Multidynamic isotope ratio analysis using MC?ICP?MS and the causes of secular drift in Hf, Nd and Pb isotope ratios

Abstract

The causes of secular drift in multicollector (MC)–ICP–MS normalized isotope ratios were investigated using multidynamic measurements of Nd, Yb, Hf and Pb isotope ratios over 2.5 year on an IsoProbe MC–ICP–MS, and of Nd and Yb by thermal ionisation mass spectrometry (TIMS). We present the first TIMS multidynamic determinations of all nonradiogenic Nd isotope ratios, which differ significantly from previously assumed reference values. Multidynamic isotope ratios measured on the IsoProbe show much less secular drift than static ratios calculated from the same mass spectrometer runs, and for example improve Hf external precision to ∼60 to 75 ppm 2S.D. for all ratios. The poor performance of static measurements cannot easily be attributed to Faraday cup deterioration because static data show cyclic and sometimes rapid secular drift. Multidynamic Nd and Yb ratios, normalized using exponential law, show systematically greater deviation from TIMS ratios the greater the difference in average mass between normalized and normalising ratios. Nearly accurate isotope ratios for Yb and for two of three sets of Nd data can be calculated using a general power law (GPL) with exponent of −0.63. This implies that our multidynamic 176Hf/177Hf and 180Hf/177Hf ratios, and those reported by most other laboratories also using normalisation by exponential law to 179Hf/177Hf=0.7325, are lower than the true isotope ratios. General power law normalisation does not reduce secular variation in multidynamic ratios. Correlations between ratios normalized by exponential or general power law can only partly be explained by variation in the general power law exponent from day to day. Exponential-law normalized 143Nd/144Nd, 142Nd/144Nd and 145Nd/144Nd lie within error of a plane in three dimensions: empirical correction of 143Nd/144Nd using this plane yields 25 ppm external precision (2S.D.) for all multidynamic 143Nd/144Nd over 2.5 years and results in a corrected ratio identical to TIMS. Linear regressions for Hf isotope ratios yield 50 ppm 2S.D. on 176Hf/177Hf over the same period. Algorithms are presented for static and multidynamic analysis of spiked Nd and Hf samples and we show that these yield the same nonradiogenic ratios as standard samples despite a wide range of matrix types.