A solution model for coexisting iron–titanium oxides

Abstract

A solution model has been developed for coexisting magnetite-ulvcispinel and hematiteilmenite solid solutions and applied to the Buddington-Lindsley (1964) geothermometer and oxygen barometer. The model is based on results from the hydrothermal experiments of Lindsley (550-1000C), gas-mixing experiments of Katsura et al. (1976) and Webster and Bright (1961) (1000-1200'C), and new hydrothermal experiments performed by Spencer and Lindsley (1978) using the Co-CoO bufer. The model assumes (1) behaves as a binary asymmetric Margules solution; (2) titanomagnetite behaves as a binary asymmetric Margules solution below Sfi)C and as an ideal binary solution above 800C; (3) configurational entropy terms can be approximated by a molecular mixing model for magnetites, and by Rumble's (1977) model B (disorder of Fe3+) for (R3) ilmenites; (4) only ordered (R3) ilmenite solutions are present. The free energy of the exchange reaction Fe3O4 + FeTiO3 : Fe2O3 * Fe2TiOa and the excess parameters for each solution were solved by least-squares fit of the experimental data. The model has successfully reproduced experimental data in the temperature-fo, range 600-1300'C, bounded (approximately) by the nickel-nickel oxide and wtistitemagnetite buffer curves. The model predicts a consolute point for Mt-Usp of -550C at -Usp 32 68. No attempt was made to estimate mathematically the Hem-Ilm two-phase field. Uncertainties in T andf6, are approximately 40-80C and 0.5-1.0 log units/e, (2o) assuming -rlVo uncerlainties in Usp,. and llm.. compositions. Introduction Since the introduction of the MtJlm. (for list of abbreviations see Table l) geothermometer-oxybarometer, most workers have utilized the graph of Lindsley (1963; Buddington and Lindsley, 1964, Fig. 5) to determine the temperature and oxygen fugacity of magnetite-ilmenite pairs. This graph has a number of drawbacks: (l) It involved interpolation between data obtained at a limited number of oxygen buffers, and this interpolation was largely intuitive; (2) for many Mt.,-Ilm,. pairs-especially those from volcanic rocks-it is necessary to extrapolate the curves to higher temperatures or to higher oxygen fugacities, or both; and (3) the treatment of such as Mn, Mg, Al, Cr, and V is necessarily arbitrary. Ideally one would like to know the effects of such constituents on the activities of Fe2O3 and FeTiO3 in llm and of Fe3Oa and Fe2TiOa in Mt.. An adequate solution model would eliminate these limitations-although additional experimental data would still be needed to assess quantitatively the effects of the minor constituents. We present here a solution model for coexisting Mt., and Ilm.r, based on a least-squares fit of thermodynamic parameters to experimental data obtained at 550 to 1200C. We also include results from a new series of experiments on pure Fe-Ti oxides done using the Co-CoO oxygen buffer, which has improved our understanding of the phase relations between Mt. and Ilm by permitting us to experiment in a region of ?-/6, space previously inaccessible to well-calibrated oxygen buffer assemblages. In the discussion that follows it is assumed that the Mt and Ilm. are pure binary Fe-Ti oxides; no minor constituents are considered. 0003-0M)uE I /l I l2-1 I 89$02.00 l lE9