Trace element partitioning between apatite and silicate melts: Effects of major element composition, temperature, and oxygen fugacity, and implications for the volatile element budget of the lunar magma ocean

Abstract

Apatite, as an accessory phase in igneous and metamorphic rocks, has important petrological significance due to its capacity to accommodate appreciable amounts of many trace elements in its mineral structure. To better constrain trace element partitioning between apatite and silicate melts, we conducted experiments that produced apatites approaching fluorapatite (FlAp), hydroxylapatite (OHAp) and chlorapatite (ClAp) endmembers separately at 1050 and 1100 °C, 1 GPa pressure, under oxygen fugacity (fO2) about one log unit below iron-wüstite buffer to four log unit above fayalite-magnetite-quartz buffer. We report the results of 12 experiments which demonstrate that ClAp exhibits lower trace element partition coefficients compared with FlAp and OHAp, especially for Rare Earth Elements (REEs) under all run conditions explored, suggesting trace element partitioning is sensitive to anion site occupancy. Divalent cations are less sensitive to anion occupancy. Positive Eu partitioning anomalies (DEu/DEu*, where Eu is the chondrite normalized abundance and Eu* is the interpolated value from neighboring elements ordered by atomic number) are observed in ClAp experiments under relatively low fO2s, whereas negative Eu anomalies are exhibited by FlAp and OHAp under the same fO2 conditions. We infer that anionic occupancies have a direct impact on the substitution mechanisms of trace elements in apatite, thereby influencing their partition coefficients. Beyond the anions, correlations of apatite compositional components (XCa, XNa, XP and XSi) with partition coefficients suggest they exert crystal chemical controls on trace element partitioning. Based on these observations, we developed parameterized lattice strain models to predict the partitioning of divalent and trivalent elements as a function of temperature and apatite composition, and an fO2-dependent apatite-melt Eu partitioning model and oxybarometer. We further developed a Eu in apatite-plagioclase oxybarometer that enables us to calculate the fO2 of apatite and plagioclase-bearing magmatic and subsolidus systems, and evaluated the influence of subsolidus reequilibration on the new oxybarometer. Applied to one of our experiments, winonaite HaH193, and samples from Sept Iles layered intrusion, the oxybarometer recovers their anticipated fO2s, ranging from about two log units below the iron-wüstite buffer to the fayalite-magnetite-quartz buffer. Using the new REE and fO2-dependent Eu partitioning models, we constrained the petrogenesis of lunar KREEP basalt and estimated the relative volatile content in the late lunar magma ocean (LMO) cumulates. The model suggests a relative depletion of Cl in the LMO cumulates, consistent with Cl isotopic analyses and volatile abundance measurements in previous work, suggesting that differential loss of volatiles occurred before or during the late-stage evolution of the LMO.