Ancient Antarctic ice reveals three near-collapses of Earth’s magnetic shield

Deep inside an ice core drilled at Talos Dome in East Antarctica, scientists have found a chemical diary of one of the most dramatic phenomena our planet can produce: the near-collapse of Earth’s magnetic field. A team led by Alexis Lamothe, then at CEREGE in Aix-en-Provence and now at Nanjing University, measured beryllium-10 concentrations in 257 samples of ice spanning the interval from 170,000 to 270,000 years ago, and published the results in the journal Geochronology. Their record, the longest beryllium-10 sequence ever extracted from an ice core, captures three distinct episodes when the planet’s protective magnetic dipole weakened sharply, allowing more cosmic rays to flood the atmosphere and leave a measurable chemical fingerprint in the snow.

The physics behind the method is elegant. Beryllium-10 is a radioactive isotope forged when galactic cosmic rays smash into oxygen and nitrogen atoms high in the atmosphere, a process called spallation. Because Earth’s magnetic field deflects charged cosmic-ray particles, the production rate of beryllium-10 rises whenever the field weakens. Between 60 and 66 percent of the isotope is produced in the stratosphere, and models indicate that roughly half of the beryllium-10 eventually deposited over Antarctica actually originates at lower latitudes. When the geomagnetic dipole moment drops, production climbs, and the isotope rains down onto the polar ice sheet, where it is locked into annual layers of snow that compress into ice over millennia.

The star of the new record is the Iceland Basin Excursion, an event around 190,000 years ago that is widely regarded as the strongest geomagnetic excursion of the Brunhes chron, the period of normal polarity covering roughly the last 780,000 years. In the Talos Dome ice, the beryllium-10 flux rises to between 1.59 and 2.08 times its background level during a plateau lasting from about 192,000 to 185,600 years ago, with an uncertainty of roughly 1,400 years on each bound. That seven-thousand-year interval of extremely low field strength is consistent with sedimentary estimates suggesting the dipole collapsed by 70 to 80 percent, and it places the Iceland Basin Excursion in the same league as the famous Laschamps event of 41,000 years ago.

Perhaps the most striking feature of the event is its asymmetry. The Talos Dome data show a rapid decline of the dipole followed by a slow, three-step recovery, a pattern already seen in the Dome Fuji ice core and in western equatorial Pacific sediments. Intriguingly, this shape is the temporal opposite of what is observed during full polarity reversals, which typically involve a slow decay of the field followed by an abrupt recovery once the new polarity establishes itself. The authors argue that this consistent asymmetry across both the Iceland Basin and Laschamps excursions may reveal a fundamental difference between the dynamics of excursions and reversals, and they urge that it be carefully considered in future geodynamo modeling efforts.

Beyond the headline event, the record resolves two weaker disturbances that had long been mired in confusing nomenclature. A prolonged dipole decrease between about 218,500 and 206,000 years ago is associated with the Pringle Falls Excursion, while a briefer minimum centred at 242,000 years ago corresponds to the Mamaku Excursion, first recorded in volcanic rocks from New Zealand’s Taupo volcanic zone. Both show beryllium-10 flux enhancements of 1.24 to 1.63 times background. This is the first time either of these moderate-amplitude events has been discussed in an ice core, and their appearance only in flux, rather than in raw concentration, underscores a technical lesson: because snow accumulation rates varied substantially across the warm substages of Marine Isotope Stage 7, concentrations alone can mask genuine production signals.

The team also confronted a subtle artifact that could have corrupted the paleomagnetic interpretation. They identified 40 short-lived minima in beryllium-10 concentration, each confined to one or two consecutive 20-centimeter samples. These minima consistently coincide with spikes in major ion concentrations, including sodium, chloride, and calcium, which originate from such diverse sources as sea spray, crustal dust, and volcanic emissions. The statistical association is highly significant, and calcium-rich conditions in particular stand out: 32 percent of the beryllium-10 minima exceed 15 parts per billion of calcium, compared with only 13 percent of background samples. Because the ions come from unrelated sources, the team rules out atmospheric explanations and instead points to post-depositional processes deep in the ice.

