An assessment of mass discrimination in MC-ICPMS using Nd isotopes

The stability of mass discrimination in multiple-collector magnetic-sector inductively coupled mass spectrometry (MC-ICPMS)-with time, between elements and between samples-makes it potentially much simpler to deal with than in TIMS. However, while the stability of mass bias across limited areas of the mass spectrum is critical to the derivation of precise isotope ratios, the fundamentals of mass bias behaviour of MC-ICPMS instruments are still incompletely characterised. In this paper, we present Nd isotope data for standards and samples with the aim of using the well-known. Nd isotopic system to obtain systematic information on the nature of the mass bias in MC-ICPMS. An extensive Nd dataset was obtained on a Micromass IsoProbe and more limited data on two different Nu Instrument machines. The IsoProbe data were obtained over 18 months between March 2000 and September 2001. The standard approach of using Nd-146/Nd-144 to normalise other Nd isotope ratios leads to both inaccurate (by around 100 ppm in the case of the exponential-law normalised Nd-143/Nd-144 ratio) and relatively imprecise (2 rsd=45 ppm for Nd-143/Nd-144) results. On the IsoProbe, this is due to the fact that the magnitude of the exponential mass bias itself varies, albeit by a small amount, over limited mass ranges such as that for Nd. The result is that the inaccuracy is much greater for isotope ratios that have an average mass further away from that of the normalising ratio-for example, > 500 ppm for Nd-150/Nd-144 vs. < 30 ppm for Nd-145/Nd-144. Both accuracy and precision increase dramatically if a normalising ratio is used that is close in mass to the ratio to be normalised. Nd-143/Nd-144 (average mass = 143.5) normalisation with Nd-145/Nd-142 (average mass = 143.5) yields a value identical to TIMS. An alternative approach is to use post-normalisation linear correlations between isotope ratios to do a secondary mass bias correction. Such an approach with the Nd-143/(144) Nd ratio using Nd-142/(144) Nd yields a value identical to the TIMS value and a long-term reproducibility of 14-20 ppm. This compares with a reproducibility of 45 ppm using simple normalisation to Nd-146/Nd-144. We have tested both these approaches on standards and samples with Nd-143/Nd-144 up to 30 epsilon units different from our in-house standard and identical results to TIMS are obtained. Post-normalisation correlations between isotope ratios obtained on the Nu Instruments MC-ICPMS are qualitatively very similar to those obtained on the IsoProbe and suggest a common cause. This, despite the very different physical characteristics of the various instruments. Furthermore, it also appears that qualitatively very similar effects, though at much smaller magnitude, are observed in TIMS. The data suggest that the quasi-empirical exponential law does not perfectly correct for mass discrimination on any mass spectrometer. This inadequacy becomes important, for precise isotope ratio analysis, when dealing with the large mass discriminations inherent in MC-ICPMS. (C) 2002 Elsevier Science B.V. All rights reserved.