The highest-energy cosmic rays may be iron nuclei

Up against the Andes in the high plains of Argentina, the 3000-km² Pierre Auger Observatory has, since 2004, been recording the particle and fluorescence showers initiated by ultrahigh-energy cosmic-ray (UHECR) particles—mostly protons and fully ionized nuclei — hitting the top of the atmosphere. Such an enormous expanse, studded with 1600 surface detectors watched over by four fluorescence-telescope complexes, is appropriate because UHECRs become ever more scarce with increasing energy. Above 10¹⁹ eV, one can expect to find only a few dozen CRs per square kilometer per century.

A lot has been learned in the past decade from Auger and similar, albeit smaller, facilities worldwide. But a newly reported Auger result challenges much of what had become conventional wisdom. ¹ And it has sent astrophysicists and particle theorists scrambling for explanations.

Seeking to determine the nuclear identities of the highest-energy CRs, the Auger collaboration examined the development of their enormous showers of secondary particles in the atmosphere. The shower observations seem to indicate a transition, at primary-particle energies E of a few times 10¹⁸ eV, from a CR flux dominated by protons to one increasingly dominated at higher energies by iron nuclei.

A problematic surprise

The new result comes as a considerable surprise: Previously most observations and arguments seemed to support proton dominance at the highest energies. And indeed, in the weeks since its publication, the Auger paper has not gone unchallenged.

The so-called ankle of the CR energy spectrum — a well-known kink near 4 × 10¹⁸ eV—is related to the transition from galactic to extragalactic predominance in the CR flux. And the most plausible extragalactic sources are active galactic nuclei (AGNs)—galaxy cores energized by particularly voracious black holes. So a straightforward reading of the new Auger shower data would be that the most energetic nuclei arriving from AGNs are mostly Fe nuclei. But that conclusion poses problems.

For example, if the highest energy CRs really are highly-charged Fe ions, how could their distribution of arrival directions be exhibiting significant correlation with the anisotropic distribution of potential sources in the local cosmos? An Auger study of such correlations concluded last year ² that for E above 5 × 10¹⁹ eV, the observed correlation was inconsistent with an isotropic flux lacking any imprint of the distribution of galaxies within a few hundred million light-years (Mly). One would expect such an imprint for protons, which at those energies are deflected only a few degrees by intergalactic and Milky Way magnetic fields. Indeed, Auger was built with the fond hope that the highest-energy protons would point back to their sources. But the trajectories of Fe²⁶⁺ ions should be severely scrambled by the intervening magnetic fields.

Another reason for presupposing protons is related to the expectation that the highest-energy CRs originate within a few hundred Mly. Above a threshold energy of about 5 × 10¹⁹ eV (the so-called GZK—Greisen, Zatsepin, Kuzmin—threshold), a proton plowing through the cosmic microwave background is expected to lose energy by creating pions in occasional collisions with CMB photons. That energy loss creates an effective horizon of about 500 Mly beyond which a proton can’t maintain an energy above the GZK threshold. The effect was predicted to produce an abrupt falloff (the GZK cutoff) of the CR energy spectrum at about 5×10¹⁹ eV. The 2007 discovery of the GZK cutoff at the expected energy by the HiRes fluorescence-telescope collaboration was taken as strong evidence of a proton-dominated UHECR flux (see Physics Today, May 2007, page 17 ).

Proton dominance was also bolstered by the presumption that the binding energies of nuclei are too puny to survive the jolts a nucleus would experience in the accelerating engine of an AGN. On the other hand, if the acceleration is gradual enough for nuclei to survive intact, the maximum energy to which a nucleus could be accelerated in an AGN would be proportional to its charge Z, and Fe has the highest Z of any abundant species.

Furthermore, Fe nuclei can travel much farther without fragmenting in collisions with CMB photons than the lighter nuclei can. In fact, the effective horizon imposed by photofragmentation of Fe nuclei above 5 × 10¹⁹ eV is roughly the same as the GZK horizon for protons. So what appears in the CR energy spectrum to be a clear GZK cutoff indicative of protons might, in fact, be a photofragmentation cutoff for Fe nuclei. Or the observed cutoff might simply be marking the acceleration limit of some class of extragalactic sources.

The Auger data

For 3754 particularly well-measured shower events, the Auger collaboration has determined the so-called shower maximum—the penetration path length X _max in the atmosphere at which the shower reaches its maximum number of secondary particles. The mean value 〈X _max〉 for a given E, and its root-mean-square fluctuation, are thought to be informative about the nuclear-species composition of the CR flux at that energy.

Fluorescence telescopes image UHECR showers streaking across the sky in the UV fluorescence of atmospheric nitrogen excited by the charged shower particles. The total luminosity of the fluorescent streak provides a good measurement of E, and X _max is deduced from where the streak is brightest. Because fluorescence imaging requires clear, moonless nights, Auger’s telescopes record far fewer events than does the surface array of detectors that record the arrival of shower particles on the ground. For its shower-maximum analysis, the collaboration accepted only so-called hybrid events—showers recorded by at least one telescope and the surface array. The surface-array data provide good geometrical reconstruction of the shower’s axis.

