1. INTRODUCTION
Recently there have been debates about the definition and the start of the Anthropocene regarding human alterations of Earth's environments (Zalasiewicz and others, Reference Zalasiewicz, Kryza, Williams, Waters, Zalasiewicz, Williams, Ellis and Snelling2014; Lewis and Maslin, Reference Lewis and Maslin2015; Monastersky, Reference Monastersky2015). The long-lived 239Pu nuclide was suggested to be considered as a globally synchronous stratigraphic marker for defining the beginning of the Anthropocene (Waters and others, Reference Waters2015). In 2009, at the 35th International Geological Congress in South Africa, the Anthropocene Working Group (AWG) first advised the International Commission on Stratigraphy (ICS) on the possibility of formally adding the Anthropocene as an interval to the International Chronostratigraphic Chart. Anthropocene was first suggested to reflect pervasive human alternations of Earth's environments (Crutzen and Stoermer, Reference Crutzen and Stoermer2000; Crutzen, Reference Crutzen2002). Since then, there have been increasing debates about defining the Anthropocene and finding the stratigraphic markers that reflect the human impacts on Earth (Waters and others, Reference Waters2014, Reference Waters2016; Lewis and Maslin, Reference Lewis and Maslin2015; Monastersky, Reference Monastersky2015; Ruddiman and others, Reference Ruddiman, Ellis, Kaplan and Fuller2015; Zalasiewicz and others, Reference Zalasiewicz2015). Recent indications are that the Great Acceleration marked by a dramatic increase in human population, large changes in natural processes (Steffen and others, Reference Steffen, Crutzen and McNeill2007; Canfield and others, Reference Canfield, Glazer and Falkowski2010; Wolfe and others, Reference Wolfe2013) and an enhanced production of new human-made materials such as plastics, organic and inorganic pollutants (Ford and others, Reference Ford, Price, Cooper, Waters, Waters, Zalasiewicz, Williams, Ellis and Snelling2014) since the 1950s has the highest support within the AWG for defining the beginning of the Anthropocene (Waters and others, Reference Waters2015; Zalasiewicz and others, Reference Zalasiewicz2015). The long-lived 239Pu nuclide resulting from the fallout of nuclear weapons testing (NWT) was subsequently suggested as the marker for a globally synchronous stratigraphic boundary for the mid-20th century (Monastersky, Reference Monastersky2015; Waters and others, Reference Waters2015; Zalasiewicz and others, Reference Zalasiewicz2015).
Plutonium in the air is now dominated by atmospheric discharge from nuclear weapons tests and the re-suspension of plutonium-bearing soil particles (Choppin and Morgenstern, Reference Choppin and Morgenstern2001). Extensive nuclear weapon tests were conducted by the USA and the former Soviet Union in the mid-20th century, and tests on a smaller scale were done later by the UK, France, China and India, leading to the most fallout of plutonium in the atmosphere during 1945–1962. The global fallout of plutonium concentration peak in sediments marked as 1963 in the northern atmosphere has commonly been used to construct the time sequence of lacustrine sediments (Ketterer and others, Reference Ketterer, Hafer, Jones and Appleby2004) and ice cores (Olivier and others, Reference Olivier2004; Wendel and others, Reference Wendel2013). Besides dating, plutonium isotope ratios, especially 240Pu/239Pu ratios have been used to differentiate the plutonium contribution from general fallout from other emission sources. Depending on emission source, the 240Pu/239Pu atom ratios vary significantly. It is well known (Krey and others, Reference Krey and Horn1976; Perkins and Thomas, Reference Perkins, Thomas and Hanson1980; Koide and others, Reference Koide, Bertine, Chow and Goldberg1985) that this ratio has a value of about 0.18 if the plutonium fallout stems from the NWT by US and Soviet tests that ejected the debris far into the stratosphere followed by transfer to the Earth surface within 1–2 years by sedimentation and wet precipitation. 240Pu/239Pu ratios from nuclear weapon-grade materials range from 0.01 to 0.07 (Micholas and others, Reference Micholas, Coop and Estep1992). Values from nuclear reactors are up to 0.4 or even higher (Chamizo and others, Reference Chamizo, Enamorado, Garcia-Leon, Suter and Wacker2008).
