High-precision multidynamic Sr isotope analysis using thermal ionization mass spectrometer (TIMS) with correction of fractionation drift

Abstract

High-precision and high-accuracy measurement of the natural mass-independent variations of 84Sr/86Sr using thermal ionization mass spectrometry (TIMS) is crucial for understanding the planetary volatile depletion history and the heterogeneous isotopic reservoirs in the early Solar System. The recently reported μ84Sr values [defined as 106 × (84Sr/86Srsample/84Sr/86Srstandard – 1)] for terrestrial rock samples, however, show a significant discrepancy between static and multidynamic TIMS measurement results, indicating that the data from one or both measurement methods are biased by analytical artefacts. Coupled with this issue, the accurate μ84Sr values of terrestrial samples are also under debate. Here we present high-precision 84Sr/86Sr and 87Sr/86Sr data for a series of natural samples and the standard NIST SRM 987 measured using a new 3-line multidynamic TIMS method, with the aims to: (1) systematically evaluate and correct the fractionation drift effect in multidynamic Sr isotope measurements, (2) investigate the origin of the discrepancy between the μ84Sr values derived from static and multidynamic measurements, and (3) determine the accurate terrestrial μ84Sr value. Our data show that the rapid drift of isotopic fractionation during TIMS measurement can significantly bias the multidynamic 84Sr/86Sr ratio, but has only minor influence on 87Sr/86Sr. The previously observed static vs. multidynamic μ84Sr discrepancy has been reproduced in our study, and can be explained by the positive biases in multidynamic μ84Sr caused by the high fractionation rates of terrestrial samples compared to the standard. After correcting the fractionation drift effect using a linear interpolation method, the average multidynamic μ84Sr value of the terrestrial samples is in complete agreement with the multistatic result. We confirm that the standard NIST SRM 987 possesses an apparent 84Sr enrichment relative to the natural samples from the Earth, and conclude that the terrestrial samples have an average μ84Sr value of −31 ± 8 ppm (2 standard error). The fractionation drift-corrected multidynamic μ87Sr and μ84Sr results of SRM 987 show a pronounced correlation that is likely to be a combined effect of ion counting random error, uncorrected (or overcorrected) fractionation drift effect, and mixing of Sr ions evaporated from variably fractionated sample reservoirs on the filament. The two independent multidynamic 87Sr/86Sr ratios returned by our method exhibit a stable, small but resolvable offset, which most likely reflects systematic errors generated by an insufficient amplifier idle time setting. Our measurement method has also been applied to three standard materials that have been used in the early Solar System 87Sr/86Sr chronology studies from the 1970s till present to examine the reliability of the normalization method used for inter-laboratory comparison and to assure accurate cross-calibrations of these standards.