Accuracy of stable Mg and Ca isotope data obtained by MC-ICP-MS using the standard addition method

The standard addition method is evaluated to verify the accuracy and precision of Mg and Ca isotope data with complex matrices, using the standard-sample bracketing technique and analysis by MC-ICP-MS. The Ca-44/Ca-42 ratio of seawater (expressed as delta Ca-44/42 relative to SRM915a) was determined as 0 93 +/- 0.093 parts per thousand (95% confidence), in agreement with estimates obtained by the double spike method. Using standard addition, the seawater Mg-26/Mg-24 ratio (expressed as delta Mg-26 relative to the DSM3 standard) was determined as -0.80 +/- 0.06 parts per thousand (95% confidence) in agreement with previous estimates. Four terrestrial silicate rocks (MORB, flood basalt. glacial flour, and granodiorite) and olivine mineral separates from an island basalt are shown to exhibit no scatter within the error of the method, averaging a delta Mg-26 of -0.20 +/- 0.05 parts per thousand (95% confidence). Although a number of silicate rock data for Mg isotope ratios have already been reported this is, the first detailed effort to validate the accuracy of such data and test for residual analytical artifact after chemical purification of samples. Data regressions were evaluated statistically using the mean square weighted deviate (MSWD), demonstrating that the uncertainty on individual data points are generally over estimated. The external two standard deviation uncertainty on individual data points is estimated by Monte Carlo simulation as <0.075 parts per thousand (about a factor of two improvement on early publications of Mg isotope data). The consistency of the standard addition estimates of delta Mg-26 in silicate rocks imply that if any residual matrix effects are present, then they must be less than the spread of the data (0.11 parts per thousand) given the diverse range of matrices in each of the samples. The delta Mg-26 values of the silicate rocks suggest that Mg isotope ratios in silicate material may only have a very restricted range. The delta Mg-26 values of silicate material in the present study falls between the average values reported by Teng et al. [Teng, F.Z., Wadhwa, M., Helz, R.f., 2007. Investigation of magnesium isotope fractionation during basalt differentiation: implications for a chondritic composition of the terrestrial mantle. Earth and Planetary Science Letters 261, 84-92. doi:10.1016/j.epsl.2007.06.004] and Wiechert and Halliday [Wiechert, U., Halliday, A.N., 2006. Non-chondritic magnesium and the origins of the inner terrestrial planers. Earth and Planetary Science Letters 256, 360-371. doi:10.1016/j.epsl.2007.01.007] and given the spread of published delta Mg-26 values for chondritic material, a chondritic composition for terrestrial Mg cannot be ruled out. We suggest that some of the small discrepancies between our data and analysis of the same samples in earlier studies, may have arisen because the chemical purification of Mg prior to analysis can easily induce analytical artifact. This method could be expanded to the isotope ratios of other elements, which also rely on correcting for mass bias using the standard-sample bracketing method, where similar analytical discrepancies may also exist. (C) 2008 Elsevier B.V. All rights reserved.