Measuring 0.01‰ to 0.1‰ isotopic variations by MC-ICPMS - Testing limits for the first time with Pb δ-iCRMs

A blind comparison on Pb-isotope δ-scale measurements by MC-ICPMS of 0.01‰ to 0.1‰ level was organised, involving five laboratories. Test samples were obtained from the series of candidate ERM-3810 δ-isotopic Certified Reference Materials (δ-iCRMs), and comprise four pairs of a material with ∼natural Pb-isotopic composition ('delta zero' or 'δ-0') and the same natural Pb progressively enriched in 207Pb (with δ207Pb values certified to ∼0.1% relative uncertainty, k = 2). Participants were free to apply the measurement strategy of their choice. A result was considered 'acceptable' only when, simultaneously, there was agreement within stated uncertainties with the corresponding reference value and the relative uncertainty stated by the participant was < 100%. This study illustrates the high degree of difficulty inherent to these δ-scale measurements by 'routine' MC-ICPMS methodologies (in this case, three participants reported 55% of their results which were deemed accurate, and the other two reported none). The closer to unity the isotope ratio value the better the results became ('acceptable' results mostly for δ7/6‰ and δ7/8‰ measurements). This first experiment of its kind demonstrates that Pb δ-scale isotopic measurements by MC-ICPMS can be reliably carried out down to 0.05‰ levels (two participants delivered accurate results above this threshold systematically for δ7/6‰, δ7/8‰ and δ7/4‰). Below this limit, at ∼0.01‰ and ∼ 0.03‰ levels, results are no longer consistent or reproducible and appear to be susceptible to a number of effects introducing error (such as short term changes in mass discrimination) which are either not well understood, or not controlled and/or not corrected for at a sufficiently low level of uncertainty. These results also suggest that 'routine' methods for absolute (calibrated) Pb-isotope ratio determination by MC-ICPMS produce relative combined uncertainties on results which are unlikely to be better than 0.05‰ (k = 2). © The Royal Society of Chemistry 2009.