Deep-ocean ferromanganese (Fe/Mn) crusts are slow-growing metallic underwater rocks found in all major oceans on Earth (Hein and Koschinsky, 2014). The growth of the here considered class of Fe/Mn crusts is induced by hydrogenetic precipitation of vernadite and feroxyhyte out of seawater onto a sediment-free rock substrate (Koschinsky and Halbach, 1995; Halbach et al., 2017). Typical growth-rates of Fe/Mn crusts are on the order of mm/Myr. This allows to investigate Earth’s history for interstellar radionuclide influxes (Knie et al., 1999; 2004; Wallner et al., 2015; 2016; 2021; Koll, 2023) on a Myr timescale in addition to deep-ocean sediments (Fitoussi et al., 2008; Ludwig et al., 2016; Wallner et al., 2016; 2020) and Antarctic ice and snow (Koll et al., 2019). Fe/Mn crusts are rich in rare-earth elements as well as in Ni and in particular Co (0.2–2%), leading to their similarly common name Co-rich Fe/Mn crusts and making them an important resource for future industrial mining (Halbach et al., 2017).

The radioactive decay of incorporated ¹⁰Be (t_1/2 = 1.39 Myr, Korschinek et al., 2010; Chmeleff et al., 2010) is most commonly used to date Fe/Mn crusts and to determine their growth-rate. Cosmogenic ¹⁰Be is produced in the upper atmosphere by galactic cosmic rays striking nitrogen and oxygen molecules, breaking the nuclides up and forming the spallation product ¹⁰Be (Masarik and Beer, 1999). A steady cosmic ray flux and composition of the atmosphere lead to a relatively constant production of ¹⁰Be over geological timescales (Ku et al., 1982; Smith et al., 2019). The freshly produced ¹⁰Be is continuously incorporated into Fe/Mn crusts. The decay of ¹⁰Be can then be used as a radioactive clock to date the Fe/Mn crust, if the incorporation of Be into the crust stays constant over time. Several studies indicate a good constancy of Be incorporation in several different Fe/Mn crusts over the last 10 Myr (Bourles et al., 1989; Segl et al., 1984; Koll and Lachner, 2023; von Blanckenburg et al., 1996b). Differences in Be uptake in deep-ocean sediments were assigned to varying Ca concentrations in them (Feige, 2014). The Ca content of Fe/Mn crusts is low and constant and does, therefore, not influence the Be uptake (Koll, 2023).

The production rate of ¹⁰Be in the upper atmosphere of ≈  6 × 10⁵ at/cm²/yr (Masarik and Beer, 1999; Nagai et al., 2000; Webber and Higbie, 2003) and the ⁹Be concentration in sea water of ≈  6 × 10⁻⁶ mg/L (Haynes et al., 2014) lead to isotopic ratios of ¹⁰Be/⁹Be ≈ 10^–7 in sea water and, hence, in Fe/Mn crusts. This intrinsic ratio is several orders of magnitude higher than the sensitivity of AMS of 10^–16. However, the ⁹Be concentration in Fe/Mn crusts is only a few ppm (von Blanckenburg et al., 1996a), leading to less than micrograms of ⁹Be for the final AMS sample. Due to the high intrinsic ratio, either a carrier-free approach can be chosen, where < μg Be are extracted and dispersed in Fe carrier (Maden et al., 2004; Horiuchi and Matsuzaki, 2015; Lachner et al., 2020), or, more commonly, a known amount of ⁹Be carrier is added to the sample (Fitoussi et al., 2008; Koll et al., 2022b). The measured ¹⁰Be/⁹Be ratio, critical for AMS, is only reduced by a few orders of magnitude compared to the intrinsic ratio through carrier addition and still easily measurable by AMS.

The chemical extraction of Be and several other elements from Fe/Mn crusts was described in detail by Koll et al. (2022b) and Koll (2023). Here, the chemistry protocol is described with a focus on Be and more recent findings. A schematic summary of the procedure can be found in Figure 1.

