Fluorescence telescopes observe the predicted ultrahigh-energy cutoff of the cosmic-ray spectrum

The flux of ultrahigh-energy cosmic rays is very small, and it falls steeply with increasing energy. From 10¹² to 10¹⁹ eV, the flux falls roughly like E ⁻³, where E is the energy of the primary cosmic-ray particle hitting the top of the atmosphere. If the cosmic-ray spectrum continued indefinitely with an E ⁻³ falloff, one would see only a few dozen cosmic rays per square kilometer per century with energies above 10¹⁹ eV. That’s why observers studying ultra-high-energy cosmic rays want detection facilities with effective areas of thousands of square kilometers (see the article by Thomas O’Halloran, Pierre Sokolsky, and Shigeru Yoshida in Physics Today, January 1998, page 31 ).

By 10¹⁹ eV, the cosmic-ray flux is dominated by protons of extragalactic origin. In 1966, not long after the discovery of the cosmic microwave background, Kenneth Greisen at Cornell University pointed out that the CMB should impose a rather abrupt cutoff on the cosmic-ray energy spectrum at about 6 × 10¹⁹ eV, even if protons emerge with much higher energies from distant extragalactic sources. Greisen argued that the center-of-mass collision energy of a 6 × 10¹⁹-eV proton hitting a millielectron-volt CMB photon would be just enough to excite the proton to its first excited state—the Δ(1232 MeV) resonance discovered by Enrico Fermi in 1952.

The excited state decays immediately to a nucleon and a pion. So 6 × 10¹⁹ eV is, in effect, the threshold energy for pion production by high-energy protons plowing through the ubiquitous CMB. A cosmic-ray proton starting out at higher energy would keep losing energy to pion production until, after 150 million light-years at most, it falls below Greisen’s threshold. A source of sufficiently energetic cosmic rays closer to us than that would be exempted from the cutoff. But within our neighborhood, thus delimited, there are few obvious active galactic nuclei of the kind that might be capable of producing 10²⁰-eV protons.

Because much the same argument was made at about the same time by Georgii Zatsepin and Vadim Kuzmin in Moscow, the predicted sharp flux downturn at 6 × 10¹⁹ eV is called the GZK cutoff. Observers have now been looking for it for 40 years. Its absence would suggest that there are covert sources of protons above the GZK energy within our neighborhood. The protons might, for example, be local decay products of as-yet unknown exotic particle species that can travel far through the CMB without losing energy. In 2003, the Akeno Giant Air Shower Array (AGASA) collaboration reported that its 100-km² ground array in Japan had found 11 events above 10²⁰ eV and no evidence of a GZK cutoff in 10 years of exposure. ¹ That negative finding provoked much theoretical speculation as to how nonstandard particle physics or astrophysics might trump the predicted cutoff.

After forty years

Now, at long last, the High Resolution Fly’s Eye (HiRes) collaboration writes that “forty years after its initial prediction, the HiRes experiment has observed the GZK cutoff.” ² The HiRes facility is a pair of atmospheric-fluorescence telescopes (HiRes-1 and HiRes-2) on hilltops 12 miles apart at the US Air Force’s Dugway Proving Ground in Utah. It was built under the leadership of Sokolsky and his University of Utah colleague Eugene Loh. Except for a seven-month hiatus when civilians were barred from the proving ground after the 11 September 2001 attacks, HiRes has been recording showers generated by cosmic-ray primaries with energies above 10¹⁷ eV since 1997.

At such energies, a cosmic-ray primary hitting a nucleus in the upper atmosphere initiates an enormous “air shower” consisting mostly of pions and their sequential decay products—muons, gammas, electrons, positrons, and neutrinos. AGASA and HiRes represent two different ways of detecting such a shower and determining the direction and energy of its initiating primary.

At the time of its final shutdown three years ago, AGASA was the world’s largest ground air-shower array. Such facilities record charged shower particles hitting the ground with water-Cherenkov or scintillator detector modules arrayed at kilometer intervals over a large area. A typical shower rains on half a dozen modules. Relative arrival times indicate the direction from which the primary came.

HiRes, by contrast, avails itself of the near-UV fluorescence emitted isotropically by atmospheric nitrogen molecules excited by charged shower particles. Each of the facility’s two telescopes has several dozen mirror modules focusing a different patch of sky onto an imaging array of fast photomultiplier tubes (see figure 1). In effect, HiRes provides a movie of the shower as it propagates across the sky.

The two techniques have complementary strengths and weaknesses. Whereas ground arrays can monitor around the clock, fluorescence telescopes require dark, clear nights. Fluorescence telescopes have the better energy resolution because their technique is more directly calorimetric. Nitrogen fluorescence accounts for only a small fraction of an extensive shower’s energy loss, which is dominated by ionization of air molecules. But laboratory experiments show that the fluorescence is closely proportional to total ionization loss.

