PARTITION-COEFFICIENTS FOR IRON BETWEEN PLAGIOCLASE AND BASALT AS A FUNCTION OF OXYGEN FUGACITY - IMPLICATIONS FOR ARCHEAN AND LUNAR ANORTHOSITES

Experimental studies were undertaken to determine D for FeOT between plagioclase and melt as a function of oxygen fugacity for a basaltic composition that occurs as matrices for plagioclase megacrysts that are similar to those in calcic anorthosites. At liquidus conditions, the D for FeOT between calcic plagioclase and tholeiitic basalt changes very little, from 0.030 to 0.044, between the very low f(O2) of the iron-wustite buffer and that of the quartz-fayalite-magnetite (QFM) buffer. At fugacities above QFM the value for D increases rapidly to 0.14 at the magnetite-hematite buffer and 0.36 in air. The increase in D results from the fact that nearly all of the Fe is in the Fe2+ state at f(O2) below QFM, and the Fe3+/Fe2+ ratio in the melt increases rapidly above QFM causing more Fe to enter the plagioclase which accepts Fe3+ more readily than Fe2+. There is essentially no change in values of D over cooling rates that vary by two orders of magnitude. The D for FeOT of 0.030 at iron-wustite is in good agreement with earlier studies at low f(O2). Terrestrial calcic plagioclase-matrix pairs display Ds of about 0.043, in excellent agreement with the experimental value of 0.044 at QFM which is the accepted fugacity for terrestrial tholeiitic basalts. Application of the Ds for FeOT and other components to terrestrial calcic anorthosites indicates that the anorthosites were in equilibrium with nearby coeval tholeiitic basalts at QFM, as was expected. Lunar anorthosites, however, have anomalously low contents of FeOT as well as MgO and would be in equilibrium with melts that are several times lower in these components than expected. The low values of FeOT and MgO in the plagioclase of lunar anorthosites are a result of subsolidus reequilibration with pyroxene, olivine, and other Fe-Mg-bearing phases during high-temperature metamorphism, as is the case in terrestrial anorthosites that have undergone granulite-grade metamorphism. The Ds for K2O, MgO, and TiO2 were determined as 0.21, 0.044, and 0.033, respectively, and show no variation with f(O2). The value for K2O is in good agreement with natural assemblages. The values for MgO and TiO2, however, are lower in natural assemblages; but the high degree of mobility of Mg and the very low contents of TiO2 in plagioclase makes the reliability of the natural values rather low. Also, the D for TiO2 may be affected by rate of cooling. Attempts to determine the D for Cr2O3 as a function of f(O2) were made difficult by the low levels of Cr2O3 (300 to 500 ppm) in the melt when saturation with a Cr spinel occurred. Because the Cr concentrations in plagioclase are nearly 100 times less than those in the melt, it was not possible to obtain reliable analyses for plagioclase. These uncertainties constrain the D for Cr to be less than 0.07 at the QFM buffer.