Radionuclide-decay dating in ice cores

Age estimates using the radioactive decay of radionuclides in ice cores have the potential to verify and extend existing age scales. The method is currently limited by unexplained variations in radionuclide concentrations and the large masses of ice needed for their measurement.

Radionuclide production

Earth is constantly bombarded by a flux of cosmic rays, which consist mainly of protons and alpha particles, traveling near the speed of light. In our atmosphere, they collide with gases and set off a cascade of nuclear reactions that result in the production of a range of radionuclides, including 10Be, 14C, 26Al, 36Cl, and 81Kr. Because the atmospheric composition is dominated by nitrogen (14N) and oxygen (16O), most reactions produce radionuclides with a relative isotopic mass below 16; heavier radionuclides are much less abundant (Beer et al. 2012; Poluianov et al. 2016).

Radioactive-decay dating is best known from radiocarbon, which is however, not suitable for most ice-core dating due to its comparably short half-life of 5.73 kyr (Beer et al. 2012). Several other radionuclides have much longer half-lives (Fig. 1a), but face two major challenges. Firstly, the low abundance of most radionuclides translates into low concentrations in ice, necessitating large sample masses for their measurement, as shown in Figure 1b. Secondly, the production of all radionuclides varies over time because the cosmic-ray flux on Earth is not constant. It is subject to modulations from the varying magnetic fields of the Sun and the Earth, since protons and alpha particles are charged particles.

Figure 1: (A) Comparison of the half-lives of different radionuclides and (B) the mass of 1-Myr-old ice needed for measuring their respective concentrations.

The advantage of radionuclide dating is that it can produce absolute age estimates, independent of other age scales and time markers. Therefore, it can also be applied to ice with a disturbed stratigraphy, making it a valuable tool for the dating of deep ice cores.

36Cl/10Be

With a half-life of t1/2 = 301 kyr, 36Cl is well suited for dating ice with ages up to 1.5 Myr, which is what the Beyond EPICA Oldest Ice Core project aims to retrieve. For a sample with an age of 1 Myr, about 1 kg of ice is needed (Fig. 1b). Variations in the production signal affect 36Cl and 10Be similarly, so taking the 36Cl/10Be ratio theoretically removes production-related variations and isolates the decay signal (Delmas et al. 2004). However, measurements from the GRIP (Yiou et al. 1997) and Dome Fuji ice cores reveal significant variations of the 36Cl/10Be ratio unrelated to radioactive decay (Fig. 2a; Kanzawa et al. 2021; ca 7450–7350 yr BP). Most likely, these variations are linked to the different chemical and physical properties of the two radionuclides, which may lead to differences in their atmospheric lifetimes, transport pathways and deposition mechanisms (Beer et al. 2012).

At low accumulation sites, 36Cl can be lost from the ice by reacting with acidic species to form volatile hydrogen chloride (HCl) after deposition, further complicating the interpretation of the measured 36Cl/10Be ratio (Delmas et al. 2004). Ice from glacial periods, however, has the potential to be unaffected by this because the glacial atmosphere contains higher concentrations of alkaline dust, which can neutralize acidic species (Röthlisberger et al. 2003; Wolff et al. 2010). Another uncertainty for age estimates is introduced by the accumulation of 10Be at grain boundaries, where it can be adsorbed on to dust particles in deep ice, creating local concentration spikes and disturbing the stratigraphic signal (Baumgartner et al. 1997; Raisbeck et al. 2006).

Nonetheless, the 36Cl/10Be ratio has been successfully used to date the deepest section of ice from the Dye3 and GRIP cores, where the age estimates agreed with other dating methods (Willerslev et al. 2007).

26Al/10Be

The advantage of using 26Al instead of 36Cl is that it quickly attaches to aerosol particles, similar to 10Be. The transport for both radionuclides should, therefore, be identical, leading to less variations in the 26Al/10Be ratio. Indeed, measurements of atmospheric air around the globe yielded the same 26Al/10Be ratio with deviations of no more than 5% and similar values for measurements in firn from several locations in Antarctica (Auer et al. 2009).

The disadvantage of using 26Al is its very low production rate, which is about 300 times lower than that of 10Be, necessitating at least 7–14 kg of ice for a single measurement of 1-Myr-old ice (Auer et al. 2009; Fig. 1b).

In a pilot study, Auer et al. (2009) measured the 26Al/10Be ratio in the deep, undated section of the EDML ice core (older than 150 kyr) and found that its values varied strongly between samples and were on average 50% higher than in samples from the modern-day atmosphere and Antarctic firn, even though the decay of 26Al (t1/2 = 717 kyr) should lead to lower values (Fig. 2b). The authors concluded that recrystallization and high pressure may result in local concentration enhancements at the bottom of the EDML core. However, these alterations and their effects on 26Al and 10Be are not understood, and the 26Al/10Be ratio, therefore, appears to suffer from a similar weakness as the 36Cl/10Be ratio: although changes in the production signal are theoretically removed, the ratio exhibits unexplained variations.

