Precise and accurate isotopic measurements using multiple-collector ICPMS

New techniques of isotopic measurements by a new generation of mass spectrometers equipped with an inductively-coupled-plasma source, a magnetic mass filter, and multiple collection (MC-ICPMS) are quickly developing. These techniques are valuable because of (1) the ability of ICP sources to ionize virtually every element in the periodic table, and (2) the large sample throughout. However, because of the complex trajectories of multiple ion beams produced in the plasma source whether from the same or different elements, the acquisition of precise and accurate isotopic data with this type of instrument still requires a good understanding of instrumental fractionation processes, both mass-dependent and mass-independent. Although physical processes responsible for the instrumental mass bias are still to be understood more fully, we here present a theoretical framework that allows for most of the analytical limitations to high precision and accuracy to be overcome. After a presentation of unifying phenomenological theory for mass-dependent fractionation in mass spectrometers, we show how this theory accounts for the techniques of standard bracketing and of isotopic normalization by a ratio of either the same or a different element, such as the use of Tl to correct mass bias on Pb. Accuracy is discussed with reference to the concept of cup efficiencies. Although these can be simply calibrated by analyzing standards, we derive a straightforward, very general method to calculate accurate isotopic ratios from dynamic measurements. In this study, we successfully applied the dynamic method to Nd and Pb as examples. We confirm that the assumption of identical mass bias for neighboring elements (notably Pb and Tl, and Yb and Lu) is both unnecessary and incorrect. We further discuss the dangers of straightforward standard-sample bracketing when chemical purification of the element to be analyzed is imperfect. Pooling runs to improve precision is acceptable provided the pooled measurements are shown to be part of a single population. Second-order corrections seem to be able to improve the precision on ¹⁴³Nd/¹⁴⁴Nd measurements. Finally, after discussing a number of potential pitfalls, such as the consequence of peak shape, correlations introduced by counting statistics, and the effect of memory on double-spike methods, we describe an optimal strategy for high-precision and accurate measurements by MC-ICPMS, which involves the repetitive calibration of cup efficiencies and rigorous assessment of mass bias combined with standard-sample bracketing. We suggest that, when these simple guidelines are followed, MC-ICPMS is capable of producing isotopic data precise and accurate to better than 15 ppm.