Ice Fishing for Neutrinos

The lensmakers in Flanders developed the optical telescope to allow traders to make an early inventory of the goods on ships crossing the English Channel. Little did they anticipate that Galileo would use the same instrument to discover the moons of Jupiter. The first rocket-borne X-ray detector was proposed to study the sun and moon; its successors revealed wonderful skies with neutron stars, quasars and black holes. The gamma ray instruments developed by the US to search for thermonuclear explosions in the Soviet Union discovered the still enigmatic gamma ray bursts. Stories like these are associated with each wavelength of light. Perhaps the most dramatic is the accidental discovery of the cosmic photon background with apparatus meant to study sky-interference with telephone communications. The ingenuity of astronomers has produced instruments which have collected information spanning 60 octaves in photon frequency, from 10^4 radio-waves to 10^-14 cm GeV-energy photons. This is an amazing expansion of the power of our eyes which scan the sky over less than a single octave just above 10^-5 cm wavelength. This article describes an attempt to do astronomy at wavelengths smaller than 10^-16 cm. As if it were not a sufficient challenge to detect the very small fluxes of very high energy photons, the Antarctic Muon and Neutrino Detector Array (AMANDA) will probe the sky in the corresponding energy region by detecting muons and neutrinos instead. Hopefully, in a not-too-distant future, short-wavelength neutrinos will contribute their own bizarre tale. Although nothing can be guaranteed, history is on our side.

Neutrinos are particles associated with radioactive phenomena. They are also produced in the decays of elementary particles such as pions. They have no electric charge and, as far we know, no mass. Their existence was postulated by Pauli as a book-keeping device to conserve energy and momentum in the radioactive decays of nuclei. He liked to say: "I have done a terrible thing, I have postulated a particle that cannot be detected." Clyde Cowan and Fred Reines proved him wrong 23 years later. The problem is that interactions of neutrinos with matter are extremely feeble. The handicap of a minuscule probability for the absorption of neutrinos in the instrumented volume of a particle detector can, of course, be overcome by exposing it to an abundant source. Cowan and Reines contributed to its construction: the atom bomb! The detector they designed was, in the end, operated in the vicinity of a more friendly source, the Savannah River nuclear reactor. In the sky there are only two intense sources of neutrinos: the sun and nearby supernovae. They have been studied (for supernovae, only for a total time of about 10 seconds in February 1987) by dedicated detectors. This has been a spectacular start. Sources of neutrinos far beyond the sun are expected to be much dimmer. It is generally believed that one must construct kilometer-scale detectors to scrutinize the skies for neutrino sources beyond the sun. It is a daunting task but worth it. Theorists have indeed already anticipated an impressive wish-list of missions for such an instrument ranging from the observation of neutrinos from bright, far away galaxies to those from the annihilation of cold dark matter particles inside our own galaxy's halo. The hope is that a particle that is almost nothing may tell us everything about the Universe.

The neutrino sky may reveal the enigmatic sources of the highest energy cosmic rays. We have known for over a decade that Nature accelerates particles to energies in excess of 10^20 electron volts. The sources of these particles are shrouded in mystery because their very existence represents a challenge to both astronomy and physics. We do not know of magnetic fields sufficiently powerful and expansive to accelerate particles to such energy. Once accelerated, such particles should readily lose energy in interactions with the cosmic microwave background, and therefore arrive at our detectors with reduced energy, thus compounding the acceleration puzzle.

The accelerator physicist's method for building a neutrino detector will typically use lead absorber to filter out all particles but the neutrinos, wire chambers and associated electronics with a combined price tag of roughly 10,000 US dollars per square meter. A 1 kilometer-square detector would cost 10 billion dollars. Realistically, we are compelled to develop methods which are more cost-effective by a factor of one hundred. The only known solution is to use a "natural" detector consisting of a thousand billion liters (a teraliter) of instrumented natural water or ice. Such instruments should be able to study our Universe far beyond the sun and watch cosmic cataclysms without having to wait for a one-in-a-century miracle like a nearby supernova.