Ground-array observers estimate the primary’s energy from the pattern of charged particles reaching the detectors. But that estimate ultimately depends on Monte Carlo simulations of hadron production in the first few shower-initiating collisions at center-of-mass energies far beyond what is reliably known from high-energy collider data.

Nowadays the trend is to create hybrid facilities with both ground arrays and fluorescence telescopes. The 3000-km² Pierre Auger ground array nearing completion in Argentina will be augmented by four fluorescence telescopes. And a 1000-km² array is being planned for deployment near the HiRes telescopes.

Because the fluorescence of a cosmic-ray shower increases with its energy, so does the distance at which HiRes can see it. At 10¹⁹ eV, the effective reach of each HiRes telescope is roughly that of a 20 000-km² ground array. But because conditions for recording fluorescence are dark and clear enough only about 10% of the time, ground arrays have a 10 times better duty cycle. Deducing the distance and energy of a given shower from its observed fluorescence involves constant monitoring of air clarity. For the subset of showers that are well measured by both HiRes telescopes, stereoscopic analysis yields the best energy resolution. But the first report of the GZK cutoff is based on what the group calls its monocular mode, which provides greater energy range and better statistics by independently analyzing the data from the two telescopes.

Seeing the cutoff

Figure 2 shows the energy dependence of the cosmic-ray flux as measured by HiRes-1 since 1997 and by the larger HiRes-2 since 1999. To clarify the structures of interest by suppressing the overall E ⁻³ falloff, the plot shows the measured flux multiplied by E ³. The spectrum reported by AGASA in 2003 is shown for comparison.

The AGASA data had shown the roughly E ⁻³ falloff continuing beyond 10²⁰ eV without any evident cutoff. But the HiRes data are well fitted by power-law segments that change abruptly to a much steeper E ^−5.1 falloff at (5.6 ± 0.7) × 10¹⁹ eV, in excellent agreement with the GZK prediction. The modest upward break at 4 × 10¹⁸ eV is well understood in terms of the lower-energy onset of electron–positron creation by cosmic-ray protons hitting CMB photons. From its earlier statistical analysis of shower profiles, the team concluded that protons do indeed predominate over heavier nuclei and gammas as cosmic-ray primaries at these high energies. ³

Figure 3 shows the steepness of the GZK cutoff more clearly. There, plotted against E, is the measured flux integrated over all energies from E to infinity, divided by the same integral calculated from the power-law fit below the GZK cutoff and its extrapolation to infinity. Thus if there were no cutoff, the plotted ratio would be consistent with unity out to the limits of measurement.

Instead, the ratio falls to 1/2 at 5.4 × 10¹⁹ eV, in good agreement with predicted details of the GZK cutoff. ⁴ The steepness of the falloff depends on the abundance and distribution of nearby sources capable of accelerating protons to energies above 10²⁰ eV. “The very steep E ^−5.1 falloff we’re seeing,” says HiRes co-spokesman Gordon Thomson of Rutgers University, “indicates that such high-energy sources are underrepresented in our intergalactic neighborhood.” Further examination of the flux beyond the cutoff, he says, should flesh out that indication.

The conflict resolved?

The HiRes collaboration estimates its monocular energy resolution to be about 12%. “The normalization of the 2003 AGASA spectrum becomes consistent with ours,” says Thomson, “if you simply lower the energy of every AGASA event by 30%.” Indeed at the Quarks 2006 conference in St. Petersburg, Russia, last year, AGASA’s Kenji Shinozaki reported that the collaboration’s ongoing reanalysis of its accumulated data favored a roughly 15% systematic decrease in the energy of every shower. That would pare AGASA’s 2003 claim of 11 events above 10²⁰ eV down to 6, he said, leaving a sample at the highest energies that is statistically insufficient to determine the presence or absence of the GZK cutoff. AGASA’s final result has not yet been submitted for publication.

HiRes, whose total exposure from 1997 through 2006 was four or five times AGASA’s, reported finding only 8 events above 6 × 10¹⁹ eV. But straightforward extrapolation of the HiRes power-law falloff at lower energy without a GZK break would lead one to expect 40 events. Calculating the accidental probability of such a deficit in the absence of a real physical break, the collaboration concludes that it has seen the GZK cutoff with a statistical significance of 4.8 standard deviations.

On 17 March, two weeks after the HiRes paper announcing the verification of his 40-year-old prediction was posted on the Web, Kenneth Greisen died at age 89. He was one of the inventors of the air-shower fluorescence technique in the late 1960s. Greisen tried the prototype in the hills above Cornell, only to be defeated by the damp climate. But the method soon proved itself in the clear, dry uplands of New Mexico.