Beryllium is extracted from 10–100 mg of milled Fe/Mn crust powder by first dissolving the powder in 15 mL of 10.2 M HCl and 0.25 mL of 30% H₂O₂ at room temperature. Seawater beryllium, and therefore, atmospheric ¹⁰Be that were incorporated in the leachable part of the Fe/Mn matrix are dissolved. Any unwanted in-situ produced ¹⁰Be as well as terrestrial Be in the insoluble mineral fraction of silicates and zircons are discarded. The dissolved Be is precipitated together with Fe, Al, and Ti hydroxides by the addition of NH₃(aq.) until pH = 8–9. Both, Mn and Co, can either partially or not precipitate due to its oxidation state ambivalence or the formation of an ammine-complex, respectively. Abundant elements such as Na, Mg or Ca do not form insoluble hydroxides at this pH and stay in the supernatant. The most important isobaric element for ¹⁰Be AMS, B, is present as anionic borate in solution and similarly separated. The pre-concentrated and purified precipitate is redissolved in 10.2 M HCl and charged to a 12 mL anion-exchange column (DOWEX 1X8). Iron is strongly retained on the column in contrast to Be, Al and Ti. After the column chromatography, the solution containing Be, Al and Ti is precipitated again by the addition of NH₃(aq.). The precipitate can directly be transformed into an AMS target by igniting it at 900°C since the Al and Ti contents are low and do not interfere with the ¹⁰Be AMS measurement. In case an isolated Al fraction is required, a cation-exchange column or a complexation at pH = 12–13 (Ochs and Ivy-Ochs, 1997) can be added to separate Al from Be (see Figure 1 and Subsection 2.3.2).

It is advantageous that the final AMS target contains both fractions, Be and Al, because ¹⁰Be and ²⁶Al are highly desired AMS radionuclides (see e.g., Nishiizumi et al., 1986; Granger, 2006). The measurement of both radionuclides from the same target would improve the sample throughput significantly. Typically, a good performance of Al targets with respect to ion beam current output is more critical than for Be due to the lower ²⁶Al concentration and worse negative ion formation efficiency in the sputter ion source. Therefore, a combined sample should be optimised for Al AMS. First tests, where the combined sample of BeO and Al₂O₃ was split into two and prepared as an individual Be-optimised target (mixed 1:4 wt/wt with Nb) and as an Al-optimised target (mixed 1:1 wt/wt with Ag) showed that both targets performed well in their respective radionuclide measurement. The next step is to optimise the sample preparation such that both measurements for ¹⁰Be and ²⁶Al perform well from the same AMS target.

The ¹⁰Be AMS technique to determine a Fe/Mn crust’s growth rate is well-established and the methodology at the DREsden Accelerator Mass Spectrometry (DREAMS) facility (Akhmadaliev et al., 2013; Rugel et al., 2016; Lachner et al., 2023) will be briefly explained. The prepared BeO target (mixed with Nb, 1:4 wt/wt) is inserted into the negative ion source. After sputtering, ^9,10BeO⁻ is selected and injected into the accelerator where electrons are stripped and molecules dissociate. The positively charged ^9,10Be ions are then selected after the accelerator and guided to the detector. Interfering ¹⁰B is separated from ¹⁰Be by differential energy loss in a SiN foil and the subsequent passage of an electrostatic analyser. Macroscopic currents of ⁹Be are measured in offset Faraday cups. Isotopic ratios of processed crust surface samples with previously added Be carrier are in the range of ¹⁰Be/⁹Be = 10^–10–10^–11. At the DREAMS facility, dating of ferromanganese crusts beyond 15 Myr is possible with high precision. Recently, a piece of the Fe/Mn crust VA13/2-237KD was ¹⁰Be-dated at the DREAMS facility till 18 Myr for the search of supernova-produced ⁶⁰Fe and r-process ²⁴⁴Pu (Koll et al., 2022b; Koll, 2023).

Atmospheric nuclear weapons test during the 1950s and 60s released huge amounts of actinides (e.g., U, Np, Pu, Am) and fission products (e.g., ⁹⁰Sr, ⁹⁹Tc, ¹²⁹I, ¹³⁷Cs) into the environment (UNSCEAR, 2010). These nuclides have spread globally and can be found on Earth’s surface in oceans, soils, sediments and plants. Furthermore, major nuclear accidents as happened in Chernobyl (1986) or Fukushima-Daichi (2011) contributed to the anthropogenic radionuclide signal in the environment (NEA/OECD, 2002; Onda et al., 2020). Additionally, local sources from nuclear facilities such as nuclear power plants, reprocessing or fuel production sites could be present. The isotopic signature of actinides in environmental samples gives the unique opportunity to distinguish different sources and is essential for nuclear safeguards and non-proliferation purposes. Thus, several actinide isotopes are of interest due to their different production pathways and hence their ability to account for various sources.

