Density of primitive lunar melts

Thermo-chemical models of the dynamics of the internal differentiation of the Moon require accurate knowledge of the density of lunar minerals and melts at a range of high pressure, high temperature conditions. Here we present a comparative study of high-pressure, high-temperature density determinations of possible primitive lunar melts, as represented by volcanic glass beads sampled during the Apollo missions. We performed molecular dynamic (MD) simulations on the density of molten low (Apollo 15C green), intermediate-high (Apollo 17 74,220 orange) and high titanium (Apollo 14 black) bearing lunar glass compositions at conditions bracketing those relevant for lunar interior melt evolution (pressures between 0 and 10 GPa and temperatures of 1723, 2073 and 2423 K). Density variations are well described by third-order Birch-Murnaghan equations of state. Isothermal bulk moduli at 1673 K, K1673, are 18.2 ± 0.2 GPa for green glass, 20.8 ± 1.0 GPa for orange glass, and 19.6 ± 0.3 GPa for the black glass composition. Corresponding pressure derivatives, K′, are 8.5 ± 0.2, 7.6 ± 0.6, and 9.2 ± 0.3, respectively. These values translate to calculated densities of 2.97-3.06 g cm (green), 3.10-3.19 g cm (orange) and 3.19-3.28 g cm (black) at liquidus temperatures and typical multiple saturation depths between 1.5 and 2.5 GPa. We compare these results with previously published ex situ and in situ experimental density measurements that were performed on synthetic analogues of these glasses. Within the lunar pressure range the results of these widely different techniques are comparable with MD simulations to within 5.9%. Irrespective of the technique used, the largest uncertainty in lunar melt density prediction at high pressure is the uncertainty in the room-pressure density values used to anchor both MD simulations and experiments, in particular for titaniumand iron-rich compositions.