Influence of non-spectral matrix effects on the accuracy of Pb isotope ratio measurement by MC-ICP-MS: implications for the external normalization method of instrumental mass bias correction

Ca, Al, Fe and Mg are identified as the dominant inorganic matrix components in Pb purified by anion exchange chromatography from a wide compositional range of silicate samples. Doping experiments demonstrate that these elements cause non-spectral matrix effects that result in inaccuracy of mass bias-corrected MC-ICP-MS Pb isotope ratios ranging from -150 to +120 ppm per amu. Although small, these inaccuracies are significant at the level of precision required for application in mantle geochemistry today. The inaccuracy results from matrix induced per mil level variation in the Tl to Pb mass bias ratio (df(Tl)/f(Pb)parts per thousand). In matrix-free standards, df(Tl)/f(Pb)parts per thousand varies from -0.30 to +0.21 parts per thousand, depending on run conditions, whereas the mean df(Tl)/f(Pb) in matrix experiments varies from -1.45 parts per thousand with an Fe matrix to 7.11 parts per thousand with a Ca matrix. The matrix also causes increased Pb and Tl sensitivity. Signal enhancement plateaus out at high matrix levels. This finite response to the matrix can be used as an advantage. With a common matrix of Mg added to both doped Pb(Tl) standards and their bracketing un-doped standards, variation in f(Tl)/f(Pb) is greatly reduced, resulting in external reproducibility of combined doped and un-doped Pb isotope ratios (30-61 ppm per amu) similar to single session external reproducibility of pure standards (20-60 ppm per amu). Addition of a common matrix to samples and standards may provide a viable alternative to thorough purification in the effort to reduce or eliminate non-spectral matrix effects in MC-IPC-MS isotope ratio measurements.