Their favored hypothesis involves the migration of impurities along ice crystal boundaries. In the deepest, warmest sections of ice cores, large ice grains and enhanced impurity relocation can bind beryllium into dust-rich aggregates at grain boundaries, which the ion-exchange chemistry used to extract the isotope does not fully recover, producing apparent concentration minima. Crucially, once these artifacts are identified and removed, the long-term flux signal remains intact, and the geomagnetic record survives. The finding carries a practical warning for projects hunting ice older than 1.5 million years, such as Beyond EPICA: highly altered deep ice may require improved extraction protocols that account for in-ice remobilization of cosmogenic nuclides.

Cross-checking against other archives proved remarkably successful. The Talos Dome flux variations agree closely with the Dome Fuji beryllium-10 record over their overlapping interval, and both mirror the authigenic beryllium-10 to beryllium-9 ratios measured in marine sediment cores. Fine-scale features, including a short-lived recovery at 182,000 years ago and a flux minimum near 178,000 years ago, appear in both ice cores and ocean sediments, offering valuable tie points for synchronizing paleoclimate archives across entirely different media. One apparent discrepancy, a roughly 3,000-year offset between the oceanic and ice records, the authors attribute to uncertainties in the marine age model rather than a genuine physical lag, noting that the offset does not vary between glacial and interglacial intervals as an ocean-circulation effect would.

The study also documents a persistent puzzle of Antarctic beryllium geochemistry: absolute fluxes at Dome Fuji run nearly twice as high as those at Talos Dome, a pattern that persists into the recent past and appears unrelated to glacial-interglacial climate variability. The team suggests regional differences in atmospheric deposition, including a transition from wet to dry deposition south of 75 degrees south latitude and enhanced stratosphere-troposphere exchange over the highest domes. Importantly, this affects amplitude but not timing, so relative variations across sites still faithfully track changes in cosmic-ray production.

The implications reach beyond paleomagnetism. A reliable, well-dated record of dipole collapses provides a global synchronization tool for aligning ice, sediment, and volcanic archives across the Pleistocene, refining age models where independent dating is scarce. The Iceland Basin Excursion, with its prolonged seven-millennium low field and interglacial setting, also offers an analogue for reassessing claims that the Laschamps event disrupted atmospheric chemistry and ecosystems. And at a deeper level, each high-resolution record of the geodynamo’s erratic behavior brings scientists closer to answering a fundamental question: whether the seemingly random fluctuations of Earth’s magnetic field are truly stochastic, or the signature of deterministic but chaotic dynamics churning in the liquid iron core, two thousand miles beneath our feet.

Subject of Research: Reconstruction of geomagnetic dipole intensity variations from 170 to 270 ka BP using atmospheric beryllium-10 in the Talos Dome East Antarctic ice core

Article Title: Atmospheric 10Be from Talos Dome (East Antarctic) ice core records geomagnetic dipole intensity from 170 to 270 ka BP

Article References: Lamothe, A., Bard, E., Thouveny, N., Auriol, E., Severi, M., Traversi, R., De Angelis, M., Wilhelms, F., Mulvaney, R., Zaidi, F., Aumaitre, G., Keddadouche, K., & Baroni, M. (2026). Atmospheric 10 Be from Talos Dome (East Antarctic) ice core records geomagnetic dipole intensity from 170 to 270 ka BP. Geochronology, 8(2), 351-371. https://doi.org/10.5194/gchron-8-351-2026

Image Credits: AI Generated

DOI: 10.5194/gchron-8-351-2026

Keywords: beryllium-10, geomagnetic excursion, Iceland Basin Excursion, Pringle Falls Excursion, Mamaku Excursion, Talos Dome ice core, East Antarctica, cosmogenic radionuclides, geodynamo, paleomagnetism, ice core chronology, Marine Isotope Stage 7