the three physicists realized, however, was that you don’t need to track the positrons in order to detect a supernova. As the ten-second neutrino flash from a supernova passes through the instrument, a large enough number of positrons should be born within ten or twenty centimeters of a large enough number of phototubes to produce an uptick in the total amount of light sensed by the array. It’s hard to tell which direction the flash came from, where in the sky the event occurred, but you should be able to tell that it did occur if you monitor the collective behavior of the entire array: the sum of all the signals coming from every single phototube. And since the electronics of the instrument respond so rapidly, on the order of a few billionths of a second, it should be possible to reproduce the shape of the neutrino flash in exquisite detail—in great contrast to IMB and Kamiokande, which detected only a handful of neutrinos altogether.

  This could provide a rigorous test of the competing theories about how a supernova explodes, each of which predicts a slightly different shape. Different features in the shape of the neutrino flash should provide evidence of the swift collapse of the iron core, the ultraluminous burst accompanying the explosive bounce-shock, the streaming out of neutrinos from the proto-neutron star that presumably revives the shock, and so on. Two interesting new possibilities also arise: The sudden cutting off of the flash would indicate that a black hole rather than a neutron star had formed at the center of the star, because a black hole would pull even the neutrinos back into itself. And more speculatively, a brief spike at a specific point during the flash might signal the formation of an exotic quark star.

  Greisen-style instruments like IMB and Kamiokande have other advantages, such as the ability to tell the rough direction of the incoming neutrinos, but it turns out that you can make the most sensitive basic detector much less expensively with the Markov design, since it reaches equivalent sensitivity with far fewer phototubes.

  In the innocent days before they met the bubbles, Francis and his students estimated that a completed AMANDA-A or DUMAND II, comprising only two hundred or so optical modules, could detect a supernova occurring in our own galaxy with some confidence, but not quite enough. Random noise from the detectors would result in a fake signal more than once a century, but you have to do better than that, because galactic supernovae are expected to occur at about the same rate. Therefore, a small Markov instrument could not “watch” for supernovae on its own; it would produce too many false positives. It would need to be backed up by an optical telescope or an instrument like Kamiokande.

  When AMANDA dropped its first four strings into the bubbles, Francis began searching desperately for “some way to make this disaster look good.” “I was thinking, ‘What can you do when you have a scattering length of less than half a meter and an absorption length of two hundred seventy meters?’ And the answer was supernova.”

  Or, as Bob Morse said at the time, “If you’ve got lemons, make lemonade.”

  In 1995, Francis and his students revisited their earlier calculations, employing the astonishingly long absorption lengths they had now observed in South Pole ice, and discovered that the tiny, four-string instrument they had already built was by far the most sensitive supernova detector on the planet. Owing to the long absorption lengths at ultraviolet wavelengths, every single one of the seventy-three working optical modules in AMANDA-A (seven had failed for one reason or another) was monitoring rough
ly the same volume as an entire Kamiokande or IMB instrument. And the sensitivity was actually enhanced by the bubbles, since they tended to keep the light inside the array! Scaling up from the eleven neutrinos that Kamiokande had detected from Supernova 1987a, the AMANDA trio estimated that their small instrument would have detected about twenty thousand neutrinos from the same event. AMANDA-A was already capable of acting as a stand-alone supernova watch for the entire Milky Way galaxy and would continue to work for a decade or more “with minimal maintenance or investment in manpower and operation.” Thus, even their handicapped instrument represented a breakthrough.

  “Suddenly our disaster had a mission,” Francis recalls.

  * * *

  Christian Spiering was following these developments closely from Zeuthen, Germany. Baikal was still going strong—the collaboration had reached another milestone by reconstructing the first-ever up-going muons in the summer of 1994—but the long-term prospects did not look good. There were serious obstacles to working in Russia, and the shallowness of the lake made for additional hurdles. He was casting about for other opportunities, and he was a strong player. Not only was his group now associated with DESY, one of the preeminent particle physics enterprises in Europe, they also had an impressive track record with Baikal. All three of Baikal’s competitors, DUMAND, AMANDA, and NESTOR, were courting him. In July 1994, he and his Zeuthen colleagues came to a decision. In the full knowledge that AMANDA was mired in bubbles, they decided to join anyway—and retain their membership in Baikal. Just as they had given a shot in the arm to the Russians, they gave another to the Swedes and Americans, who were preparing a second grant proposal to NSF: a stronger collaboration made for a stronger proposal.