In addition, recent interest mostly from marine scientists arose in 236U as a tracer for modelling ocean circulation (Sakaguchi and others, Reference Sakaguchi2012; Winkler and others, Reference Winkler, Steier and Carilli2012). The reason is the extremely conservative behaviour of this radionuclide in the marine environment that makes it an ideal tracer for such studies. The average residence time of uranium in a marine environment in the form of UO2(CO3)3 4− is expected to be very long, about 0.3–0.5 million years (Bloch, Reference Bloch1980; Dunk and others, Reference Dunk, Mills and Jenkins2002). One drawback for application of this ‘new’ tracer in environmental research is, however, the poor knowledge of its deposition rates on the Earth surface. 236U was mostly ejected into the atmosphere during NWT. Estimated values are ~1 tonne (Sakaguchi and others, Reference Sakaguchi2009). Other sources of anthropogenic 236U are the reprocessing plants in La Hague and Sellafield, respectively, with a total release of ~100 kg (Casacuberta and others, Reference Casacuberta2014; Christl and others, Reference Christl2015a). The amount of natural 236U is much lower, ~35 kg (Steier and others, Reference Steier2008). This results in an estimated pre-anthropogenic 236U/238U value of ≈10−13 in ocean water (Christl and others, Reference Christl2012).
In 2005, two ice cores to drilled bedrock (58.7 m for core 1 and 57.6 m for core 2) were recovered from a dome on the Miaoergou Glacier, eastern Tien Shan, Central Asia (43°03′19″N, 94°19′21″E, 4512 m a.s.l.) (Fig. 1). In Liu and others (Reference Liu2011), a dating was suggested down to 17 m for core 2 based on annual layer counting of the assumed seasonality of δ 18O and crustal species (Ca2+, Ba2+). Moreover, a total β activity measurement was performed that depicted two peaks at ~10.5 and 12.5 m depth, respectively. The maximum at 12.5 m depth was assigned to the nuclear weapons fallout peak in 1963 and the younger maximum at 10.5 m to possible fallout from the nearby Chinese nuclear test site Lop Nor, which is at a distance of ~400 km SW of the glacier drill site. From 16 October 1964 to 29 July 1996, 23 atmospheric and 22 underground nuclear tests were conducted at Lop Nor (Yang and others, Reference Yang, North and Romney2000). Later, an additional dating was performed using 210Pb (Wang and others, Reference Wang2014). This measurement enabled to date the core down to 41.5 m. In general, the agreement between the two dating attempts was observed, though the accuracy was rather limited with a time resolution of no more than ~5 years.
Fig. 1. Location of the Miaoergou flat-topped glacier in eastern Tien Shan, China. The black dot shows the position of the drilling site (43°03′19″N, 94°19′21″E, 4512 m a.s.l.).
Later discussions questioned the assignment of the first β activity peak found in Liu and others (Reference Liu2011) to fallout from Lop Nor. In order to verify this assignment, we decided to analyse the 240Pu/239Pu atomic ratio for the segment of the core that exhibited increased total β activity, hence between ~8 and 15 m depth. The 236U analysis was also included in order to increase knowledge on 236U deposition rates. So far, only one 236U measurement exists using an ice core from Svalbard (Wendel and others, Reference Wendel2013). It is well known that fallout from NWT exhibits a significant deposition rate dependence as a function of latitude. Hence, the measured value from (Wendel and others, Reference Wendel2013) at 79.83°N of 1.63 × 108 atoms·cm−2 does not represent expected values at mid- or low latitudes, which underlies the request for additional measurements in non-polar areas.
Given the extremely low concentrations of all three radionuclides (239,240Pu and 236U) in ice from glaciers, the highest sensitivity is required for their analysis. This prerequisite is met best by accelerator mass spectrometry (AMS). Only one out of the three radionuclides, 239Pu, has so far been measured in ice cores with another mass spectrometric technique, ICP-SFMS (Gabrieli and others, Reference Gabrieli2011; Arienzo and others, Reference Arienzo2016).
