The new generation of multi-collector mass spectrometers (e.g. ARGUSVI) permit ultra-high precision (<0.1%) 40Ar/39Ar geochronology of rocks and minerals. At the same time, the 40Ar/39Ar method is limited by relatively large uncertainties (>1%) in K-40 decay constants and the ages of natural reference minerals that form the basis of the technique. For example, reported ages for widely used 40Ar/39Ar reference materials, such as the ca. 28 Ma Fish Canyon Tuff sanidine (FCTs) and the ca. 1.2 Ma Alder Creek Rhyolite sanidine (ACRs), vary by >1%. Recent attempts to independently calibrate these reference minerals have focused on K-Ar analyses of the same minerals and inter-comparisons with astronomically tuned tephras in sedimentary sequences and U-Pb zircon ages from volcanic rocks. Most of these studies used older generation (effectively single-collector) mass spectrometers that employed peak-jumping analytical methods to acquire 40Ar/39Ar data. In this study, we reassess the inter-calibration and ages of commonly used 40Ar/39Ar reference minerals Fish Canyon Tuff sanidine (FCTs), Alder Creek Rhyolite sanidine (ACRs) and Mount Dromedary biotite (MD2b; equivalent to GA-1550 biotite), relative to the astronomically tuned age of A1 Tephra sanidine (A1Ts), Faneromeni section, Crete (Rivera et al., 2011), using a multi-collector ARGUSVI mass spectrometer. These analyses confirm the exceptional precision capability (<0.1%) of this system, compared to most previous studies. All sanidine samples (FCTs, ACRs and A1Ts) exhibit discordant 40Ar/39Ar step-heating spectra, with generally monotonically increasing ages (similar to 1% gradients). The similarity in these patterns, mass-dependent fractionation modeling, and results from step-crushing experiments on FCTs, which yield younger apparent ages, suggest that the discordance may be due to a combination of recoil loss and redistribution of 39ArK and isotope mass fractionation. In contrast to our previous inferences, these results imply that the sanidine samplesAre suitable 40Ar/39Ar reference materials, provided appropriate corrections Are included for differential recoil loss of 39ArK and contributions from xenocrysts/antecrysts can be resolved. Relative to an age of 6.943 +/- 0.005 Ma for A1Ts, we calculate astronomically tuned ages for FCTs, ACRs and MD2b of 28.126 +/- 0.019 (0.066%) Ma, 1.18144 +/- 0.00068 (0.058%) Ma and 99.125 +/- 0.076 (0.077%) Ma, respectively (95% internal errors). These results are consistent with recent U-238/Pb-206 age data from these localities, but are marginally younger (similar to 0.2%) than previous 40Ar/39Ar ages inter-calibrated with astronomically tuned tephra from the Mediterranean, and distinctly younger (0.6%) than results optimized against a broad array of U-238/(206) Pb zircon ages. Consideration of published and assumed recoil loss 39ArK proportions (0.18-0.40%), yields recoil-corrected age estimates of 28.187 +/- 0.019 Ma, 1.18404 +/- 0.00068 Ma and 99.204 +/- 0.076 Ma, respectively. This comparison indicates inherent uncertainties of >0.1% in the 40Ar/39Ar ages of reference minerals without consideration of recoil artefacts, thus limiting the benefits of high precision multi-collector analyses. Significant improvement to the accuracy of the 40Ar/39Ar method (<0.1%) will require further inter-laboratory 40Ar/39Ar studies utilizing multi-collector mass spectrometry, additional constraints on recoil 39ArK loss from reference minerals, further resolution of discrepancies between astronomically tuned sedimentary successions and refinement of the U-238/Pb-206 zircon age cross-calibration approach. (C) 2016 Elsevier Ltd. All rights reserved.
