Publications Repository - Helmholtz-Zentrum Dresden-Rossendorf

1 Publication

Accuracy of 9Be-data and its influence on 10Be cosmogenic nuclide data

Merchel, S.; Bremser, W.; Binnie, S.; Bourlès, D. L.; Czeslik, U.; Dunai, T.; Erzinger, J.; Kummer, N.-A.; Leanni, L.; Merkel, B.; Recknagel, S.; Schaefer, U.

The method of choice for the determination of 10Be (t1/2 = 1.378 Ma) is accelerator mass spectrometry (AMS) by measuring 10Be/9Be as low as 10-16. A typical AMS target consists of ~0.5 mg BeO. As most to be radiochemically treated samples are too low in natural Be, a carrier solution containing Be of known concentration is added for sample preparation. The additional amount of 9Be-atoms is taken into account to calculate the number of 10Be-atoms.
Besides, for marine and terrestrial sediments that have absorbed atmospherically-produced 10Be being investigated for dating purposes, and lately also suggested for erosion-rate studies, the determination of 9Be in every individual sample at the ng/g-level is essential.
Thus, for all 10Be-AMS involving 9Be-carrier and/or 9Be measurements, the 10Be-data cannot be more accurate than the 9Be-data of the carrier and/or the samples.
During the recent installation of a new AMS facility [1], special attention has been addressed to the preparation of a 9Be-carrier of low 10Be from Be2SiO4 from a deep mine [2], as earlier studies had shown that commercial solutions contain high amounts of 10Be at the 10-14-level [3]. The 9Be-value is to be determined by a round-robin exercise.
Experimental
The resulting slightly acidic (HCl) Be-solution has been diluted to yield a concentration of ~2000 µg/g checked by three-time repeated gravimetry measurements. Further aliquots of around 1 g each have been sent to seven laboratories experienced with atomic absorption spectrometry (AAS), inductively coupled plasma optical emission spectrometry (ICP-OES) and/or mass spectrometry (ICP-MS). Two labs still owe results; two labs submitted results by two methods each, i.e. ICP-MS & ICP-MS using standard addition (Lab #4) and AAS & ICP-MS (Lab #6). Alls labs reported results as µg/g or µg/per aliquot to exclude density uncertainties.
Results
All hitherto results are shown with their associated stated uncertainty in ascending order in Fig. 1. Lab #1 is of most distant, but the distance is covered by the large uncertainty stated. Thus, no Grubbs outlier at the significance level of a = 0.01 is identified. Lab #2 is not compatible with four other lab results, which is most probably due to an underestimation of its own uncertainty. Despite this, the weighted mean being metrologically the very best estimate, is also shown with the weighted standard deviation (Fig. 1). These values may change when the last two missing values will be taken into account.
There is no clear dependence on the analytical method. However, it is remarkable that the labs using two methods produced very similar results. So, systematic errors due to handling, e.g. further dilution, of the sample might be more influential than the actual measurement accuracy.
Figure 1: Round-robin 9Be-results.
Conclusion
As the maximum deviation from a single lab result from the weighted mean of this round-robin exercise is ~8%, it seems absolutely necessary for all labs using non-commercial 9Be-carrier to have them analyzed at more than a single lab. Otherwise, their 10Be-results are incorrect at the same order as the used 9Be-carrier.
It seems very likely that the same problem arises if measuring individual samples, thus, constant quality assurance checks by e.g. taking part in round-robin exercises are absolutely necessary. The effect might be even more prominent at the ng/g-level.
References
[1] S. Merchel et al., this meeting.
[2] Kindly provided by C. Varajão, Univ. Federal de Ouro Preto.
[3] S. Merchel et al., Nucl. Instr. Meth. B 266 (2008) 4921-4926.

Keywords: AMS; ICP-MS; ICP-OES; AAS; round-robin; quality assurance

  • Poster
    24th Seminar Activation Analysis and Gamma-Spectroscopy (SAAGAS 24), 26.-28.02.2013, Garching, Deutschland

Permalink: https://www.hzdr.de/publications/Publ-17980
Publ.-Id: 17980