On the reliability of target element data for cosmogenic nuclide exposure dating

Radioactive or stable cosmogenic nuclides are products of nuclear reactions induced by cosmic rays. The development of the interdisciplinary field of the quantification of cosmogenic nuclides has been increased dramatically in the last decades. Especially, the progress in the field of accelerator mass spectrometry (AMS) improved detection limits and accuracy and allows nowadays the determination of radionuclide concentrations as low as of 10⁴-10⁵ atoms/(g rock).
In particular, in-situ produced cosmogenic nuclides – so-called terrestrial cosmogenic nuclides (TCN) - have proved to be valuable tools for quantifying Earth's surface processes. Here, the work-horses are ¹⁰Be and ²⁶Al in quartz, and ³⁶Cl in Ca- or K-rich minerals.
In siliceous environments both radionuclides, ¹⁰Be and ²⁶Al, are pure high-energy spallation products, thus, the influence on the secondary neutron field by changing trace element concentrations in the original bulk matrix is negligible. Another advantage: Usually, quartz can be easily cleaned from other mineral phases making the normalization to (g SiO₂)^-1, i.e. only two target elements, very straightforward. There is usually no need for a full chemical analysis.

In contrast, ³⁶Cl can be produced by several different nuclear reactions: Spallation on different target elements, mainly Ca and K (to a lesser extent Ti and Fe), induced by high-energy neutrons and muons. Thus, every sample that will be analysed for TCNs has to be chemically analyzed for the main target elements, too. Additionally, as ³⁶Cl can be produced by thermal neutron-capture on ³⁵Cl, trace elements influencing the thermal neutron field have to be unavoidably also measured. This includes U and Th as “background”-neutron emitter, all elements with high neutron absorption cross sections like B, Gd, and Sm, and all light elements, which take part in (gamma,n)-reactions. Thus, a complete bulk rock analysis (for the neutron field) and measurements of the main target elements in the dissolved fraction are absolutely necessary. As ³⁶Cl concentrations will be normalized to (g Ca)^-1, (g K)^-1 etc., the overall result cannot be more precise then the corresponding target element data.

We have selected typical samples with different fractions of silicate- and calcite-rich phases to have them analyzed by the method of choice of most TCN-user: inductively coupled plasma optical emission spectrometry (ICP-OES). Unfortunately, the two French CNRS laboratories involved, CEREGE and CRPG-SARM, did not produce concordant results for all elements of interest. Thus, we have tested the capability of two non-destructive activation analysis methods: We performed Instrumental Neutron Activation Analysis (INAA) ourselves at Vienna and we sent aliquots for Prompt Gamma Activation Analysis (PGAA) to Budapest.

The four laboratories performed differently depending on the analyte. For example, all methods produce CaO-data with unacceptable high uncertainties. PGAA seems to underestimate the true CaO-value of some samples. K₂O-data (@ 0.1-0.7%) by ICP-OES has exceptional high uncertainties and is constantly lower as corresponding INAA- and PGAA-data. As a conclusion, it seems advisable to use more than a single analytical method, if precise TCN applications are intended.