Cosmogenic radionuclides reveal an extreme solar particle storm near a solar minimum 9125 years BP

Solar energetic particle (SEP) events arise from solar eruptive phenomena (CMEs and flares) that accelerate primarily protons, which can reach Earth along heliospheric magnetic field lines. Modern society is vulnerable to extreme solar storms due to impacts on communication, power systems, satellites, aviation, and astronaut safety (e.g., 2003 Halloween storms; potential hazard of the August 1972 event for Apollo missions). SEPs also affect atmospheric chemistry (e.g., ozone depletion) with possible climate implications. Prior to spaceborne observations (since the 1960s) and neutron monitor records (since the 1950s), past SEP events are inferred from cosmogenic radionuclides (14C, 36Cl, 10Be) produced by cosmic rays and modulated by solar and geomagnetic fields. Event strength is commonly characterized by fluence above 30 MeV (F30) and spectral hardness (relative proportion of >200 MeV protons). The largest hard GLE observed (GLE no.5, 1956) likely raised global 10Be production by ~5% but left little to no clear imprint in firn due to deposition noise. Three unambiguous ancient events were identified previously (774/5 CE, 993/4 CE, 660 BCE), with 774/5 CE exceeding any instrumental-era event. This study presents high-resolution multi-core 10Be and 36Cl data revealing an extreme SEP event ca. 9125 BP (7176 BCE), assesses its spectral hardness via 36Cl/10Be, estimates its fluence by scaling modern GLE spectra, evaluates timing within the 11-year solar cycle, and discusses implications for chronology synchronization.

Past work established SEP detectability in cosmogenic radionuclides and identified extreme events at 774/5 CE, 993/4 CE, and ~660 BCE using multi-radionuclide evidence. Instrumental-era GLEs, notably 1956 (hard spectrum) and 1972 (soft spectrum), have characterized 10Be and 36Cl production responses and 36Cl/10Be enhancement ratios. Theoretical production functions link spectral hardness to relative 36Cl vs 10Be yields (36Cl peaking near ~30 MeV due to 40Ar resonance; 10Be near ~200 MeV). Deposition noise and geochemical behaviors complicate interpreting absolute concentrations; relative enhancements and cross-archive synchronization mitigate these issues. Time-scale synchronization between ice-core GICC05 and tree-ring IntCal via cosmogenic radionuclides suggests a −54±6 year adjustment; high-resolution coincidence of peaks can further reduce uncertainty.

Data sources and sampling: High-resolution 10Be and 36Cl were measured in Greenland and Antarctic ice cores. NGRIP: 10Be at ~1-year resolution (11 cm sampling), 36Cl at ~4-year resolution (combining four samples). EGRIP: 10Be at ~0.85-year resolution using Continuous Flow Analysis (CFA) excess water aliquots. GRIP (~6-year) and EDML (~5-year) 10Be provide lower-resolution corroboration. Timescale: All records plotted on GICC05 adjusted by −54 years following Adolphi & Muscheler to match IntCal; EDML aligned similarly. Laboratory procedures: Ice samples processed for 10Be and 36Cl via ion exchange chromatography or direct precipitation (EGRIP), addition of carriers, AMS measurements at ETH Zurich. Blanks and standards were applied; blank corrections were negligible except for EGRIP 10Be. Baseline and enhancement computation: For each record, a baseline (mean excluding peak values) and its uncertainty (standard deviation plus measurement error) were determined over ~40–50-year windows. The event-related increase was quantified as the time-integrated concentration above baseline (integrated enhancement). Enhancement factor was defined as integrated enhancement divided by baseline; for NGRIP 36Cl (4-year resolution), corresponding four adjacent 10Be samples were used for consistent averaging. Spectral hardness assessment: The 36Cl/10Be enhancement ratio was derived using NGRIP paired data. A common baseline ratio was estimated (0.212±0.038) using the approach of O'Hare et al.; the 36Cl baseline was set as 0.212×10Be baseline, ensuring baseline variability counted once. The resulting excess ratio categorized spectral hardness relative to known GLEs. Fluence reconstruction: Assuming Greenland ice-core relative changes reflect global production changes (dominated by well-mixed stratospheric component), the average 10Be enhancement factor (3.69±0.43) was used to scale event-integrated fluence spectra of modern GLEs whose modeled 36Cl/10Be ratios matched the observed ratio. Modern GLE fluence spectra and increases in global 10Be and 36Cl production were taken from Raukunen et al., Cliver et al., and Mekhaldi et al. Scaling coefficients equaled (event 10Be enhancement)/(GLE-modeled annual 10Be production increase X). Scaled fluences F30, F200, F430 were averaged across matching GLEs to estimate the ancient event fluence, with uncertainties propagated from enhancement factors and inter-spectrum variability. Sensitivity tests: Alternative spectral reconstructions (Koldobskiy et al.) and baseline solar modulation potentials (φ=650 MV vs. φ≈300 MV) were tested; geomagnetic shielding corrections were deemed unnecessary for relative enhancements. Solar cycle timing analysis: The normalized high-resolution ice-core 10Be series (excluding the peak) around 9125 BP and a 774/5 CE stack were time-shifted and correlated with normalized modeled annual 10Be production (from neutron monitors, 1960s–2000s) to identify phase alignment with the 11-year cycle; statistical significance assessed via t-tests. Timing of the event was constrained with 14C production data and considerations of stratospheric residence time.