My involvement with neutrino astronomy took a circuitous road. Luck had it that I gave a talk at the University of Kansas where glaciologists happen to be part of the department of physics. I remember my excitement when Ed Zeller first told me of the existence of a small array of radio-antennas searching for signals of cosmic neutrinos interacting in the Antarctic ice at the Russian Vostok station. By then the idea was 30 years old. High energy neutrinos interacting with ice produce a burst of light. Some of these photons will knock electrons out of atoms and produce a spark, a few meters in length, which may occasionally contain a million electrons or more. My interest peaked with his comment that the Russian physicists had been unable to compute the power in the signal. With an enthusiasm reminiscent of graduate electromagnetism, Enrique Zas, Todor Stanev and myself solved the "hardest Jackson problem ever", using superior computer power not available to our Russian colleagues. The answer was a bit disappointing. A practical detector could only detect the very highest energy neutrinos, eliminating the possibility to do the science that speaks to the imagination of a particle physicist, e.g. the search for cold dark matter possibly composed of the stable particle predicted by supersymmetry. In physics jargon: the detector may be large but the threshold was disappointingly high.

Why not detect the burst of light directly? Sure there is an electric spark, but neutrino interactions also radiate a huge glow of blue light. It has the same origin as the ghostly glow of blue light emanating from the water covering a nuclear reactor. Who else to ask other than John Learned who was at the time preparing to instrument large volumes of the clear, deep ocean water off the coast of Hawaii to trap neutrinos. Frantic electronic mail followed for several weeks and the idea become increasingly compelling when more aspects of the problem were considered. The AMANDA idea is, in principle, very simple: sink an array of photomultiplier tubes (PMTs) deep into the polar ice cap where the pressure of the overburden squeezes all the bubbles out. Light from neutrino interactions will be transported over large distances in this clear ice to illuminate the array of PMTs. As a theorist all I knew about PMTs was that they looked like a lightbulb and operate like a lightbulb in reverse: light goes in and electricity comes out. But what a lightbulb! The operating AMANDA PMTs amplify the faint signals by over a hundred million times.

It did not take long to realize the advantages of this technique. Polar ice is a totally sterile medium, free of radioactive material such as potassium which is ubiquitous in sea-water. Its decay products are responsible for ten thousand false signals every second in the sensitive PMTs looking for the faint neutrino-light. In the absence of radioactivity off-the-shelf commercial electronics can be used to identify the footprints of neutrino interactions in the array, a great simplification and cost-saver. Also, you can walk on your experiment. Only the PMT is deployed. It is connected by a simple kilometer-long coax cable to the data acquisition which remains at the surface, accessible for repair and updating. In deep ocean experiments the deployment of active electronics in an unfriendly environment is inevitable because the raw signals would be totally degraded after transmission over twenty kilometers of cable from experiment to shore. Conveniently, glaciologists had already invented the technology to cheaply deploy equipment in ice: hot water drilling. In summary, the experiment can be made technologically simple and cheap, critical factors when one has the ultimate ambition to instrument kilometers of relatively inaccessible Antarctic ice. It turned out that there were two more advantages which we did not appreciate at the time. Ice is transparent as diamond in the blue wavelength region where the PMTs operate; more about that later. The National Science Foundation operates a research station at the South Pole with an infrastructure reminiscent of a national laboratory. With ASA, the Antarctic Support Associates, it has created an effective and surprisingly friendly environment to operate an experiment like AMANDA whose natural milieu is the large high energy physics laboratories of the US or Europe. A fifty foot crane, heavy-duty vehicle or snowmobile is only a phone call away.

