It is now widely accepted that the Moon formed as the result of a collision between two planet-sized objects. An event generally referred to as the giant impact. But that’s more or less where the consensus ends. There is currently little agreement on the relative size of the colliding bodies or the energy of the impact. So, there are all sorts of models out there, from relatively low energy encounters, to multiple impacts producing a family of moonlets, high-energy fast spinning collisions, head on crashes between half-sized Earths and ultra-high energy impacts that resulted in the formation of extended vapour-rich discs. It’s quite a list.
The classic giant impact model for the formation of the Moon
The classic version of the giant impact model for was put forward by Canup and Ashaug (2001). It envisaged an oblique collision between the proto-Earth and a Mars-sized impactor. The resulting debris disc, from which the Moon formed, would have contained up to 70% impactor-derived material. So, if the two colliding bodies were isotopically different then the Moon should show some differences from the Earth. It was a very nice model, being constrained by the present angular moment of the Earth-Moon system. And of course, like all good scientific models, it could be tested, and was.
The problem for the classic giant impact model was that, for a range of isotopic systems, the Earth and the Moon are almost identical. It was a real case of back to the drawing board. Or was it?
One way round this problem would be if the impactor and Earth were essentially identical isotopically before the encounter. This is a suggestion that has often been made, most recently in the study of Schiller et al. (2018) with respect to calcium isotopic variation in terrestrial and lunar rocks (see recent post).
Oxygen isotopes and the giant impact
Oxygen isotopic variation is an interesting case. Most recent studies couldn’t detect a difference between the Earth and Moon, with one notable exception. Herwartz et al. (2014) suggested that there was a 12 ± 3 ppm ∆17O difference between lunar and terrestrial rocks. It was a difference that the later study of Young et al. (2016) was unable to detect.
Samples were analysed from every Apollo mission (image: NASA)
The oxygen isotope evidence was puzzling, so we decided to look at it again. We analysed a large suite of both terrestrial and lunar rocks and came up with a very small Earth-Moon difference of only 4 ppm.
Wow! that’s almost nothing!
In fact, if you look at the predictions of high-energy models by Cuk and Stewert (2012) and Canup (2012), half of that 4 ppm difference could be explained in terms of slight differences in the extent of mixing during the collision itself. The latest high-energy model for the giant impact by Lock et al. (2018), which envisages the formation of an extended vapour-rich disc in the aftermath of the collision, would seem to require complete mixing between the Earth and impactor.
Was the planet that hit the Earth an aubrite?
But what about that possibility that the Earth and impactor were almost identical before the collision? Well, we looked at that by taking the aubrites as proxies for the impactor. In fact, the aubrites are a pretty good candidate in terms of their chemical composition, based on the criteria given in the study of Wood and Wade (2016). They were previously considered to be virtually identical in oxygen isotope composition to terrestrial and lunar rocks. But our study shows that there is a 22 ppm average ∆17O difference between aubrites and terrestrial rocks. Using the parameters of the standard giant impact model, an impactor with the oxygen isotope composition of aubrites would produce an easily detectable 15 ppm difference in ∆17O between the Earth and Moon. Unlike the possible case of calcium isotopes, when it comes to oxygen (∆17O), it seems to require very special pleading to invoke the possibility that the impactor and proto-Earth were identical prior to impact. The inescapable conclusion must be that the isotopic match between the Earth and Moon resulted from mixing and homogenisation in the aftermath of the high energy collision itself.
But what about that 4 ppm ∆17O Earth-Moon difference?
The Earth continued to grow after the Moon-forming giant impact. (image: BBC/ Solarseven)
Well, we know that Earth’s accretion didn’t stop with the giant impact. Based on highly siderophile element data it is estimated that the Earth grew further by an amount equivalent to 0.5% of its present mass. In contrast, the Moon grew by only a negligible amount. This most likely reflects the addition to Earth of a limited number of relatively large bodies, possibly with a diameter greater than 1,500 km. What was the isotopic composition of these bodies? To preserve the very small Earth-Moon difference they would have to have been extremely close to the ∆17O isotopic composition of the present-day Earth and Moon. The enstatite chondrites are good candidates. In fact, the aubrites, which are related to the enstatite chondrites, are even better, being the products of a differentiated body with a core, crust and mantle.
One problem though is that enstatite chondrites and aubrites are dry and if you want to add a significant amount of water to the Earth after the giant impact, as some have suggested, you have a big problem. Material from water-bearing asteroids arriving on Earth form part of a primitive class of meteorites known as carbonaceous chondrites. Such materials generally have very distinct oxygen isotope compositions compared to terrestrial rocks. So, based on our new oxygen isotope data, you can only add a limited amount of this material to the Earth after the giant impact. Oxygen is not the only evidence for a small amount of hydrated material in this so called “late veneer”. Ru isotopes also seem to point to dry enstatite chondrite-like material as being the major component of the late veneer.
So what proportion of Earth’s water could have arrived after the late veneer?
Well, now we have a problem. Estimates of the global water content are all over the place, from 2 to 12 ocean equivalents. Our data suggests that you can add as little as 5% of the Earth’s water, to as much as 30%, depending on which estimate you use. But the inescapable conclusion must be that most of Earth’s water was there before the giant impact. A recently published numerical study by Nakajima and Stevenson has reached similar conclusions. It seems entirely possible that at the time of the giant impact the Earth already had oceans.
At the time of the giant impact the Earth may have already had oceans (image: Wikipedia)
So, Earth’s water arrived early and survived the giant impact. We know that such energetic collisions also take place on some exoplanets. By implication they would also retain any water that they had accreted when experiencing the final energetic phase of planet building.
Artist’s impression of a planetary collision observed by NASA’s Spitzer Space Telescope around the young star NGC 2547-ID8
So how and when did Earth get most of its water?
The recent study of Schiller et al. (2018) indicates that inner solar system bodies were receiving a steady input of carbonaceous chondrite material throughout their accretion history. This shows up as an apparent correlation between μ48Ca values and the mass of an asteroid or planet. Schiller et al. (2018) suggest that outer solar system material was leaking into the inner solar system, progressively shifting the μ48Ca composition of the growing inner solar system bodies with time. Some of these outer solar system materials would also have been water-rich. In other words, water would have been added to the growing inner solar system bodies throughout their main accretion phase. If correct, this means that water was always a component within the inner solar system and of course a vital ingredient for the eventual emergence of life.