Many studies focused on the isotopic ratio of the plutonium isotopes ²⁴⁰Pu/²³⁹Pu, which is only accessible with mass spectrometric techniques due to the similar α energies of both nuclides (Kelley et al., 1999). Although the Pu isotopes with higher mass numbers (i.e., ²⁴¹Pu, ²⁴²Pu and ²⁴⁴Pu) are produced in much lower quantities, their assessment holds potential for a deeper understanding of the source of Pu in the environment (Steier et al., 2013). Especially the shorter-lived ²⁴¹Pu (t_1/2 = 14.3 years) and its daughter ²⁴¹Am may be used for dating the last chemical separation of both elements or the date of irradiation (Steier et al., 2013; Corcho Alvarado et al., 2022. In recent years, the rare uranium isotopes ²³³U and ²³⁶U and their atomic ratio have gained much attention, especially looking in ocean water and sediments worldwide (Hain et al., 2020; Lin et al., 2021; Qiao et al., 2022).

A compilation of expected isotope ratios for global nuclear fallout in comparison to measurements from the Chernobyl debris can be found in Table 1. Within this work we present our chemical work-up protocol for the determination of actinides (i.e., U, Pu, Am) from different environmental samples like soil, sediments or plant material. The required sample size strongly depends on the isotopes to be investigated. Several studies reported the concentration of ²³⁹Pu of anthropogenic origin for top soil layers in the order of 50–500 fg/g (Harley, 1980; Kelley et al., 1999; Muramatsu et al., 2000; Tighe et al., 2021). This corresponds to 10⁸–10⁹ atoms ²³⁹Pu/g soil (1 fg Pu = ̂ 2.5 × 10⁶ atoms Pu). As shown in Table 1 all other relevant actinide isotopes are present in smaller amounts compared to ²³⁹Pu. However, state-of-the-art AMS facilities dedicated for actinide measurements reach detection efficiencies of 1% and more (Hotchkis et al., 2019; Koll et al., 2022a; Koll, 2023) enabling routine analysis of gram-sized soil samples even for less-abundant isotopes like ²³³U and ²⁴⁴Pu.

As already introduced in Section 1, AMS is an ultra-trace analytical method measuring isotopic ratios between rare and abundant isotopes with high sensitivity. In the case of actinides there are no abundant isotopes present in nature despite the long-lived ²³²Th and ^235,238U and their daughter nuclides. Thus, for determining the content of the rare uranium isotopes ²³³U and ²³⁶U, the concentration of the primordial uranium can be analysed with less sensitive analytical methods like ICP-MS. However, the measurement of transuranium elements requires the addition of a (preferably mono-isotopic) spike of known quantity as a reference. For plutonium determination, different isotopes can be added as tracers (i.e., ²³⁶Pu, ²⁴²Pu or ²⁴⁴Pu). Isotopic purity, availability and interference with other nuclides are the major factors for selecting an appropriate isotopic spike. For many actinides the limited number of long-lived and readily accessible isotopes prevents the use of an isotopic spike. Instead, neighbouring elements may be used as non-isotopic spikes (e.g., ²⁴³Am as spike for ²⁴⁷Cm or ²⁴²Pu for ²³⁷Np). The prerequisite for using non-isotopic spikes is a similar chemical behavior during sample preparation and a well-characterised standard material to correct the AMS measurement for differences in their ionisation efficiency and transmission through the accelerator.

The general work-up scheme for different environmental samples is shown in Figure 2. Commonly, the samples are first dried in a drying cabinet at 95°C to constant mass content. Subsequently, larger grains are removed by sieving (2 mm). Afterwards, the samples are weighed and ashed in a muffle furnace at 550°C for 5 h in order to destroy all organic materials present and determine the loss-on-ignition. After ashing the residues are leached in 10 mL 8 M HNO₃ per gram sample for 2 h under reflux conditions. Although this acid leaching process does not dissolve the whole sample (silicates, zircons and refractory oxides remain) it has been shown that the majority of the targeted anthropogenic actinides are accessible (Krey and Bogen, 1987).

The separation of the actinides from the bulk sample is done by means of chemical purification steps like co-precipitation and column chromatography. For environmental samples we perform a co-precipitation of the actinides with Fe and Al hydroxides by raising the pH of the acidified solutions from leaching to pH = 8–9 with diluted NH₃(aq.). The alkali and earth-alkali metals, with the valuable exception of Be, do not form hydroxide precipitates at these pH-values and stay in the supernatant. High concentrations of iron(III) can subsequently be removed by dissolving the precipitate in 7 M hydrochloric acid and liquid-liquid extraction with diisopropyl ether. Another major constituent of soil, Al, can be decreased by making use of it’s amphoteric character, forming soluble hydroxide complexes [Al(OH)₄]⁻ at pH > 12. Generally, less matrix material facilitates a smooth column chromatography setup and avoids column blockage.

