Experimental determinations of trace element partitioning between plagioclase, pigeonite, olivine, and lunar basaltic melts and an fO2 dependent model for plagioclase-melt Eu partitioning

We report partition coefficients for Sr, Sc, La, Ce, Pr, Nd, Pm, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Y, and Th for plagioclase, pigeonite, and olivine experimentally grown from a lunar basaltic liquid at 1 bar and two different oxygen fugacities (fO(2); fayalite-magnetite-quartz (FMQ) and the iron-wustite (IW) oxygen buffers) at temperatures (Ts) of 1100-1125 degrees C. These results expand the partitioning database for plagioclase-silicate melt partitioning at lunar relevant conditions and compositions. Variations of partition coefficients for rare earth elements (REEs) are highly systematic with respect to ionic radii, in agreement with lattice strain theory. We use the lattice strain model to evaluate partition coefficients for highly incompatible elements such as the light (or in the case of plagioclase, heavy) REEs, for which direct determinations are experimentally and analytically difficult. Our experiments suggest that within analytical and experimental uncertainty, over the range of explored fO(2) (FMQ to IW), T (1100-1125 degrees C) and An# conditions (An# >94.5, where An#=Ca/(Ca + Na) x 100, in moles), fO(2) exerts an important but secondary control on Eu partitioning in anorthitic plagioclase. In contrast, the global plagioclase-melt partitioning dataset highlights the first-order role of fO(2) in determining plagioclase-melt Eu partition coefficients over a wide range of An#s. Using experimental results and predictive models for divalent and trivalent element plagioclase-melt partitioning in the plagioclase ring site, we parameterized an fO(2)-dependent plagioclase-melt Eu partitioning model. Within a factor of two, the model reproduces measured Eu partition coefficients for a wide range of experimental conditions (750-1400 degrees C) and compositional systems (basalts to dacites) over > 15 orders of magnitude variation in fO(2). When equilibrium partitioning of Eu between plagioclase and melt can be established, the model may be applied as a plagioclase-melt oxybarometer. Preliminary application to plagioclase-hosted melt inclusions in mid-ocean ridge basalts suggests they equilibrated near the FMQ buffer. We used the new Eu partitioning model to evaluate the petrogenesis of a lunar anorthositic impactite that hosts plagioclase with both positive and negative Eu anomalies. Our investigation suggests the distribution of REEs and Eu in this sample is consistent with subsolidus reequilibration after impact-induced heating. (C) 2020 Elsevier Ltd. All rights reserved.