Niobium and tantalum concentrations in NIST SRM 610 revisited

Niobium and Ta concentrations in MPI-DING and USGS (BCR-2G, BHVO-2G, BIR-1G) silicate rock glasses and the NIST SRM 610-614 synthetic soda-lime glasses were determined by 193 nm ArF excimer laser ablation and quadrupole ICP-MS. Measured Nb and Ta values of MPI-DING glasses were found to be consistently lower than the recommended values by about 15% and 25%, respectively, if calibration was undertaken using commonly accepted values of NIST SRM 610 given by Pearce et al. Analytical precision, as given by the 1 s relative standard deviation (% RSD) was less than 10% for Nb and Ta at concentrations higher than 0.1 mu g g(-1). A significant negative correlation was found between logarithmic concentration and logarithmic RSD, with correlation coefficients of -0.94 for Nb and -0.96 for Ta. This trend indicates that the analytical precision follows counting statistics and thus most of the measurement uncertainty was analytical in origin and not due to chemical heterogeneities. Large differences between measured and expected Nb and Ta in glasses GOR128-G and GOR132-G are likely to have been caused by the high RSDs associated with their very low concentrations. However, this cannot explain the large differences between measured and expected Nb and Ta in other MPI-DING glasses, since the differences are normally higher than RSD by a factor of 3. Count rates for Nb and Ta, normalised to Ca sensitivity, for the MPI-DING, USGS and NIST SRM 612-614 glasses were used to construct calibration curves for determining NIST SRM 610 concentrations at crater diameters ranging from 16 (im to 60 mu m. The excellent correlation between the Nb/Ca-1 mu gg(-1) signal (Nb represents the Nb signal intensity; Ca-1 mu g g(-1) represents the Ca sensitivity) and Nb concentration, and between the Ta/Ca-1 mu g g(-1) signal (where Ta represents the Ta signal intensity; Ca-1 mu g g(-1) represents the Ca sensitivity) and Ta concentration (R-2 = 0.9992-1.00) in the various glass matrices suggests that matrix-dependent fractionation for Nb, Ta and Ca was insignificant under the given instrumental conditions. The results confirm that calibration reference values of Nb and Ta in NIST SRM 610 given by Pearce et al. are about 16% and 28% lower, respectively. We thus propose a revision of the preferred value for Nb from 419.4 +/- 57.6 mu g g(-1) to 485 +/- 5 mu g g(-1) (1 s) and for Ta from 376.6 +/- 77.6 mu g g(-1) to 482 +/- 4 mu g g(-1) (Is) in NIST SRM 610. Using these revised values for external calibration, most of the determined average values of MPI-DING, USGS and NIST SRM 612-614 reference glasses agree within 3% with the calculated means of reported reference values. Bulk analysis of NIST SRM 610 by standard additions using membrane desolvation ICP-MS gave Nb = 479 +/- 6 mu g g(-1) (1 s) and Ta = 468 +/- 7 mu g g(-1) (1 s), which agree with the above revised values within 3%.