The penetration length X _max to the shower maximum is given as an atmospheric column density. For the dependence of 〈X _max〉 on E and nuclear mass A, a simplified model of how UHECR showers develop in the atmosphere gives

$$〈X_{max}〉=\alpha (logE-〈logA〉+\beta ),$$

where 〈log A〉 is the mean logarithm of A in the CR flux at E, and the hadronic-interaction coefficients α and β are presumed to have no significant E dependence. The depth to which a CR shower penetrates the atmosphere before reaching its maximum development increases with the primary’s energy and decreases with its nuclear mass.

Figure 1 shows the observed energy dependence of 〈X _max〉 for the 3754 Auger hybrid events. The slope, proportional to 1 − d〈log A〉/d log E, is well described by a straight line with a kink at 2 × 10¹⁸ eV, very close to the spectral ankle. The most straightforward implication of the 〈X _max〉 data, when compared in figure 2(a) with model simulations for pure-proton and pure-Fe fluxes, is that with increasing energy, the mean nuclear mass of the extragalactic CR flux becomes heavier and heavier. The simulations don’t consider intermediate nuclear species abundant in the galactic CR flux because, above 10¹⁹ eV, their intergalactic photofragmentation horizons are all much closer than iron’s.

Figure 2(b) shows Auger’s measurements of the complementary observable RMS(X _max), the rms fluctuation of X _max from event to event at a given E. The fluctuation variable has the virtue that its expected composition dependence relies less than that of 〈X _max〉 on shower-model assumptions. In essence, one expects that the bigger the incident CR nucleus, the less random will be the hadronic cascade it engenders in the atmosphere. And indeed, the Auger fluctuation results appear to make an even stronger case for Fe dominance at the highest energies than do the 〈X _max〉 data.

Challenges and speculations

Just a month after the publication of Auger’s shower-maximum results comes a paper from the HiRes collaboration that reaches a different conclusion. ³ From 1999 to 2006 the HiRes team took UHECR data with a pair of fluorescence telescopes on hilltops 13 km apart in the Utah desert. The paper reports the final X _max analysis of the team’s total of 814 events stereoscopically recorded by both telescopes. HiRes had no surface array, but its stereoscopic capabilities served a similar function for geometric shower reconstruction.

The HiRes 〈X _max〉 observations are, within uncertainties, consistent with what Auger reports. But the HiRes data analysis, taking account of systematic effects like the growth of effective stereo aperture with increasing E, leads the team to conclude that proton dominance persists to the highest CR energies. The HiRes fluctuation data, shown in figure 3, make the disagreement with Auger quite clear. Unlike figure 2(b), they show no evidence of Fe.

Ironically, another disagreement between the two collaborations seems to point in the opposite direction. A recent HiRes study of the arrival directions of CRs above the GZK cutoff finds ⁴ a distribution consistent with isotropy and no correlation with galaxy distributions within a few hundred Mly. Isotropy is puzzling for protons and comforting for Fe dominance.

“Of course it’s possible that the northern and southern CR fluxes really are different,” says the University of Utah’s Pierre Sokolsky, who led the HiRes collaboration. That could be the case if, as some suggest, the highest CR energies are dominated by Fe from just a few local AGNs like Centaurus A—only 10 Mly away in the southern sky. “But,” says Sokolsky, “we’ll have to have a much firmer understanding of the systematics of stereo and hybrid observing before coming to such a conclusion.” For more than a year now, Sokolsky and collaborators have, in fact, been making hybrid UHECR observations of the northern sky at the new Telescope Array Project in Utah, an 800-km² successor to HiRes and Japan’s pioneering AGASA surface-array observatory.

One speculative resolution of Auger’s intramural tension between its X _max results and its claims of directional correlations is of particular interest to particle theorists. “Suppose that at the highest energies we’re seeing not iron but protons behaving in atmospheric collisions more and more like heavy nuclei,” says theorist Glennys Farrar (New York University), a member of the Auger team. The center-of-mass collision energy of a 10¹⁹-eV proton hitting a nitrogen nucleus in the atmosphere is about 500 TeV, far beyond anything that could be studied at CERN’s new Large Hadron Collider.

There’s considerable wiggle room in extrapolating the standard model of hadronic interactions to that terra incognita. But the observation that 10¹⁸-eV protons still seem to behave normally in atmospheric collisions makes it questionable that anything short of an abrupt onset of new physics beyond the standard model could account for the requisite doubling of the proton’s effective width. In the past, cosmic-ray observations have famously contributed to fundamental particle physics. “It would be wonderful if that’s happening yet again,” says Farrar.