The Moon-forming giant impact extensively melts and partially vaporizes the silicate Earth and delivers a substantial mass of metal to Earth's core. Subsequent evolution of the magma ocean and overlying atmosphere has been described by theoretical models but observable constraints on this epoch have proved elusive. Here, we report calculations of the primordial atmosphere during the magma ocean and water ocean epochs and forge new links with observations to gain insight into the behavior of volatiles on the early Earth. As Earth's magma ocean crystallizes, it outgasses the bulk of the volatiles into the primordial atmosphere. The redox state of the magma ocean controls both the chemical composition of the outgassed volatiles and the hydrogen isotopic composition of water oceans that remain after hydrogen loss from the primordial atmosphere. Whereas water condenses and is retained, molecular hydrogen does not condense and can escape, allowing large quantities (~10^2 bars) of hydrogen - if present - to be lost from Earth in this epoch. Because the escaping inventory of H can be comparable to the hydrogen inventory in the early oceans, the corresponding deuterium enrichment can be large with a magnitude that depends on the initial H2 inventory. By contrast, the common view that terrestrial water has a carbonaceous chondrite source requires the oceans to preserve the isotopic composition of that source, undergoing minimal D-enrichment via H2 loss. Such minimal enrichment places upper limits on the amount of primordial H2 in contact with early water oceans (pH2<20 bars), implies oxidizing conditions for outgassing from the magma ocean, and suggests that Earth's mantle supplied the oxidant for the chemical resorption of metals during late accretion.

Despite very high temperatures accompanying lunar origin, indigenous water in the form of OH has been unambiguously observed in Apollo samples in recent years. Such observations have prompted questions about the abundance and distribution of lunar hydrogen. Here, we investigate the related question of the origin of lunar H: is the hydrogen observed a remnant of a much larger initial inventory that was inherited from a wet Earth but partly depleted during the process of origin, or was hydrogen quantitatively lost from the lunar material, with water being delivered to lunar reservoirs via subsequent impacts after the origins sequence? Motivated by recent results pointing to a limited extent of hydrogen escape from the gravity field of the Earth during lunar origin, we apply a newly developed thermodynamic model of liquid-vapor silicates to the proto-lunar disk to interrogate the behavior of H as a trace element in the energetic aftermath of the giant impact. We find that: (1) pre-existing H-bearing molecules are rapidly dissociated at the temperatures considered (3,100-4,200 K) and vaporized hydrogen predominantly exists as OH(v), H(v) and MgOH(v) for nearly the full range of thermal states encountered in the proto-lunar disk, (2) despite such a diversity in the vapor speciation, which reduces the water fugacity and favors hydrogen exsolution from co-existing liquids, the equilibration of the vapor atmosphere with the disk liquid results in significant dissolution of H into proto-lunar magmas, and (3) equilibrium H isotopic fractionation in this setting is limited to < 10 per mil and the terrestrial character of lunar D/H recently inferred should extend to such a precision if liquid-vapor equilibration in the proto-lunar disk is the process that gave rise to lunar hydrogen. Taken together, these results implicate dissolution as the process responsible for establishing lunar H abundances.

The Moon is generally thought to have formed from the debris ejected by the impact of a planet-sized object with the proto-Earth towards the end of planetary accretion. Modeling of the impact process predicts that the lunar material was disaggregated into a circumplanetary disk and that lunar accretion subsequently placed the Moon in a near equatorial orbit. Forward integration of the lunar orbit from this initial state predicts a modern inclination at least an order of magnitude smaller than the lunar value, a long-standing discrepancy known as the lunar inclination problem. Here we show that the modern lunar orbit provides a sensitive record of gravitational interactions with Earth-crossing planetesimals not yet accreted at the time of the Moon-forming event. The excited lunar orbit can naturally be reproduced via interaction with a small quantity of mass (corresponding to 0.0075-0.015 ME eventually accreted to the Earth) carried by a few bodies, consistent with constraints and models of late accretion. While the process has a stochastic element, the observed value of the lunar inclination is among the most likely outcomes for a wide range of parameters. The excitation of the lunar orbit is most readily reproduced via collisionless encounters with strong tidal dissipation on the early Earth. This mechanism obviates the necessity of previously proposed (but idealized) excitation mechanisms and can constrain the planet formation context of the Moon-forming event and the pristineness of the dynamical state of the Earth-Moon system.

Simulations of the moon-forming impact suggest that most of the lunar material derives from the impactor rather than the Earth. Measurements of lunar samples, however, reveal an oxygen isotope composition that is indistinguishable from terrestrial samples, and clearly distinct from meteorites coming from Mars and Vesta. Here we explore the possibility that the silicate Earth and impactor were compositionally distinct with respect to oxygen isotopes, and that the terrestrial magma ocean and lunar-forming material underwent turbulent mixing and equilibration in the energetic aftermath of the giant impact. This mixing may arise in the molten disk epoch between the impact and lunar accretion, lasting perhaps 10^2-10^3 yr. The implications of this idea for the geochemistry of the Moon, the origin of water on Earth, and constraints on the giant impact are discussed.

Despite its importance to questions of lunar origin, the chemical composition of the Moon is not precisely known. In recent years, however, the isotopic composition of lunar samples has been determined to high precision and found to be indistinguishable from the terrestrial mantle despite widespread isotopic heterogeneity in the Solar System. In the context of the giant-impact hypothesis, this level of isotopic homogeneity can evolve if the proto-lunar disk and post-impact Earth undergo turbulent mixing into a single uniform reservoir while the system is extensively molten and partially vaporized. In the absence of liquid-vapor separation, such a model leads to the lunar inheritance of the chemical composition of the terrestrial magma ocean. Hence, the turbulent mixing model raises the question of how chemical differences arose between the silicate Earth and Moon. Here we explore the consequences of liquid-vapor separation in one of the settings relevant to the lunar composition: the silicate vapor atmosphere of the post-giant-impact Earth. We use a model atmosphere to quantify the extent to which rainout can generate chemical differences by enriching the upper atmosphere in the vapor, and show that plausible parameters can generate the postulated enhancement in the FeO/MgO ratio of the silicate Moon relative to the Earth's mantle. Moreover, we show that liquid-vapor separation also generates measurable mass-dependent isotopic offsets between the silicate Earth and Moon and that precise silicon isotope measurements can be used to constrain the degree of chemical fractionation during this earliest period of lunar history. An approach of this kind has the potential to resolve long-standing questions on the lunar chemical composition.