Solubility of molecular hydrogen in silicate melts and consequences for volatile evolution of terrestrial planets

We present experiments from 0.7 to 3 GPa that quantify solubility of H2 in silicate melts under controlled hydrogen fugacities (fH2). Two experimental series, one on synthetic basalt+COH and other with a synthetic andesite+OH, were conducted using a double capsule technique to impose a range of fH2, on the samples. Quenched glasses were analyzed by FTIR and SIMS. Both series follow simple solubility laws in which molecular H2 concentrations are proportional to fH2 and with a partial molar volume of molecular H2 of 11 cm3/mole. Solubilities in andesitic melt are systematically greater than in basaltic liquid in a relationship consistent with control by the ionic porosity (IP) of the melts. Extrapolation based on IP allows estimation of the solubility of H2 in peridotitic melts applicable to magma oceans. The H2/(H2+H2O) ratio in silicate melts (where H2O includes molecular H2O and OH−) increases as conditions become more reduced, with increasing pressure, and with increasing total H. Under some conditions prevailing in the early Earth and terrestrial planets as well as in the deep Earth today, H2 can be a significant fraction of the dissolved H and at high pressure it may exceed “water” (H2O and OH−). Therefore, magmatic H2 may influence the initial distribution of volatiles and the redox evolution of terrestrial planets, as well as the ongoing formation and fate of hydrous melts in the deep Earth today. Hydrous species in melts in equilibrium with Fe-rich alloy at high pressure, for example during core formation from a magma ocean, could be chiefly H2, rather than H2O. Hence, delivery of H2 to the core by removal of Fe hydride need not be coupled to oxidation of the residual mantle. Although lunar basalts are much reduced, the fraction of H dissolved as molecular H2 is small owing to low total H concentrations. Extrapolation to conditions of potential hydrous partial melting in the deep Earth suggests that the chief magmatic volatile may be H2 rather than H2O. The very small partial specific density of magmatic H2 (0.18 g/cm3 at low pressure) may provide significant positive buoyancy to deep partial melts.