• Multi-core detection: Sharp, synchronous peaks at ~9125 BP in NGRIP, EDML, GRIP, EGRIP 10Be and NGRIP 36Cl, consistent with 14C production data.
• Enhancement factors (integrated enhancement/baseline):
  • 10Be: NGRIP 3.85±0.68; EDML 4.21±1.10; GRIP 3.74±0.77; EGRIP 2.98±0.70; average across 10Be records: 3.69±0.43.
  • 36Cl (NGRIP): 6.09±1.21.
  • 14C production enhancement factor: 4.5±0.5 (about 20% higher than average 10Be enhancement).
• Spectral hardness: NGRIP-based 36Cl/10Be excess ratio 1.59±0.38, indicating a hard spectrum comparable to GLE no.5 (1956).
• Duration and deposition: EGRIP 10Be peak spans ~3 years, consistent with stratospheric residence times and potential clustering of one or several short events.
• Fluence estimates (average scaled spectrum, assuming φ≈650 MV):
  • F30 = 1.64(±0.53)×10^11 protons/cm^2.
  • F200 = 1.06(±0.19)×10^10 protons/cm^2.
  • F430 = 1.80(±0.35)×10^9 protons/cm^2.
  • These values imply the event was possibly up to two orders of magnitude larger than GLE no.5 (1956). Using 14C enhancement yields ~20% higher F30; using Koldobskiy spectra gives F30 ≈ 1.27(±0.48)×10^11 with similar F200.
  • Using a lower Holocene baseline solar modulation (φ≈300 MV) increases F30 to ~2.17(±0.81)×10^11.
• Solar cycle timing: Correlations between normalized ice-core 10Be and modeled GCR-driven 10Be production show clear 11-year cycles and indicate the 9125 BP event occurred near solar minimum (EGRIP r=0.45, NGRIP r=0.51, p<0.01). A similar analysis for 774/5 CE (stacked records) yields r=0.69 (p<0.01), also near solar minimum.
• Chronology: Confirms −54-year adjustment of GICC05 to IntCal and reduces synchronization uncertainty around 9125 BP from ~6 years to ~1 year due to the global time marker from coincident 10Be and 14C peaks.

The coincident, large-amplitude 10Be and 36Cl peaks across multiple Greenland and Antarctic ice cores, together with elevated 14C production, robustly indicate an extreme SEP event around 9125 BP. The 36Cl/10Be excess ratio near ~1.6 identifies a hard proton spectrum akin to the 1956 GLE, but the reconstructed fluence suggests a vastly larger magnitude. The ~3-year signal duration matches expected stratospheric residence and mixing, supporting global representativeness of relative production changes captured in ice cores. Fluence reconstructions, while model-dependent, consistently point to an event exceeding instrumental-era extremes, underscoring underestimated space weather risk. The event's occurrence near solar minimum, corroborated also for 774/5 CE, challenges the expectation that the largest SEP events cluster around solar maxima and suggests that extreme events may not follow the same phase dependence as more common SEPs. The identified peak also serves as a precise global chronological marker, significantly improving ice-core to tree-ring timescale synchronization in this interval. Differences of ~20% between 10Be- and 14C-derived enhancements likely reflect production yield function uncertainties and geochemical differences rather than polar production biases. Overall, results refine understanding of SEP spectral characteristics, timing within the solar cycle, and the potential for rare, extreme events.

This study documents an extreme SEP event at ~9125 BP (7176 BCE) using high-resolution 10Be and 36Cl ice-core records corroborated by 14C production. The event exhibits one of the largest relative short-term 10Be enhancements observed, a hard energy spectrum (36Cl/10Be excess ratio ~1.6), and an event-integrated fluence potentially up to two orders of magnitude greater than the strongest instrumental-era GLE. Analyses indicate the event likely occurred near solar minimum, as also inferred for the 774/5 CE event, suggesting that the most extreme SEPs may preferentially occur during low solar activity phases. The synchronous cosmogenic radionuclide peak provides a precise global time marker, reducing ice-core to tree-ring timescale synchronization uncertainty to ~1 year around this period. Future work should identify and analyze additional events at similarly high resolution to test the apparent association with solar minima, better constrain spectral shapes and yields, refine fluence reconstructions under varying solar modulation, and improve hazard assessments relevant to space missions and technological infrastructure.

• Fluence reconstruction depends on scaling modern GLE spectra to an ancient event; the 36Cl/10Be ratio constrains hardness between ~30–200 MeV but not full spectral shape, introducing uncertainty.
• Baseline solar modulation potential (φ) during 9125 BP is uncertain; different φ assumptions shift fluence estimates (estimates likely lower limits at φ=650 MV).
• Production yield functions, especially for 14C below ~500 MeV, have substantial uncertainties, contributing to the observed ~20% discrepancy between 14C and 10Be enhancements.
• Ice-core records contain deposition/weather noise, potential smoothing (notably in CFA sampling), and stratosphere–troposphere exchange variability; absolute deposition is site-dependent, necessitating reliance on relative enhancements.
• Peak shapes are not diagnostic due to stochastic climatic and depositional factors; stratospheric residence time may shift observed signals by ~1 year.
• Limited resolution in some cores (e.g., GRIP, EDML) and non-negligible blanks for certain datasets may affect precision.
• While volcanic influences can affect 10Be, concurrent 36Cl and 14C peaks argue against such a cause here, but residual confounding cannot be entirely excluded.