The list of advantages kept growing to the point that the only mystery, in retrospect, is why we did not proceed to do the experiment. We had excuses, John had his hands full with his own project in Hawaii and I am, after all, a theorist. Like a true theorist, I was happy to write a paper... All this changed about a year later when I received a call from an irate NSF program officer telling me that "two guys from Berkeley had been caught trying to sneak a string of photomultipliers into Antarctica to detect muons in the ice," and asking whether I might have put any crazy ideas in their heads. I had in fact never heard of them. Doug Lowder and Andrew Westphal had decided that they would volunteer their labor to a Caltech group drilling holes at the deep ice glacier "Upstream B" in return for the opportunity to lower a muon detector, rigged up with Steve Barwick, into one of them. I actually talked the NSF officer into letting them get away with it and I am happy to report that for the last four years I have had the pleasure to collaborate with Buford Price and his "young Turks", as they were introduced to me, on a officially funded experiment. We were also joined by my colleague Bob Morse. Attracted by the long Antarctic nights and clear, dry, cold skies, he had been operating a telescope at the South Pole. He had also been toying with the idea to use shallow ice as a cosmic ray muon detector. With a Berkeley graduate student, Tim Miller, he did the first transparency measurements of ice in Greenland, "borrowing" the 3 kilometer hole drilled by a collaboration of glaciologists for the GISP-project in 1990.

In the 1992 Antarctic summer the embryonic AMANDA-collaboration deployed an odd collection of PMTs at the South Pole, positioning them at various depths to try out the technology. The equipment was deployed using a hot water drilling technique pioneered by glaciologists. The drill has been compared to a rather heavy bathroom shower head. Gushing out hot water, it melts its way down the ice steered only by gravity. In its free fall it will deviate by less than 1 meter from vertical when reaching the 1 kilometer mark. The melted ice is not removed from the hole; the hot water is continuously recirculated in order to keep the hole from refreezing. This task is not too difficult --- the surrounding solid ice is a great insulator. After the stream of hot water is interrupted, the water will not refreeze for several days, leaving plenty of time for deployment of the equipment. All of the equipment deployed at the time is still working, ticking away like Swiss watches frozen in time in ten thousand year old ice. That is the good news. The drilling itself was a nightmare. The method turned out to be time- and fuel-intensive and "plateau-ed" at 800 meters, a depth too shallow for our ambitious project.

I guess we learned our lessons well. At Wisconsin, efficient, computer-controlled drills were developed in collaboration with drill-meister Bruce Koci from the Polar Ice Coring Office (PICO). Steve Barwick, who had left the "Price-group" in Berkeley for the University of California at Irvine, designed the first AMANDA detector with such efficiency and foresight that, even after 2 years of operation of the first 80 modules, essentially nothing has been changed for future deployments. Prior to the first construction in the 93-94 Antarctic summer the project was joined by a group of Swedish particle physicists from Stockholm and Uppsala under the leadership of Per Olof Hulth. They had abandoned their own project to convert clear Northern Scandinavian lakes into particle detectors. With incredible efficiency the NSF-ASA support people moved all detector and drilling equipment to the pole, either via Los Angeles and the Antarctic coastal McMurdo base by boat, or via Christchurch, New Zealand, and McMurdo by plane. The latter road is followed by people who, after stints of a few months in the minus thirty degree Antarctic summer, often disappear for weeks into the more tropical parts of New Zealand.

The first string of 80 PMTs was successfully deployed on Christmas eve 1993. It took 10 people 17 hours in minus 30 degrees to assemble the string. Some tasks, involving connectors and fiber optic cable, have to be handled without the protection of gloves. It does help to have heaters and to shield the open working area against wind by a suspended parachute. The AMANDA strings are assembled in-situ by attaching the PMTs with carabiners, familiar to mountaineers, to the cable carrying the signals. Little did we know that 2 days later the ambient pressure in the hole, about 100 atmospheres at 1 kilometer, would exceed five times that value or more; the dynamic range of our pressure meters was insufficient. The water column breaks the symmetry and is responsible for a nasty overpressure just before the ice finally turns solid. We lost one PMT and damaged a few connectors. Where no theory existed, the intuition of Bruce Koci came to the rescue. More heat was pumped into the bottom of the next holes which refroze very gently reaching only twice ambient pressure. The last 2 strings were essentially flawless. In the end, 73 tubes were operating as expected and only 3 were fatally damaged. On to science!