2. EXPERIMENTAL
2.1. Ice core treatment
The ice cores were transported in a frozen state after drilling to the State Key Laboratory of Cryospheric Sciences, Chinese Academy of Sciences in Lanzhou. For further analysis, the 57.6 m long ice core – also analysed for 210Pb (Wang and others, Reference Wang2014) – was used. The core segments from 8 to 15 m depth were cut into 11 0.5–1 m long pieces. Each piece weighted between ~1 and 2 kg. The core segments were then subjected to chemical treatment, first in Lanzhou (part a) and then at Bern University (part b), prior to measurement of the wanted radionuclides at the AMS facility at ETH Zürich, see below.
2.2. Chemical treatment
Part (a)
Each ice sample was melted and acidified to pH <1 with HNO3 (65%) of ultrapure quality. The solutions were stored for ~12 h (overnight) and then filtered to remove mineral dust. A 100 mg KMnO4 (dissolved in H2O) was added and left for 30 min in order to oxidize all Pu species to the oxidation state VI. Subsequently, NaOHconc was added until pH reached a value of ~9. Then, 200 mg NaHSO3 (dissolved in H2O) was added to form MnO2 and left for ~1 h to let the precipitate settle. The precipitate was filtered through a membrane filter. The solution was subjected again to the entire procedure as described above and both precipitates were stored in a plastic container for transport to Switzerland. The chemical treatment closely followed a described procedure from (Olivier, Reference Olivier2004).
Part (b)
Both filters of each sample were dissolved in 20 mL 1 M HNO3 together with 100 mg Mohr's salt [(NH4)2Fe(SO4)2·6H2O]. Twenty microlitres of 233U and 100 µL 242Pu tracers were added to the solution. The standards contained 91.29 pg·g−1 233U and 9.16 pg·g−1 242Pu, respectively. Hence, the total amounts of added standards were 1.826 pg 233U and 0.916 pg 242Pu, respectively. It was therefore assumed that the chemical treatment described in part (a) was quantitative (100% yield).
Then 20 mL concentrated HNO3 and 0.5 g NaNO2 were added and the solution gently heated until the formation of nitrous gases stopped.
Purification and preparation of Pu for AMS measurement
A DIONEX 1 × 8 column (6 g) was pre-conditioned with 40 mL 8 M HNO3. The solution from above was loaded onto the column, then the beaker rinsed four times with 10 mL 8 M HNO3 and all solutions eluted through the column. All eluates were combined for further preparation of uranium (see below).
The column was then rinsed with 50 mL HClconc to remove traces of Fe, Am, Th and Ca prior to eluting Pu with 50 mL freshly prepared 0.1 M NH4I/9 M HCl solution. The eluate was evaporated to dryness, fumed three times with 5 mL HNO3/0.5 mL HClconc and then three times with 5 mL HClconc. The residue was taken up with 2 mL HClconc and then 7 mL H2O added that contained 1 mg Fe3+. Twenty-five per cent NH3 solution was added dropwise until Fe(OH)3 precipitation formed. Through heating in a water bath, the pH was reduced to a value between 8 and 9. The precipitate was centrifuged for 15 min and the supernatant decanted. The precipitate was dried in an oven at 100 °C for 2 h, and then transferred to 1.5 mL Eppendorf vials.
This part of the chemical separation also closely followed a procedure described by Olivier (Reference Olivier2004).
Purification and preparation of U for AMS measurement
A UTEVA column (0.5 g; 100–150 m) was preconditioned with 3 M HNO3. The U solution in 8 M HNO3 from above was loaded onto the column. U was eluted from the column with 30 mL 0.01 M HCl and the eluate evaporated to dryness. The residue was taken up with 2 mL HClconc followed by the same procedure as described above for Pu.
2.3. AMS measurement
The AMS measurements of 236U and Pu isotopes were performed with the compact low-energy system Tandy at ETH Zürich. The system is well suited for the sensitive detection of ultra-trace amounts of actinides. Details of the AMS setup for actinide measurements have been described previously (Christl and others, Reference Christl2013). The compact AMS system at ETH Zürich combines high transport efficiency (35–40% transmission) (Vockenhuber and others, Reference Vockenhuber2011) with the highest abundance sensitivity (of the order of 10−12) (Christl and others, Reference Christl2015b), so that detection limits at the sub-femtogram level can be reached for the actinides (Dai and others, Reference Dai, Christl, Kramer-Tremblay and Synal2011) even if an intense potential interference is present on a neighbouring mass (e.g. 238U interfering with 239Pu or 235U with 236U).
