Empirical and experimental constraints on Fe-Ti oxide-melt titanium isotope fractionation factors

• Published in: Geochimica et cosmochimica acta 2022-06, Vol.326, p.253-272
• Main Authors: Hoare, Liam, Klaver, Martijn, Muir, Duncan D., Klemme, Stephan, Barling, Jane, Parkinson, Ian J., Lissenberg, C. Johan, Millet, Marc-Alban
• Format: Article
• Language: English
• Subjects: Fractionation factors; Ilmenite; Isotope fractionation; Magmatic differentiation; Mantle melting; Micro-mill; Rutile; Titanium isotopes; Titanomagnetite
• ISSN: 0016-7037; 1872-9533
• DOI: 10.1016/j.gca.2022.02.011

The Titanium (Ti) isotope compositions of felsic rocks are heavier than their mafic counterparts, and alkaline magmas develop heavier Ti isotope compositions compared to other magma series during magmatic differentiation. Both observations are interpreted to reflect the preferential sequestration of light Ti isotopes in Fe-Ti oxides, such as rutile, ilmenite and titanomagnetite. However, such interpretations so far rely on whole rock studies of cogenetic magmatic samples and the detailed mechanics of oxide-melt equilibrium on the Ti isotope composition of magmas is poorly constrained. To address this, we have measured the Ti isotope composition of co-existing Fe-Ti oxides and groundmass or silicate melt in both natural lavas from contrasting tectonic settings (Heard Island and Santorini), and experimental run products (rutile-melt). All Fe-Ti oxide phases are consistently isotopically lighter than their respective host groundmass or silicate melt, with the magnitude of Δ49/47Tioxide-melt increasing from rutile to ilmenite, and titanomagnetite. The observed difference in Ti isotope fractionation between rutile-melt experiments and ilmenite-groundmass pairs is primarily reflective of small differences in their TiO bond length, with ilmenite being isotopically lighter (Δ49/47Tiilmenite-melt extrapolated to 1000 K = −0.600 ± 0.035‰) compared to rutile (Δ49/47Tirutile-melt at extrapolated to 1000 K = −0.404 ± 0.099‰) due to slightly longer TiO bonds in ilmenite. In contrast, the variation in Δ49/47Ti observed between titanomagnetite-groundmass pairs increases as a function of increasing TiO2 content (increasing ulvöspinel component) in titanomagnetite, with Δ49/47Tititanomagnetite-melt values extrapolated to 1000 K ranging from −0.811 to −1.451‰ in Ti-rich titanomagnetite (21–23 wt.% TiO2; Usp66-73) from Heard Island compared to −0.673‰ to −0.863 in Santorini (14–15 wt.% TiO2; Usp45-49). We interpret this to result from a weaker and distorted crystal lattice due to changes in the local cationic environment resulting from exchange of smaller Fe3+ ions with larger Fe2+ ions during magnetite-ulvöspinel solid solution. Our results are consistent with fractionation factors inferred from Ti isotopic analyses of mineral separates and ab-initio calculations. We use these fractionation factors in combination with Ti isotope fractionation factors for silicate minerals recalculated from previous studies, along with petrologic information to model the behaviour of Ti isoto