A joint Fermilab/SLAC publication

Astronomy on ice: IceCube from start to finish


With a detector encompassing a cubic kilometer of ice almost directly beneath the South Pole, IceCube is the world’s largest neutrino telescope. Like many other neutrino detectors, IceCube uses photomultiplier tubes to peer through a clear medium, looking for faint streaks of blue Cherenkov radiation – shockwaves from particles that are moving faster than light can travel in that medium. Most of what any such detector sees is background, but occasionally an energetic neutrino collides with an atomic nucleus and produces a muon headed in the same direction – the main quarry in the hunt for neutrinos from beyond the Sun.

Francis Halzen, a theoretical physicist at the University of Wisconsin-Madison and IceCube’s principal investigator, says his inspiration for the giant telescope came in the mid-1980s, while in Hawaii with other theorists plotting ways to find the top quark. A project named DUMAND (Deep Underwater Muon and Neutrino Detection) had been underway off the coast of the Big Island since 1976, and he got interested.

A string of Digital Optical Modules, the devices used to transmit information in IceCube, goes into the ice.

A string of Digital Optical Modules, the devices that IceCube uses to detect shockwaves from particles, goes into the ice.

“Since the 1970s, everyone knew an effective neutrino telescope had to be at least a cubic kilometer in volume,” says Halzen. “The first detectors were underwater. But I saw that DUMAND was really struggling.” (Indeed, DUMAND was shut down in 1995.) Halzen made a suggestion he now deprecates as “theorists showing they were smart”: the right medium for a neutrino telescope is ice, he said – not least because it doesn’t slosh around.

“Showing we were stupid came later,” he jokes, once it became evident that the optical properties of ice were poorly understood. “People had studied bubble-free ice in the lab, but we didn’t know what we’d find in the field.” In 1992 Halzen joined with Steve Barwick, Buford Price, and others at the University of California at Berkeley to propose a pilot project  called AMANDA (Antarctic Muon and Neutrino Detector Array) to the National Science Foundation. AMANDA was intended to show that a detector in the South Polar ice would work; if it did, it would be incorporated into the full cubic-kilometer version.

“We soon learned that AMANDA was too shallow; there were too many bubbles in the ice,” Halzen says. A sufficiently sensitive detector would have to start a kilometer and a half below the surface. Still, at depth, “the ice turned out great. How clear it was had not been anticipated. We were incredibly lucky.”

Ice clarity was only one question AMANDA addressed. Drilling deep holes with hot water and sinking strings of photomultiplier tubes inside pressure-resistant glass spheres would prove especially challenging when AMANDA’s 19-string design began expanding to IceCube’s 86. But prerequisite were the optical modules themselves, and their communications with the surface.

David Nygren at Lawrence Berkeley National Laboratory is a detector designer whose inventions include the Time Projection Chamber used in high-energy colliders worldwide. When the Superconducting Super Collider in Waxahatchie, Texas was canceled in 1993, Nygren felt he faced a choice: move to CERN to work on the LHC or find a new venue. He remembers a timely call from a friend at Caltech who told him, “Dave, you gotta get involved in neutrino astronomy.”

Berkeley Lab’s Robert Stokstad recalls that “Dave went to a neutrino conference in Southern California, and I went to a Fermilab conference on cosmic-ray detectors. We compared notes, and it seemed like high-energy neutrinos would be more fun.” After flirting with the idea of detectors in the ocean off the Southern California coast, the Berkeley Lab group applied to join the AMANDA Collaboration.

At first, light emitters carried in AMANDA’s modules took the place of Cherenkov radiation; Nygren says it was lucky the blue LED had been invented in late 1993. Detecting a blue flash, a distant module would send an analog waveform that stretched and diminished as it traveled up a coaxial cable to the surface.

“Next they tried to improve the signal with optical fibers,” Nygren says. “But a hole drilled with hot water freezes from the top down and starts pushing stuff around. Optical fibers were too fragile, not to mention too expensive.”

The rugged alternative was a twisted pair of copper wires, but analog signals would degrade too much during the slow trip to the surface. Copper called for “disruptive technology,” says Nygren: the waveform had to be digitized before it left the optical module.

For his Master’s thesis, a Berkeley grad student working in Berkeley Lab’s Engineering Division, Stuart Kleinfelder, had adapted a digitizer he’d first devised for time projection chambers: the result was an integrated circuit that could digitize an analog waveform while sampling it in three-nanosecond slices. Using several channels at once, it recovered an unprecedented amount of information. What’s more, it operated at very low power.

A Digital Optical Module

A Digital Optical Module for IceCube

With help from engineer Jerry Przybylski and others, Kleinfelder’s innovative chip grew to become a Digital Optical Module (DOM). In 2000, a year after Halzen proposed the full IceCube, 41 DOMs were demonstrated on one of AMANDA’s final strings, the storied String 18 that proved IceCube could work better than promised.

To make best use of the torrent of data from the digitizers, clocks on the DOMs had to be synchronized throughout the array. Stokstad came up with “reciprocal active pulsing,” sending pulses to all the DOMs several times a second, which each DOM digitized and sent back to the surface. Among other advantages, this allows the intensity and arrival times of Cherenkov light from a particle moving through the ice to be resolved within two nanoseconds – even when a DOM’s only connection to the surface is 2.5 kilometers of telephone wire.

Many AMANDA partners were aboard IceCube by the time construction began in the winter of 2004/2005, including staunch German and Swedish supporters, and many technical challenges had been met. Yet the first seasons were cliff-hangers.

“I was sure we had a failure on our hands,” Halzen says. “We’d designed IceCube on the assumption we could drill a hole two and half kilometers deep in two days. The first season we drilled only one hole.” Given the short Antarctic summer – planes can reach the South Pole fewer than four months of the year – building IceCube at that rate would have taken over a quarter century.

The rate picked up as the team kept drilling, testing DOMs, assembling strings, and getting them down the hole before the water froze. “Coordinating all the different things that have to happen at the right time is really like a ballet,” Halzen says. By the end of the six-year construction period, IceCube was drilling and installing three strings a week. “The only real difference from when we started was the experience of the drillers.”

IceCube and AMANDA have collected meaningful data throughout their construction, but the full cubic-kilometer neutrino detector opens a new energy range, everything from the most explosive events in the universe to cold dark matter lurking in the heart of the Sun. IceCube’s 86th and final string went into the ice on December 18, 2010.

For more information:

“IceCube: An instrument for neutrino astronomy,” by Francis Halzen and Spencer R. Klein

IceCube completed, University of Wisconsin press release

Ice Cube completed, Berkeley Lab press release

IceCube website

Latest news articles

The LHCb collaboration announced the discovery of a new system of five particles all in a single analysis


Dianna Cowern—a.k.a. Physics Girl—has one of those invent-it-yourself jobs that exist only in the age of the internet.


The PICO bubble chambers use temperature and sound to tune into dark matter particles.


An animated take on dark matter, voiced by Janna Levin.