Measured 236U/233U (Pu-isotopic) ratios were normalized to the ETH Zürich in house standard ZUTRI (CNA) (Christl and others, Reference Christl2013) and corrected for impurities carried by the 233U (242Pu) tracer. The reported one sigma uncertainties (Table 1) take into account counting statistics, the scatter of the isotopic ratios during repeated measurements of the same sample, the uncertainty of standard normalization and the uncertainty of the blank correction.
Table 1. Concentrations of 239Pu, 240Pu and 236U in the Miaoergou ice core (uncertainties represent counting statistics (1 σ))
a Concentration of plutonium in the plutonium fraction.
b Concentration of plutonium in the uranium fraction.
c No Pu was detected in the uranium-fraction.
d Sample lost due to technical problems during chemical separation.
Pu isotopes were measured in both, the Pu and U fraction of the ice core samples (Table 1). While the 240Pu/239Pu ratios were indistinguishable in both fractions (Table 3), the x Pu/242Pu ratios (x = 239, 240) in those fractions were significantly different for some samples. Possible explanations for this observation are described below.
The fact that during chemical preparation the Pu spike obviously did not behave like the Pu isotopes in the sample makes the straight forward application of the isotope dilution method for the calculation of Pu concentrations impossible. Nevertheless, concentrations of Pu isotopes in each fraction could be estimated using the following procedure. To make different AMS runs (U and Pu fractions) comparable, the counting rate of the in house CNA standards was used for normalization. Under the assumption that no 242Pu was lost during chemical preparation (all 242Pu is found either in the U or Pu fraction), the amount of 242Pu present in either fraction was estimated for each sample by comparing the normalized average counting rates of 242Pu during each (U and Pu) run. With the estimated amounts of 242Pu in each respective fraction, the isotopic dilution method now could be applied to calculate 239Pu and 240Pu concentrations in the U and Pu fractions separately (Table 1). The uncertainty of this estimation is most probably larger than the reported analytical uncertainties reported in Tables 1–3. However, it cannot be easily estimated since it rather represents a systematic error affecting the accuracy of the results. To get an impression of the uncertainty associated with the above described procedure, it was applied twice for some samples. The results of the two independent experiments differ by 15% on average.
Table 2. Total 239Pu, 240Pu and 236U concentrations and activities
a Sample lost due to technical problems during chemical separation.
Table 3. 240Pu/239Pu (at·at−1), 236U/239Pu (at·at−1) and 239Pu/242Pu (at·at−1) ratios
a 240Pu/239Pu or 239Pu/242Pu in the plutonium fraction.
b 240Pu/239Pu, 236U/239Pu or 239Pu/242Pu in the uranium fraction.
c Pu was not detectable in U fraction.
d Uranium fraction was lost.
3. RESULTS AND DISCUSSIONS
The measurements were performed for: (i) the samples prepared as described above, (ii) for spike samples and (iii) for blank samples in a well-defined protocol. This enabled measurement at the highest possible level of accuracy and under clean conditions.
Therefore, Table 1 which first lists the total weight of each sample then summarizes all results of the 11 ice core segments. This not only includes measured concentrations of 239Pu, 240Pu and 236U in the corresponding chemical fractions, but also 239Pu and 240Pu concentrations in the 236U fraction. The reason why after absolutely identical chemical procedures part of plutonium was found in the uranium fraction is unclear. A tentative assumption is that in some samples, part of plutonium did exist in an unexpected chemical form.
Table 2 lists the total mass values for 239,240Pu and for 236U in each segment, but also gives the resulting activities. The reason is that in the literature, published data have been presented sometimes in mass units and sometimes in activity values. Therefore, comparing our results with literature values becomes easier.
The total 239Pu concentration along the ice core is depicted in Figure 2. It clearly shows a double peak structure. We assign the maximum at ~10.5 m to fallout from ~1963, while the maximum at ~12.5 m is assigned to the pre-moratorium fallout peaking at ~1958. In the previous work by Liu and others (Reference Liu2011), the total β activity peak assigned at ~12.5 m to fallout from 1963 should be modified and the annual layer counting result in that work was overestimated by ~5 years. In Figures 2 and 3, we applied the new time scale as best guess from the annual layer counting, the 210Pb dating (Wang and others, Reference Wang2014) and the nuclear weapons fallout (Pu&U) peaks at 1963 and 1958 in the same ice core, respectively. The uncertainty for each date is estimated as ±1 year. From the two time markers, an approximate annual average deposition rate of 25.4 cm results, or ~20.5 cm w.e., in good agreement with the values of 22.9 cm w.e. observed for a much longer time period, 1851–2005 AD (Wang and others, Reference Wang2014).