Prior to separation of the actinides from the sample using column chromatography, their oxidation state has to be adjusted. Under the given oxidising conditions in diluted nitric acid medium, uranium exclusively exists as U(VI) and americium as Am(III). Hence, only the oxidation state of Pu needs to be adjusted to Pu(IV) in order to separate the most relevant actinides according to their oxidation state and their connected chemical behavior. Addition of solid NaNO₂ and boiling for 30 min is sufficient for this purpose (Koll et al., 2022a).

In general, Pu is extracted from the remaining matrix as [Pu^IV(NO₃)₆]²⁻ complex on anion exchange resin DOWEX 1X8, U as phosphonate complex on UTEVA resin and finally Am as TODGA complex on DGA resin. Hence, a stacked three column approach is used to sequentially extract the actinides of interest. After sample loading, the columns are washed with 8 M and 4 M HNO₃ to remove matrix elements (e.g., Al, Fe, Mn). Thereafter the columns are separated and treated differently to finally obtain pure fractions of the actinides. The anion exchange column is subsequently washed with 10.2 M HCl to remove the coextracted Th. Finally, Pu is eluted by reducing it to its trivalent oxidation state with 0.1 M NH₄I - 9 M HCl. Evaporation of the solution and fuming with HNO₃ and HCl destroys the remaining ammonium iodide. U and Am can be eluted from UTEVA resp. DGA columns by 0.1 M HCl.

The final AMS target preparation is similar for all separated actinides which are subsequently either co-precipitated with iron as iron(III)-hydroxides or dissolved in HNO₃ acidic iron solution and dried. The dried residues are transferred to quartz crucibles which are calcined in a muffle furnace at 600°C for 2 h to form Fe₂O₃ with embedded actinides. The resulting iron oxide powder is mixed with metal powders (e.g., Ag, Nb) and pressed into cathodes for sputtering in negative ion sources. Nowadays, actinides are more efficiently measured at compact AMS systems Vockenhuber et al., 2013; Hotchkis et al., 2019; Steier et al., 2019. Here, we exemplify the AMS setup for the two AMS facilities Vega (ANSTO) and VERA (Univ. of Vienna). For plutonium measurements at ANSTO (Wilcken et al., 2015; Hotchkis et al., 2019), mixtures of Fe₂O₃:Nb (1:1 wt/wt) are pressed into aluminium target holders. Negatively charged PuO⁻ ions are extracted and accelerated by setting the terminal voltage to 0.86 MV. On the high-energy side Pu³⁺ ions are counted in a gas-ionization chamber. Fast isotope cycling is applied to the low- and high-energy magnets to inject up to 8 masses of different isotopes quasi-simultaneously (Hotchkis et al., 2019). For actinide (An) measurements at the VERA facility (Steier et al., 2019; Hain et al., 2020) the iron oxide material is mixed with silver in 1:1 weight ratio. Again, AnO⁻ ions are extracted and stripped to An³⁺ ions in the accelerator using 1.65 MV terminal voltage.

Stars freshly synthesise chemical elements in their interior by nuclear reactions. The life cycle of massive stars concludes with a stellar explosion, a supernova, that is ejecting most of the stars material into space, including freshly produced radionuclides. While the solar system traverses the interstellar medium, it captures interstellar dust containing live radionuclides from former supernovae (Ellis et al., 1996; Athanassiadou and Fields, 2011). These dust particles can accumulate on Earth over millions of years, leaving their isotopic fingerprints in terrestrial archives to be uncovered.

Latest endeavours focus on establishing a new geological archive for nuclear astrophysics—the ice of Antarctica. Antarctic ice serves as a unique geological reservoir for the study of extraterrestrial material due to its isolated location and the climatic conditions that preserve materials chronologically over millennia. In this project, we analyse radionuclide concentrations in continuous-flow analysis (CFA) water, extracted from an Antarctic ice core. The ice core is sourced from a drilling site near the Kohnen Station in Dronning Maud Land, Antarctica.