Any science was of course held hostage to calibrating the detector, i.e. one has to quantitatively understand how light propagates through kilometer-deep polar ice. At this depth the absorption length, which measures how far a typical photon travels, was supposed to be 8 meters and the scattering on remnant bubbles predicted to be negligible. Bubbles of air should be squeezed by pressure to sub-micron size at a depth of 1 kilometer. Our detector simulation programs, based on these expectations, totally failed to reproduce the observations of the well-understood beam of cosmic ray muons which we used for calibration purposes. We, for instance, observed over one hundred cosmic ray events in coincidence with the Bartol-Leeds air shower detector at the surface, where only 2 were expected. Resolving this puzzle took some time as we had to disentangle the combined effects of two complete surprises: the bubbles were not small, more like 50 microns, and the absorption length was not 8 but over 300 meters! Shortly after the deployment Gregorii Domagatzky, the leader of an experiment in Lake Baikal which is chasing similar science, showed me a paper at a conference in Venice which described the observation of large bubbles as deep as 1 kilometer in an ice core extracted at the Russian Vostok station not too far from the South Pole. This helped, but did not resolve our problems. It became a private joke among the people doing the detector simulation that all our problems would go away if only light travelled hundreds of meters in ice...

It does. With every PMT a small plastic ball had been deployed which is connected to the surface by a fiber optic cable. Using a laser, pulses of light can be pumped into the deep ice and its propagation studied with the PMT array. It turned out that the absorption length of deep South Pole ice had the astonishingly large value of ~310 m for the blue light to which the PMTs are sensitive. A value of only 8 m had been anticipated from laboratory measurements. For many applications, such as the detection of supernovae, the detector volume scales linearly with the absorption length. The discovery turned AMANDA into a supernova detector. It has been monitoring our galaxy since February 1995. A group from the German national high energy physics laboratory DESY, headed by Christian Spiering, produced the dedicated data acquisition electronics in record time. Since then they have been dividing their efforts between the Baikal and South Pole experiments, a partial marriage which will hopefully be beneficial to both efforts.

Though at first surprising, the large transparency is, in retrospect, understandable in terms of conventional optics. The PMTs are sensitive to a range of blue colors where neither atomic nor molecular excitations absorb the light. So why wouldn't the ice be infinitely transparent? The absorption characteristics of transparent solids such as LiF, CaCl, diamond, BaTiO3, and ice, are probably due to scattering from small defects, e.g. dust in polar ice, not genuine absorption. The large transparency is just a reflection of the high purity of the ice, condensed from snow that fell 18000 years ago, close to the last ice age. Our results are consistent with extrapolation of measurements of the absorption length of highly purified water, e.g. the water produced at Carleton University in Ottawa for the future SNO-neutrino observatory.

We now finally can understand the response of AMANDA to the cosmic ray beam and can quantitatively simulate all aspects of the detector response. In the meantime AMANDA has been collecting signals at the rate of about 1 every second for almost 2 years. The task of sifting through the vast amount of accumulating data for interesting science has just begun. It is, in fact, seriously slowed down by the frantic preparations for deploying another 120 PMTs during the 95-96 season Antarctic summer. They will not only expand the telescope but focus on mapping the properties of polar ice between 1 and 2 kilometers. Barring any more surprises the AMANDA group will subsequently take a first stab at the construction of a kilometer scale instrument. This effort has already been funded by the National Science Foundation as a research infrastructure grant to the University of California at Irvine and the University of Wisconsin at Madison. With the generous support of both universities, the NSF Polar and Physics Divisions, the Swedish Wallenberg Foundation and DESY our fishing expedition continues.

Like every fishing expedition, we are dreaming of that big catch!

-Francis Halzen

Note: This article was written in 1995. Subsequently, during the 1995-96 season we did in fact deploy 86 OMs at depths of 1600 - 2000 m. We find that at the larger depth, there are no bubbles, and absorption lengths remain very long. Go here for papers on preliminary results from the new detectors.
- Doug Lowder

Back to the AMANDA Home Page
Douglas M. Lowder
Department of Physics
University of California
Berkeley, CA 94720