Fig. 2. Concentrations of 239Pu in the Miaoergou ice core as a function of depth (left axis). The right axis indicates the ‘best guess’ age (see text). Error bars refer to 1 σ.
Fig. 3. Concentrations of 236U in the Miaoergou ice core. The axes are identical to those in Figure 2. Error bars refer to 1 σ.
The total deposition of 239Pu integrated over the NWT period amounts to 1.55 × 109 atoms·cm−2, obtained from the data summarized in Table 1 and the diameter of the ice core of 9.4 cm. This value is higher compared with values reported for glaciers from the European Alps with 0.9 × 109 atoms·cm−2 at Col du Dome, a site close to Mont Blanc in France and 0.7 × 109 atoms·cm−2 at Colle Gnifetti, Swiss Alps (Gabrieli and others, Reference Gabrieli2011) but lower than 3.6 × 109 atoms·cm−2 obtained from an analysis of an ice core at Belukha, Altai (Russia) (Olivier and others, Reference Olivier2004). Other literature values are 1.7 × 109 measured in a Greenland ice core (Koide and others, Reference Koide, Michel, Goldberg, Herron and Langway1982) or 0.54 × 109 at the Agassiz ice cap (Kudo and others, Reference Kudo2000), respectively. In a recent publication (Arienzo and others, Reference Arienzo2016), a comparison was made between 239Pu concentration in Arctic and Antarctic samples, indicating that they differ by about a factor of three (Northern hemisphere (NH) to Southern hemisphere ratio). The maximum value for 239Pu from several Arctic sites (~5 mBq·kg−1) agrees well with our maximum value of 5.5 mBq·kg−1 (see Table 2).
All values listed above for the total flux of 239Pu during NWT in NH vary within a factor of about seven and do not scale with latitude. This indicates that local influences such as annual deposition rates, altitude, wind erosion, dry/wet deposition significantly influence the inventories. With the higher fallout of 239Pu in the mid-latitudes, the ice cores from the eastern Tien Shan and Altai may be very suitable for defining the beginning of Anthropocene.
Figure 3 depicts the deposition rates of 236U. Due to technique problems, one sample was lost during handling. Unfortunately, it turned out to be the sample with the presumed highest mass (or activity). We decided to extrapolate the missing value assuming the shape of the peak being identical to the shape of the 239Pu peak. The reason is the same nuclear process that forms both radionuclides in an explosive scenario (neutron caption with uranium isotopes). Therefore, the 239Pu/236U ratios from the two neighbouring samples were used to estimate the missing 236U value. We indicate the uncertainty in the described procedure using dashed lines in Figure 3. On the basis of this assumption, a total fallout of 236U from NTW can be estimated to be 3.5 × 108 atoms·cm−2. This value agrees rather well with 1.63 × 108 atoms·cm−2 (Wendel and others, Reference Wendel2013) from the Arctic site Svalbard 79.83°N based on the expected trend of deposition rate with latitude. However, as explained above for 239Pu, also for 236U the deposition rates are expected to vary largely from site to site.
Table 3 lists the 240Pu/239Pu atomic ratios in both the plutonium and uranium samples, as well as the 236U/239Pu atomic number ratios in the uranium samples. The values for 239Pu/242Pu atomic ratios are also listed, which were measured in the plutonium and uranium fractions to underlie the large variability. Within uncertainties, in all except one (sample 10), the 240Pu/239Pu ratios average ~0.18 ± 0.02. This value is well known to represent average fallout from the NWT where bomb debris from US and USSR tests were injected into the stratosphere, mixed due to the long residence time of ~2 years prior to re-entering the troposphere again and deposition on the Earth surface mostly by wet precipitation. Hence, no indication is found for debris from the nearby Lop Nor Chinese test site. Soil measurements in the vicinity of Lop Nor yielded average values of 0.158 (Bu and others, Reference Bu, Ni, Guo, Zheng and Uchida2015).