Our objective is to investigate an ongoing interstellar ⁶⁰Fe (t_1/2 = 2.6 Myr, Rugel et al., 2009; Wallner et al., 2015) influx by covering a time period between 50,000 and 80,000 years ago. Earlier studies showed ⁶⁰Fe activity about 2–3 Myr ago (Knie et al., 2004; Ludwig et al., 2016; Wallner et al., 2016; 2021; Koll, 2023), as well as an older influx 7–8 Myr ago (Wallner et al., 2016; 2021; Koll, 2023), both of which are attributed to interstellar dust containing traces of supernova-produced ⁶⁰Fe. Furthermore, a recent interstellar ⁶⁰Fe influx was discovered in Antarctic surface snow (Koll et al., 2019) and independently in deep-sea sediments covering the past 35 kyr (Wallner et al., 2020). A study of Antarctic ice for ⁶⁰Fe will, therefore, yield time-resolved data between the previously investigated timescales (Koll et al., 2020). In addition to ⁶⁰Fe, the atmospheric radionuclides ¹⁰Be and ²⁶Al are used to characterise the Antarctic ice sample further, whereas the interplanetary isotopes ⁴¹Ca and ⁵³Mn are utilised to distinctively differentiate interstellar ⁶⁰Fe from cosmic-ray produced ⁶⁰Fe (Koll et al., 2022a).

The rarity of the influx and deposition of interstellar radionuclides onto Earth constrains the usability of potential environmental archives. In order to clearly detect ⁶⁰Fe, a minimum of 25,000 atoms of ⁶⁰Fe must be present in a sample in order to obtain 10 detector events at the Heavy-Ion Accelerator Facility (HIAF) at the Australian National University (Wallner et al., 2023). Furthermore, the atomic ratio ⁶⁰Fe/Fe must exceed 2 × 10⁻¹⁶, which is 10 times the lowest blank level. Addressing the first condition, we consider the known recent ⁶⁰Fe deposition rate in Antarctica of 1 at/cm²/yr (Koll et al., 2019) and the pre-industrial accumulation rate of 66 mm w.e. yr⁻¹ (Oerter et al., 2000; Birnbaum et al., 2006; Medley et al., 2018). A minimum of 165 kg of Antarctic ice is necessary to contain the required 25,000 ⁶⁰Fe atoms. A typical terrestrial Fe flux of 1 mg/m²/yr (Conway et al., 2015), yields 2.5 mg of Fe in 165 kg of Antarctic ice. This results in a ratio of ⁶⁰Fe/Fe = 9 × 10⁻¹⁶, thereby fulfilling the second condition. A higher accumulation rate of 110–250 mm w.e. yr⁻¹ (Karlsson et al., 2016) and an up to 20 times higher iron content (Xiao et al., 2020) in Greenland ice would lead to a reduction of the ⁶⁰Fe/Fe ratio beyond detectability, primarily driven by the higher stable iron content.

The relatively low accumulation rates in Dronning Maud Land, however, still require high sample volumes to reach the sensitivity required to detect a few single atoms from an interstellar dust influx. Additionally, a chemically pure sample is essential to prevent a dilution in the Fe concentration. The chemical procedure to extract the aforementioned radionuclides from such large ice samples is based on the established procedure for a deep-ocean ferromanganese crust (Koll et al., 2022b) and meteorites (Merchel and Herpers, 1999). It is an update to the previously employed procedure for Antarctic snow (Koll, 2018; Koll et al., 2019). The chemical protocol is depicted in the flowchart in Figure 3.

A total of 300 kg of CFA water is acidified to pH = 2 by adding HNO₃ to prevent microbial contamination and ensure sample stability. An aliquot of the total volume is analysed by ICP-MS for already dissolved elements. The volume is then significantly reduced through evaporation over three to 4 weeks in two insulated and covered 10 L glass beakers under temperature-controlled conditions at 80–90 °C to prevent boiling. A 10 L blank sample from purified (18 MΩ) water is placed next to the beakers to monitor the introduction of contaminants. The progressing evaporation gradually increases the acid concentration of the sample, prompting repeated checks of the element concentrations by ICP-MS until the sample reaches 50 mL. The increasing acid concentration continuously leaches terrestrial elements from solid particles in the sample. If necessary, a filtration step could be included to restrict further leaching at the expense of potential radionuclide losses. Here, the low stable element concentrations did not require filtration even in concentrated HNO₃. The addition of H₂O₂ reduces any Mn(IV) to soluble Mn(II) in HNO₃. After evaporation, a final aliquot from the remaining 50 mL sample is taken for stable element determination before carrier solutions are added to the sample and the blank. The low natural concentrations of Be and Mn in Antarctic ice (Koll, 2018) require the addition of 2 mg Be and 3 mg Mn, whereas natural Al, Ca and Fe concentrations are high enough for carrier-free chemistry of the Antarctic ice sample. The typically added Be carrier mass in AMS is in the range of 200–500 μg, but here, this large natural sample with a considerable atmospheric ¹⁰Be concentration requires further dilution to reduce the expected ¹⁰Be/⁹Be ratio.