High-precision Mg isotope analyses of low-Mg rocks by MC-ICP-MS

We present a method for precise measurement of Mg isotope ratios for low-Mg rock samples (where MgO <1 wt.%) by multi-collector inductively coupled plasma-mass spectrometry (MC-ICP-MS). The efficiency of Mg purification is significantly improved by using a newly calibrated HNO3 + HF step to remove undesired matrix elements (such as Ti, Al, Fe, and K) in low-Mg samples. We also establish that increasing the amount of Mg loaded to the chromatographic column minimized blank effects of organics leached from cation resin. All parameters that could affect the accuracy and precision of Mg isotope analyses were rigorously examined by two independent laboratories in Beijing and Hefei. The delta Mg-26 of mono-elemental Mg standard CAM-1 measured in the two laboratories were -2.597 +/- 0.042 parts per thousand (2 sigma, n = 49) and -2.598 +/- 0.039 parts per thousand (2 sigma, n = 79), respectively; in house standard IGGMg1 were -1.742 +/- 0.041 parts per thousand (2 sigma, n = 53) and -1.749 +/- 0.049 parts per thousand (2 sigma, n = 72), respectively. The average delta Mg-26 over ten months of two synthetic standards, made by doping IGGMg1 and IGGMg2 with matrix elements, agrees well with their recommended values, within error. The robustness of our method was further assessed by replicated analyses of sixteen rock standards with MgO contents from 0.28 wt.% to 49.4 wt.%. The delta Mg-26 of USGS rhyolite standards RGM-1 and RGM-2 are -0.188 +/- 0.031 parts per thousand (2 sigma, n = 35) and -0.182 +/- 0.041 parts per thousand (2 sigma, n = 72), respectively; granite standard GA is -0.165 +/- 0.038 parts per thousand (2 sigma, n = 57), G-2 is -0.129 +/- 0.045 parts per thousand (2 sigma, n = 34), GS-N is -0.204 +/- 0.059 parts per thousand (2 sigma, n = 33), GSP-2 is 0.042 +/- 0.020 parts per thousand (2 sigma, n = 15), and GSR-1 is -0.234 +/- 0.016 parts per thousand (2 sigma, n = 17). Based on repeated analyses of standards, the long-term external precision of our method is better than +/- 0.05 parts per thousand for delta Mg-26. This precision allows us to distinguish the fractionation of Mg isotopes in low-Mg granites and rhyolites as well as that between mantle minerals. (C) 2014 Elsevier B.V. All rights reserved.