Calibration of cosmogenic noble gas production based on 36 Cl‐ 36 Ar ages. Part 2. The 81 Kr‐Kr dating technique

We calibrated the 81 Kr‐Kr dating system for ordinary chondrites of different sizes using independent shielding‐corrected 36 Cl‐ 36 Ar ages. Krypton concentrations and isotopic compositions were measured in bulk samples from 14 ordinary chondrites of high petrologic type and the cosmogenic Kr component was obtained by subtracting trapped Kr from phase Q . The thus‐determined average cosmogenic 78 Kr/ 83 Kr, 80 Kr/ 83 Kr, 82 Kr/ 83 Kr, and 84 Kr/ 83 Kr ratiC(Lavielle and Marti 1988; Wieler 2002). The cosmogenic 78 Kr/ 83 Kr ratio is correlated with the cosmogenic 22 Ne/ 21 Ne ratio, confirming that 78 Kr/ 83 Kr is a reliable shielding indicator. Previously, 81 Kr‐Kr ages have been determined by assuming the cosmogenic production rate of 81 Kr, P( 81 Kr) c , to be 0.95 times the average of the cosmogenic production rates of 80 Kr and 82 Kr; the factor Y = 0.95 therefore accounts for the unequal production of the various Kr isotopes (Marti 1967a). However, Y should be regarded as an empirical adjustment. For samples whose 80 Kr and 82 Kr concentrations may be affected by neutron‐capture reactions, the shielding‐dependent cosmogenic ( 78 Kr/ 83 Kr) c ratio has been used instead to calculate P( 81 Kr)/P( 83 Kr), as for some lunar samples, this ratio has been shown to linearly increase with ( 78 Kr/ 83 Kr) c (Marti and Lugmair 1971). However, the 81 Kr‐Kr ages of our samples calculated with these methods are on average ~30% higher than their 36 Cl‐ 36 Ar ages, indicating that most if not all the 81 Kr‐Kr ages determined so far are significantly too high. We therefore re‐evaluated both methods to determine P( 81 Kr) c /P( 83 Kr) c . Our new Y value of 0.70 ± 0.04 is more than 25% lower than the value of 0.95 used so far. Furthermore, together with literature data, our data indicate that for chondrites, P( 81 Kr) c /P( 83 Kr) c is rather constant at 0.43 ± 0.02, at least for the shielding range covered by our samples ([ 78 Kr/ 83 Kr] c = 0.119–0.185; [ 22 Ne/ 21 Ne] c = 1.083–1.144), in contrast to the observations on lunar samples. As expected considering the method used, 81 Kr‐Kr ages calculated either directly with this new P( 81 Kr) c /P( 83 Kr) c value or with our new Y value both agree with the corresponding 36 Cl‐ 36 Ar ages. However, the average deviation of 2% indicates the accuracy of both new 81 Kr‐Kr dating methods and the precision of the new dating systems of ~10% is demonstrated by the low scatter in the data. Consequently, this study indicates