Radiokrypton dating finally takes off

Methods of age determination belong to the fundamental toolkit of modern Earth and environmental sciences, as well as archeology. Radiometric dating, based on the well-known radioactive decay of certain isotopes, is the gold standard among the dating methods, with radiocarbon (¹⁴C) as the most famous example. However, many more radioisotopes are necessary to cover the wide range of dating applications. Among them, ⁸¹Kr, or radiokrypton, has long been recognized as a desirable tool, especially for the dating of old groundwaters and ancient polar ice, but this goal has remained elusive. In PNAS, Buizert et al. (1) present, to my knowledge, the first successful ⁸¹Kr dating of polar ice. This breakthrough, along with two recent applications of ⁸¹Kr in groundwater (2) and thermal fluids (3), signals to me that the dream of radiokrypton dating has finally become reality.

The contribution of Buizert et al. (1) is important because it addresses the prominent climate archive made up of the polar ice sheets. More than 3-km-long vertical drill cores from Greenland and Antarctica have provided a wealth of information on past temperatures and greenhouse gas concentrations up to 800 ka (4). In addition to the ice cores, which are limited in number and amount of ice, old ice is accessible in abundance at locations near ice sheet margins and in so-called blue ice areas, where old ice upwells, whereas accumulation of snow is prevented. Some of the blue ice areas of Antarctica may contain the much-sought “oldest ice” on Earth, expected to be about 1.5 Ma (5).

So-called horizontal ice cores from blue ice areas lack the obvious stratigraphy of vertical ice cores, which makes the availability of an independent dating tool even more important. However, the common radiometric methods largely fail when it comes to the dating of polar ice. Radiocarbon is not only limited to a dating range of merely 50 ka, but it is also afflicted by in situ production of ¹⁴C in ice. Other tools are limited to horizons with datable material. In contrast, ⁸¹Kr is present in air inclusions throughout the ice and its half-life of 229 ka perfectly fits to the expected age range of Antarctic ice. Furthermore, as a noble gas isotope, ⁸¹Kr has nearly ideal tracer properties: It is not involved in complicating geochemical reactions and it essentially has a single, well-mixed, and steady source, the atmosphere.

Buizert et al. (1) did not attempt to find extremely old ice, but they could demonstrate that ice from the Eemian interglacial (about 120 ka) is present at their study site, the Taylor Glacier in Antarctica. The

The contribution of Buizert et al. is important because it addresses the prominent climate archive made up of the polar ice sheets.

independent and absolute ages derived from ⁸¹Kr anchor available age models based on stratigraphic matching and verify that scientifically valuable ice from the Eemian and the preceding penultimate deglaciation can be sampled in large quantities at this site. Furthermore, the presence of such old ice puts constraints on the ice flow dynamics of Taylor Glacier.

As important as these glaciological findings are, the main reason for excitement about the study of Buizert et al. (1) is that it finally fulfills an old vision. Already in 1970, Loosli et al. (6), the Swiss pioneers in the detection of noble gas radionuclides, wrote “The dating of ice in coastal areas of Antarctica is an important problem that could be studied by using ⁸¹Kr.” Why then did it take more than 40 y to achieve this goal? The answer lies in the next sentence of Loosli et al. (6): “To obtain 50 cm³ of krypton, however, the present minimum volume required, 10³ tons of ice would be required.”

The only available analytical method at the time, low-level counting (LLC) of ⁸¹Kr decays, needs huge amounts of sample material to extract a sufficient number of the rare and slowly decaying ⁸¹Kr atoms. Similar restrictions plague the use of the other long-lived noble gas radioisotope, ³⁹Ar, which is also of great interest for many dating applications (see Table 1 for relevant data on these isotopes). For practical purposes, ⁸¹Kr has remained beyond the reach of LLC, whereas ³⁹Ar has been measured at the University of Bern ever since Loosli's pioneering work, even if that laboratory still remains the only one worldwide capable of performing this task.

Table 1.

Properties and sample requirements of the long-lived noble gas radioisotopes

Property	39Ar	81Kr
Half-life (y)	269	229,000
Decay constant (s−1)	8.17 × 10−11	9.59 × 10−14
Isotope concentration in air (39Ar/Ar, 81Kr/Kr)	8.1 × 10−16	5.2 × 10−13
Concentration in modern water* (atoms/L)	7,500	1,100
Concentration in modern ice† (atoms/kg)	18,000	1,400
Minimum sample size for ATTA‡ (mL Ar or Kr)	500	0.005
Corresponding number of atoms in modern sample	1 × 107	70,000

^*Calculated for air-equilibrated water at 15 °C.

^†Calculated assuming an air content of 90 cm³STP/kg.

^‡Current status for Kr (5, 13) and Ar (19). If contamination issues can be solved, the requirement for ³⁹Ar is expected to drop to roughly 1 mL of Ar or 20,000 ³⁹Ar atoms (20).