Sample 10 indicated the deposition ~1953 ± 1 according to the cross-check result of multiple dating methods in the ice core. The first high-yield nuclear device (Ivy Mike. 5.7 Mt fission) was detonated in November 1952 at Enewetak Atoll in the Marshall Islands with an estimated locally deposited fission energy of 2.9 Mt (Lindahl and others, Reference Lindahl2011). The mushroom cloud created by the explosion rose to an altitude of 37 km reaching the stratosphere (Machta and others, Reference Machta, List and Hubert1956). After the rapid transmission in the stratosphere and stratosphere–troposphere exchange, the fallout deposited globally. A higher 240Pu/239Pu ratio of 0.46 was observed for the Ivy test (Lindahl and others, Reference Lindahl2011). A more recent study of nuclear weapons produced 236U, 239Pu and 240Pu archived in a Porites Lutea coral from Enewetak Atoll (Froehlich and others, Reference Froehlich, Tims, Fallon, Wallner and Fifield2017) showed that in this archive the measured 240Pu/239Pu ratio was between 0.6 (1952) and 0.2 (1954) (i.e. significantly higher compared with younger samples) and the 236U/239Pu ratio varied for the same time period between 0.04 and 0.1 (i.e. lower than for younger samples). Both observations are in line with our data for sample 10 with the ratios for 240Pu/239Pu of 0.27 and for 236U/239Pu of 0.12, respectively.
The scatter of the 236U/239Pu ratios is larger than for the 240Pu/239Pu ratios. The average 236U/239Pu ratio of 0.27 ± 0.09, however, is in good agreement with information from the literature, e.g. 0.18–0.33 for an Arctic ice core (Wendel and others, Reference Wendel2013), 0.235 ± 0.014 from global fallout (Sakaguchi and others, Reference Sakaguchi2009) or 0.05–0.50 for Stratospheric fallout (Ketterer and others, Reference Ketterer, Groves and Strick2007). It is interesting to note that this ratio is significantly higher in aquatic systems, e.g. 1–12 in river water samples (Eigl and others, Reference Eigl, Srncik, Steier and Wallner2013) or ~0.7 in the deep Arctic Ocean [236U data in Casacuberta and others (Reference Casacuberta2016), Pu data unpublished]. This nicely reflects the fact that 236U is extremely conservative in water, while Pu is not. Pu strongly adsorbs to surfaces, hence is depleted in water flows. This leads to higher values of the 236U/239Pu ratio.
4. CONCLUSIONS
Concentrations, atom ratios and the total deposition flux during NWT of 239Pu, 240Pu and 236U were obtained for an ice core from eastern Tien Shan, Central Asia. The observed 240Pu/239Pu atom ratios were 0.18 ± 0.02, except for one sample, which represents the global fallout ratio of 0.18. No indication for emission from local sources (Lop Nor) was found. The 236U/239Pu ratios were between 0.12 and 0.43 (average 0.27), in good agreement with values from the literature. Our measured deposition rate for 239Pu of 1.55 × 109 atoms·cm−2 during the NWT period lies within measured values for other sites in the NH though the scatter of reported values is quite large (factor of seven). The higher fallout of 239Pu in eastern Tien Shan glacier may be ideal for defining the Anthropocene. The expected trend of deposition rates with latitude was not observed (lower values at higher latitudes), obviously due to local meteorological conditions that influence deposition rates. The measured 236U deposition rate of 3.5 × 108 atoms·cm−2 agrees within the variability of such measurements with another measurement at an Arctic site (Svalbard, 79.83°N; 24.02°E; 750 m a.s.l.) of 1.6 × 108 atoms·cm−2 (Wendel and others, Reference Wendel2013).
ACKNOWLEDGEMENTS
The effort of the field personnel on the scientific expedition to the Miaoergou glacier in 2005 is highly appreciated. We thank for the support from E. Vogel and S. Szidat during a chemical procedure at the Bern University. The excellent performance of the AMS beam facility at ETH Zürich is highly appreciated. This work was supported by the Natural Science Foundation of China (grant numbers 41330526 and 41711530148), Chinese Academy of Sciences (grant number XDB03030101-4).
Open access
