South Pole radio array sets new limits on extreme-energy neutrinos

Ultrahigh-energy cosmic rays are particles that reach Earth with extraordinary energies. Magnetic fields can bend their paths, making it hard to trace them back to their origins. Neutrinos produced in cosmic-ray interactions offer another way to investigate those sources: they are essentially undeflected and largely unattenuated as they travel. Their expected flux, however, remains uncertain.

The ARA Collaboration reports the array’s first search for a diffuse flux of ultrahigh-energy neutrinos. The analysis uses data collected from 2013 through 2023. It found zero candidates after all selection steps, against an estimated background of 0.13 events, with statistical and systematic uncertainties.

Listening for a radio impulse

Relativistic particle cascades in a material such as ice can produce a very brief radio impulse called Askaryan emission. ARA’s antennas, deployed in ice at stations near the South Pole, are designed to detect such signals. The array has five stations; one also hosts a compact phased-array detector. Not every neutrino interaction would produce a signal detectable by ARA.

The team combined data from different operating periods, accounting for changes as stations came online and the array’s sensitivity evolved. The dataset amounts to 28 station-years of livetime. After data-quality cuts, 10.6 years of array-wide livetime remained. Researchers simulated neutrino signals and used selection criteria to distinguish possible signals from backgrounds, including thermal noise, radio emission associated with cosmic rays, and human-made interference such as aircraft and snowmobiles.

To reduce the risk of tuning the selection around the result, the team developed it using a randomly selected 10% of the data. The remaining 90% was kept blinded until the selection and background estimates were finalized. Simulations accounted for the array’s changing configurations.

A null search and a flux limit

One event passed the selection before the array-wide time-isolation cut. On investigation, the team judged it likely to have come from an electrostatic discharge near a detector string. The collaboration removed it with an additional, after-the-fact cut. Thus, the final count of zero candidates is after that extra removal—not because no event ever passed the selection.

With no candidates remaining, the collaboration calculated a 90% confidence-level upper limit on the diffuse neutrino flux using the Feldman–Cousins method, with a correction described in the paper. The authors say the limit is the most stringent across experimental techniques from 50 to 1,000 exa-electronvolts (EeV). An upper limit sets a boundary on the flux consistent with the search; it is not a detection or evidence that the flux is zero.

The limit tests specific models of ultrahigh-energy cosmic-ray sources. The authors report that one model with a proton-dominated population extending beyond the Greisen–Zatsepin–Kuzmin threshold is excluded at roughly 90% confidence. That applies to the specified model, not every possible source scenario. Several other models considered predict fewer detectable events and are not excluded by this analysis.

The paper also compares its results with the high-energy neutrino candidate reported by KM3NeT. For the KM3NeT flux model, the paper’s table gives a 0.8-sigma Feldman–Cousins exclusion level. Separately, assuming an E^-2 spectrum across the energy range associated with the candidate, ARA would expect 1.6 detected events; the paper describes the non-observation as a 1.6-sigma tension in that comparison. These are different ways of characterizing related comparisons, and neither amounts to a decisive contradiction.

What the result cannot settle

A flux limit depends on estimates of the detector’s exposure and response. The paper identifies uncertainties in models of radio-wave attenuation and refraction in ice, Askaryan emission, and the neutrino–nucleon cross section. It characterizes the Askaryan-model uncertainty as dominant at low energies, at about 50%, and the cross-section uncertainty as dominant at high energies, at about 150%. These uncertainties affect how the search translates its result into a flux constraint.

This was a search for diffuse emission. It does not settle whether a particular source or brief event produces neutrinos; the authors note that targeted searches could add constraints on transient scenarios. Nor does the result identify cosmic-ray sources or rule out every proposed source model.

The analysis demonstrates that ARA’s distributed stations can be analyzed together as one observatory while controlling backgrounds. The authors present this as a practical step relevant to larger proposed radio-neutrino detectors. The source is an arXiv paper; the supplied material does not establish its peer-review status. The collaboration points to a companion paper for fuller methodological detail. The findings and model constraints should be understood in light of the analysis assumptions and uncertainties reported by the authors.