GSA Annual Meeting, November 5-8, 2001

Paper No. 0
Presentation Time: 8:15 AM

IRON-TITANIUM OXIDE THERMOBAROMETRY: NEW CALIBRATION EXPERIMENTS IN THE FE-TI-O SYSTEM AT 1 BAR AND 1000–1300°C


LATTARD, Dominique, Mineralogisches Institut, Universität Heidelberg, INF 236, Heidelberg, 69120, Germany, dlattard@min.uni-heidelberg.de

Although the current formulations of the FeTi-oxide thermobarometer (Andersen & Lindsley, 1988; Ghiorso & Sack, 1991) are based on sophisticated solid solution models, they repeatly fail to reproduce experimental results, in particular at high oxygen fugacities or high temperatures. Not surprisingly, experimental calibration data are lacking in these conditions.

We have synthesized titanomagnetite-ilmenite (Tmag-Ilm) pairs in the Fe-Ti-O system at 1 bar in the sub-solidus temperature range 1000-1300°C. The oxygen fugacities (in the range FMQ-4 to +5) were fixed either with CO/CO2 mixtures or by using solid oxygen buffers placed together with the sample in evacuated silica glass ampoules. The starting materials were mixtures of TiO2, Fe2O3, ± Fe°. Run durations were > 24 h at 1300°C and up to >120 h at 1000°C. Runs were terminated by drop-quenching in water. The compositions of the coexisting phases in the run products were analyzed by electron microprobe. The attainment of equilibrium is indicated by the compositional homogeneity of the phases within each sample and by the consistency of the results obtained over the whole range of temperature and fO2.

At all temperatures there is a good agreement between the compositions of our synthetic phases and those predicted by the current models in the middle fO2 range. At low fO2 (e.g. < FMQ-1 at 1300°C), however, the models yield too low XTi values, especially for Tmag, because they do not account for cationic vacancies. At high temperatures this leads to large fO2-underestimates. In contrast, at high fO2 (e,g, > FMQ+1 at 1100°C) the formulation of Ghiorso & Sack (1991) significantly overestimates XTi in Ilm, leading to fO2-overestimates of at least 1 log unit. Apparently, the model does not adequately predict XTi in Ilm in the compositional region of the R-3c/R-3 phase transition.