Cross calibration of K–Ar standard minerals using an unspiked Ar measurement technique

We have developed an analytical procedure to intercalibrate the 40Ar content of 40Ar/39Ar and 40K/40Ar standard minerals. This procedure is based on an unspiked volumetric standardisation method developed for K–Ar dating. Argon extracted from independent aliquots of minerals are compared to atmospheric argon doses taken repeatedly from a finite reservoir, leading to a progressive depletion of this reservoir through time. Depletion lines relative to each standard, giving Ar amount versus number of doses, are then fitted by a least mean squares regression; confidence intervals are determined on a common basis for all standards by using the jacknife method. The procedure has been applied to four standards, namely MMhb-1, LP-6, HD-B1, and GL-O. For MMhb-1, LP-6 and HD-B1, the results were compared to their most widely used 40Ar∗ per gram published values. Concerning HD-B1, a recent published 40Ar∗ per gram content, called HD-B1new, was also considered. It appears that intercalibrated 40Ar∗ per gram contents of MMhb-1, LP-6 and HD-B1old are self-consistent to within less than 0.06% and fully concordant with the most generally accepted values, HD-B1new is compatible within confidence intervals, but GL-O presents a discrepancy greater than 1%. This lead us to propose to adjust its recommended 40Ar∗ per gram content from (6.68±0.06)·1014 to (6.585±0.016)·1014. Using the published K content of 6.56 wt.% for GL-O, this would shift the age of GL-O from its recommended value (95.0±1.1 Ma) to 93.6±0.9 Ma. Conversely, considering GL-O true would result in shifting the Ar content of MMhb-1, LP-6, HD-B1 by ≈1%. The procedure may be applied to the assessment of confidence intervals of Ar content of standard minerals in a statistically better way than from interlaboratory variance.