A method has been developed for the precise and reproducible measurement of Ti isotopes in natural materials using high resolution MC-ICPMS. Instrumental mass fractionation is internally corrected using 49Ti/47Ti. Replicate measurements of synthetic standard solutions, terrestrial rocks, and the carbonaceous chondrite Allende yield a long-term reproducibility (2σ) of 0.28ε, 0.34ε, and 0.28ε for 50Ti/47Ti, 48Ti/47Ti, and 46Ti/47Ti, respectively. Isobaric interferences from 46,48Ca+, 50V+, 50Cr+, and doubly charged Zr can be corrected for reliably in separated Ti solutions with Ca/Ti 14N36Ar+, or do not compromise the measurements. The method has been successfully applied to terrestrial rocks, meteorites, and various mineral separates. Terrestrial samples and standards agree within analytical uncertainties but are consistently different from the recommended values of Niederer et al. [F.R. Niederer, D.A. Papanastassiou, G.J. Wasserburg, Geochim. Cosmochim. Acta 45 (1981) 1017] with the largest effect on 50Ti/47Ti. The new results provide evidence that the recommended terrestrial 50Ti/47Ti is not well constrained; our data are higher by ∼13ε than the recommended value. Better agreement is found with the values recommended by Heydegger et al. [H.R. Heydegger, J.J. Foster, W. Compston, Earth Planet. Sci. Lett. 58 (1982) 406]. Our best estimate for the isotopic composition of terrestrial Ti (relative to 49Ti/47Ti = 0.749766) is: 50Ti/47Ti = 0.73010, 48Ti/47Ti = 10.06884, and 46Ti/47Ti = 1.09325. This corresponds to an atomic weight of 47.877, significantly different from the value of 47.867 recommended by IUPAC. A 50Ti/47Ti anomaly for bulk Allende of 3.37 ± 0.51ε is found, while for 48Ti/47Ti and 46Ti/47Ti the data are identical to the terrestrial value within the uncertainties. This is consistent with other recently reported differences between the bulk Earth and chondrites in neutron-rich nuclides. © 2006 Elsevier B.V. All rights reserved.