As for all long-lived radioisotopes, it is more efficient to count ⁸¹Kr or ³⁹Ar atoms themselves rather than their rare decays. However, this faces the obstacle of separating the desired isotopes from the much more abundant stable isotopes of the respective element. Modern air has an ⁸¹Kr/Kr ratio of only 5 × 10⁻¹³ and an even lower ³⁹Ar/Ar ratio of 8 × 10⁻¹⁶ (Table 1). Conventional MS is unable to distinguish such rare isotopes from the overwhelming neighboring masses, and even resonance ionization MS (RIMS) requires isotope enrichment to lift the ⁸¹Kr/Kr ratio to a measurable level (7). The first attempt to date Antarctic ice with ⁸¹Kr (8) was based on RIMS, but the method never delivered fully convincing results for noble gas radioisotopes.

The only MS technique capable of the exquisite isotope selectivity required for rare radioisotopes is accelerator MS (AMS), which has revolutionized the applicability of the radiocarbon method in the last decades. However, traditional tandem accelerators cannot be used for noble gases, which do not form stable negative ions (9). In principle, this restriction can be overcome by using cyclotron accelerators, as demonstrated 30 y after the pioneering work of Loosli with the first successful ⁸¹Kr dating of groundwater (10). However, this extremely expensive technique is hardly conceivable as a routine method for dating applications.

Also around the turn of the millenium, a new analytical technique called atom trap trace analysis (ATTA) was developed at the Argonne National Laboratory (11). Five years later, a first successful application of ATTA for ⁸¹Kr dating of groundwater with ages up to 1 Ma was published (12). However, still another 10 y of further development were needed to improve the reliability and sensitivity of this analytical technique until it now proved capable of performing the task of ⁸¹Kr dating of ice. Meanwhile, ATTA can be regarded as firmly established for ⁸¹Kr, and the laboratory at Argonne is performing these analyses on a routine basis (13).

ATTA is based on the techniques of laser cooling and trapping that have been very prolific in atomic physics during the last decades. Atoms are slowed down and trapped by interaction with laser light, precisely tuned to the energy of a selected electronic transition of one specific isotope. The captured isotopes spend roughly a second in the magneto optical trap, enabling counting of single atoms via the fluorescence induced by the trapping laser beams (14). The crucial point is that many thousands of photons have to be absorbed by an atom to be trapped and detected, which is only possible for the chosen isotope that is close to resonance with the laser light. For this reason, ATTA is the only trace analysis technique that is completely free of interferences with any other species (5). The difficulty of ATTA is not to sort out single ⁸¹Kr atoms from the background of 10¹³ times more abundant stable Kr isotopes, but to obtain sufficiently high count rates.

ATTA is not applicable to all elements, but fortunately it works well for noble gases and can thus cover a blind spot of AMS. In my view, ATTA has the potential to become for noble gas radioisotopes what AMS is for radiocarbon today: a very powerful and widely used analytical technique. However, ATTA needs further improvements to fulfill this promise. The ice samples from the Taylor glacier weighed about 350 kg (1), which is far above any reasonable limit for ice cores. However, there is little doubt that the efficiency of the analytical method can be further increased (5, 14) to reduce sample size requirements and improve the analytical precision. This development will decide the fate of ⁸¹Kr as a tool of future ice core research. For applications of ⁸¹Kr dating in groundwater, sample size is less of a restriction, and several studies in large aquifer systems are currently in progress (5).

ATTA can also be used to detect other radioisotopes, such as ⁴¹Ca (15) and in particular ³⁹Ar (16, 17), which has the same advantages as ⁸¹Kr but covers a different age range with its shorter half-life of 269 y (Table 1). The dating range of ³⁹Ar fills a gap between various methods available for samples younger than about 50 y and ¹⁴C dating, which works best for ages above 1,000 y. Because this age range fits perfectly to the time scale of deep ocean circulation, ³⁹Ar is the ideal dating tracer for physical oceanography. In the past, large-volume ocean samples have been collected and analyzed for ³⁹Ar by LLC (18). However, regular use of ³⁹Ar in oceanography is only feasible with sample volumes of 10 L or smaller (5).

The reason that ATTA analysis is more challenging for ³⁹Ar than for ⁸¹Kr lies in the three orders of magnitude lower isotopic abundance (Table 1), requiring an even higher efficiency of the setup to achieve a reasonable count rate. At Heidelberg University, the Oberthaler group has succeeded in optimizing ATTA for ³⁹Ar, and in collaboration with my group, we recently could date the first few groundwater samples by ³⁹Ar-ATTA (19, 20). At present, sample sizes on the order of 1,000 L of water are required for such an analysis, but in principle, a reduction to only a few liters should be possible (20). ³⁹Ar-ATTA in Heidelberg has not yet reached the routine applicability of ⁸¹Kr-ATTA at Argonne, but there are no fundamental reasons to prevent this from happening. Assuming that sample size requirements can be reduced as expected, ³⁹Ar dating is poised to expand its scope in a similar way as we now witness for ⁸¹Kr.