Sm-Nd and U-Pb systematics of single titanite grains

Abstract

Combined Sm-Nd and U-Pb isotopic systems in single titanite grains from Archean rocks may be used to obtain Nd isotope composition at the time of titanite growth. The closed Sm-Nd system requirement is applied to titanite, and its integrity is monitored using the U-Pb system, in a similar way as U-Pb in zircon is used to interpret Lu-Hf data. The Nd isotope composition at the time of titanite formation can be used to infer initial Nd isotopic values of the rocks at the time of zircon crystallization. This approach does not involve assumption that the Sm-Nd system in the rocks remained closed since their formation, whereas such assumption underlies all initial Nd determinations based on whole rock analyses. Titanite is a common accessory mineral in felsic rocks and a major concentrator of REE and uranium. Titanite populations are in many cases heterogeneous, and resolving the isotopic characteristics of old and young titanite populations requires isotopic analysis of multiple single grains. Here I report Sm-Nd and U-Pb data from single titanite grains from four early Archean gneisses previously studied for Lu-Hf and U-Pb systematics of zircon. Gneisses from the Isua area contain heterogeneous populations of titanite, which form roughly linear arrays in the U-Pb concordia diagram, with ca. 3600 Ma upper intercepts, and ca. 2600-2700 Ma lower intercepts, or contain individual concordant grains of the latter age. Two of these gneisses also contain titanite grains of a high-U variety, which plot below these discordia lines, probably as a result of post-Archean loss of radiogenic Pb. Titanite from a gneiss from the Nuuk area is concordant at 2520-2600 Ma. Sm-Nd analyses of several single titanite grains from the Isua area gneiss GGU163263 yielded an isochron age of 3555 ± 150 Ma, identical to the upper intercept U-Pb age of 3601 ± 15 Ma. The εNd(Ttit) values of these grains, calculated using their207Pb/206Pb ages, are uniform for the oldest titanite fractions with Nd content > 1000 ppm, and give the average εNd(Ttit) = + 0.32 ± 0.20 at 3601 Ma. Extrapolation from this value to 3708 ± 25 Ma, an average of the ages of two zircon populations, using an average crustal147Sm/144Nd = 0.12 ± 0.04 yields εNd(Tzir) = + 1.4 ± 0.6 at the time of crystallization. Zircon from the same rock has average εHf(Tzir) = + 1.9 ± 1.0. Titanite grains from two other gneisses from the Isua area show more complex Sm-Nd patterns. Predominant pale-brown to colorless populations have variable147Sm/144Nd ratios reaching as high as 0.7, and do not form isochrons, suggesting that recrystallization of titanite was possibly accompanied by modification of the Sm-Nd system in the whole rock. No reliable initial Nd value has been obtained from these gneisses. The Sm-Nd analyses of titanite fractions from the gneiss from the Nuuk area yielded an average εNd(T) = - 9.00 ± 0.37. Extrapolation to the zircon crystallization age of 3610 ± 10 Ma yields an εNd(T) value of + 1.8 ± 5.4, which has low precision due to extrapolation over a billion-year period. The results presented here show that precise and tenable initial Nd values for early Archean gneisses can be obtained from analyses of best preserved titanite grains in favorable cases, even if these grains represent a small fraction of an otherwise young or altered titanite population.