Figure 2: Variations of (A) the 36Cl/10Be ratio in the Dome Fuji ice core with the theoretical production rate value of 0.0862 shown as a dashed horizontal green line and (B) the 26Al/10Be ratio in the EDML ice core with the modern-day mean atmosphere value of 1.89 ± 0.05 x 10-3 shown as a horizontal magenta line. Data from Kanzawa et al. (2021) and Auer et al. (2009), respectively. Vertical bars indicate analytical errors.

Using only the two deepest measurements, Auer et al. (2009) arrived at an age estimate of 670 kyr with an uncertainty of almost 40% for the deepest EDML sample, which is approximately the minimum possible uncertainty of 26Al/10Be dating with 7 kg of ice, mainly due to the low measurement efficiency.

81Kr

Contrary to the other radionuclides discussed so far, 81Kr is a noble gas, which is largely inert and remains in the atmosphere for most of its lifetime, resulting in a globally well-mixed atmospheric 81Kr/Kr ratio.

The approach for taking production variabilities into account is also different for 81Kr. Instead of using a second radionuclide to correct the signal, a reconstruction of geomagnetic field intensities is used to calculate the theoretical 81Kr/Kr ratio of the last 2 Myr (Zappala et al. 2020). This introduces uncertainties from the past cosmic-ray flux, the absorption cross sections of krypton, and the half-life of 81Kr, t1/2 = 229 ± 11 kyr. Nonetheless, the calculation of the theoretical present-day ratio agrees with measurements of modern air (Zappala et al. 2020).

First measurements of 81Kr in three deep samples of the undated Talos Dome ice-core section yielded age estimates with 9–16% uncertainty, and indicated a disturbed stratigraphy, because the deepest sample had a younger 81Kr age than the second deepest (Crotti et al. 2021).

Due to the low abundance of stable krypton (the target element for 81Kr production in the atmosphere) the production rate of 81Kr is even lower than that of 26Al. Measurements with less than 10 kg of ice became feasible only recently (Crotti et al. 2021) and current improvements aim to reduce the required sample mass to just 1 kg of 1-Myr-old ice (Ritterbusch et al. 2022).

Outlook

Several radionuclides have the potential to assist conventional dating of ice cores, especially in the deepest section, where the stratigraphy may be disturbed.

Two main issues complicate the use of radionuclide dating: uncertainty and required sample mass. For the 36Cl/10Be and 26Al/10Be ratios, variations occur over time and are not well understood, while 81Kr suffers from uncertainties connected to the calculation of the historic 81Kr/Kr ratio and its half-life. Because radioactive decay is used for dating, the required sample mass increases exponentially with age. To measure a radionuclide with consistent precision, the required sample mass doubles for each half-life.

Nonetheless, measurement techniques are constantly improving to reduce the required sample size, making radionuclide dating a more viable solution for dating old ice. Simultaneously, research aimed at a better understanding of climatic influences and post-depositional effects on the 36Cl/10Be and 26Al/10Be radionuclide ratios, as well as improved calculations of the historic 81Kr/Kr ratio, are expected to reduce the uncertainties of these three radionuclide dating methods.

affiliations

1Department of Geology, Lund University, Sweden

2Department of Earth Sciences, University of Cambridge, UK

contact

Niklas Kappelt: niklas.kappelt@geol.lu.se

references

Auer M et al. (2009) Earth Planet Sci Lett 287: 453-462

Baumgartner S et al. (1997) Nucl Instrum Methods Phys Res B 123: 296-301

Beer J et al. (2012) Cosmogenic Radionuclides. Springer, 293 pp.

Crotti I et al. (2021) Quat Sci Rev 266: 107078

Delmas R et al. (2004) Tellus Ser B 56: 492-498

Kanzawa K et al. (2021) J Geophys Res Space 126: 10

Poluianov S et al. (2016) J Geophys Res Atmos 121: 8125-8136

Raisbeck GM et al. (2006) Nature 444: 82-84

Ritterbusch F et al. (2022) 3rd IPICS Open Science Conference, Crans-Montana, Switzerland

Röthlisberger R et al. (2003) J Geophys Res Atmos 108: 1-6

Willerslev E et al. (2007) Science 317: 111-114

Wolff E et al. (2010) Quat Sci Rev 29: 285-295

Yiou F et al. (1997) J Geophys Res Oceans 102: 26783-26794

Zappala J et al. (2020) Geophys Res Lett 47: 1-7