This is chapter 8 in the Solar System’s story. We’ll go into the final stages of growth of the rocky planets, which included the giant impact that formed the Moon.
Earth has the biggest Moon relative to its size of any planet in the Solar System. The Moon is about a quarter the size of Earth and 1/80th the mass (due to its much smaller iron core). It is similar in composition to Earth’s mantle, although it is quite dry. Of particular importance, the Moon has nearly identical Oxygen isotopes to Earth (as we’ll see below).
When and how did the Moon form? Let’s jump into the “how” part first. There have been a number of discarded hypotheses related to how the Moon formed such as fission (the Moon being a blob that was expelled by the fast-spinning young Earth), capture of a wandering protoplanet, co-accretion of the Earth and Moon, and a nuclear explosion (to launch a piece of Earth into orbit).
The Moon-forming impact
The giant impact hypothesis is currently the leading model for the Moon’s formation. A giant impact between the proto-Earth and another planetary embryo could have spun out a disk of rock vapor around the young Earth from which the Moon coalesced. Here is a computer simulation of a Moon-forming collision:
There are several flavors of Moon-forming impacts (all the gory details here). The “canonical” Moon-forming impact envisions the collision between the almost-formed Earth (with about 90% of its current mass) and a Mars-mass planetary embryo. The canonical impact produces a Moon with the right orbit, but most of the Moon’s mass comes from the impactor, called Theia. To explain why the Earth and Moon have such similar compositions (especially in those pesky Oxygen isotopes), Theia must have grown very close-by to Earth. That type of outcome does indeed happen in simulations of terrestrial planets but it’s pretty rare.
A more energetic impact would mix the proto-Earth and Theia together such that the Earth and Moon’s compositional match comes from the fact they were both made from the same mix. More energetic impacts can involve either planetary embryos that were spinning very fast before the impact, or an impact between roughly equal-mass planetary embryos. Either case can explain the Earth-Moon compositional similarity (at least in broad strokes).
The giant, single-impact is currently favored, but there are plenty of wrinkles and alternate scenarios in the impact hypothesis. For example, the Synestia hypothesis proposes that the impact was so energetic that it generated a giant torus of rock vapor, whose evolution determined the Moon’s final properties. Another model suggests that the Moon formed not from a single giant impact but from a series of smaller ones.
When did the Moon-forming impact take place?
The Moon-forming impact was probably the final giant impact in the formation of the Earth. Other, later impacts of planetary embryos would likely have destabilized the Moon from its orbit. The time of the Moon-forming impact therefore measures the end of Earth’s growth (apart from the late accretion phase — see below).
The Hafnium-Tungsten radioactive system can be used to measure the timing of the Moon-forming impact. Hafnium (182Hf) decays into Tungsten (182W) with a half-life of 9 million years. In geology-speak, Hafnium is lithophile, which means it “follows the rocks”. But Tungsten is siderophile and “follows the iron”. This matters because Earth is differentiated. Most of Earth’s rocks are in the mantle and crust, but most of its iron is in the core.
The amount of Tungsten in Earth’s crust and mantle can be used as a clock to tell us when Earth differentiated. Imagine Earth had differentiated very quickly, long before any Hafnium had the time to decay into Tungsten. In that case, Earth’s core would have “closed” early, and all of the Tungsten would have remained in the mantle and crust, even though it is siderophile and wanted to follow the iron. Now instead, imagine that Earth differentiated very late, long after all of its Hafnium had decayed into Tungsten. In that case, all of Earth’s Tungsten would have sunk to the core and there would be none left in the mantle or crust.
Reality is in between those extremes, and things are never as simple as we would like. Earth likely differentiated multiple times, and some of the impactors into the growing Earth were likely differentiated themselves (such as planetary embryos) whereas others were not (like smaller planetesimals). It’s also uncertain to what degree the Hafnium-Tungsten system equilibrates, or resets upon impact. Nonetheless, the Hafium-to-Tungsten (or Hf/W) ratio can measure Earth’s final differentiation event. Since the Moon-forming impact was the last giant impact, and it almost certainly had enough energy to trigger differentiation of the Moon, these two events must be linked.
The current estimate is that the Moon-forming impact took place about 40 to 150 million years after the start of planet formation. This was the final giant impact in Earth’s growth and likely the last giant collision in the inner Solar System. Simulations of terrestrial planet formation can match this timescale relatively easily, at least in planetesimal-driven scenarios from chapter 6 (e.g. the Grand Tack, Low-mass Asteroid belt and Early Instability models).
After the Moon formed, its orbit relative to Earth was controlled by tidal evolution. The Moon’s orbit expanded dramatically as it was tidally pushed away from Earth. The Moon and Earth’s spins were also affected. The Moon is still tidally evolving today, moving away from the Earth at a rate of about 3.8 cm (1.5 inches) per year, about the same rate at which fingernails grow.
“Late accretion”: planetesimals crashing into Earth after the Moon-forming impact
As we saw, the Moon-forming event represents the last time that Earth differentiated. After this point, any siderophile elements were stuck in the mantle or crust, despite their “iron-loving” nature.
Planetesimals that crashed into Earth delivered siderophile elements to Earth’s mantle and crust. But that was reset every time there was a giant (embryo) impact, because it triggered a differentiation event and all the siderophile elements sunk into the core. The final “reset” was the Moon-forming impact. All of the siderophile elements in Earth’s mantle and crust today came from planetesimal impacts after the Moon-forming impact. Here is an animation to illustrate this point:
The term “late accretion” refers to planetesimal impacts after the Moon-forming collision. