Nitrogen solubility in the deep mantle and the origin of Earth's primordial nitrogen budget

Abstract

The solubility of nitrogen in the major minerals of the Earth's transition zone and lower mantle (wadsleyite, ringwoodite, bridgmanite, and Ca-silicate perovskite) coexisting with a reduced, nitrogen-rich fluid phase was measured. Experiments were carried out in multi-anvil presses at 14 to 24 GPa and 1100 to 1800 °C close to the Fe–FeO buffer. Starting materials were enriched in 15N and the nitrogen concentrations in run products were measured by secondary ion mass spectrometry. Observed nitrogen (15N) solubilities in wadsleyite and ringwoodite typically range from 10 to 250 μg/g and strongly increase with temperature. Nitrogen solubility in bridgmanite is about 20 μg/g, while Ca-silicate perovskite incorporates about 30 μg/g under comparable conditions. Partition coefficients of nitrogen derived from coexisting phases are DNwadsleyite/olivine = 5.1 ± 2.1, DNringwoodite/wadsleyite = 0.49 ± 0.29, and DNbridgmanite/ringwoodite = 0.24 (+0.30/−0.19). Nitrogen solubility in the solid, iron-rich metal phase coexisting with the silicates was also measured and reached a maximum of nearly 1 wt.% 15N at 23 GPa and 1400 °C. These data yield a partition coefficient of nitrogen between iron metal and bridgmanite of DNmetal/bridgmanite ∼ 98, implying that in a lower mantle containing about 1% of iron metal, about half of the nitrogen still resides in the silicates. The high nitrogen solubility in wadsleyite and ringwoodite may be responsible for the low nitrogen concentrations often observed in ultradeep diamonds from the transition zone. Overall, the solubility data suggest that the transition zone and the lower mantle have the capacity to store at least 33 times the mass of nitrogen presently residing in the atmosphere. By combining the nitrogen solubility data in minerals with data on nitrogen solubility in silicate melts, mineral/melt partition coefficients of nitrogen can be estimated, from which the behavior of nitrogen during magma ocean crystallization can be modeled. Such models show that if the magma ocean coexisted with a primordial atmosphere having a nitrogen partial pressure of just a few bars, several times the current atmospheric mass of nitrogen must have been trapped in the deep mantle. It is therefore plausible that the apparent depletion of nitrogen relative to other volatiles in the near-surface reservoirs reflects the storage of a larger reservoir of nitrogen in the solid Earth. Dynamic exchange between these reservoirs may have induced major fluctuations of bulk atmospheric pressure over Earth's history.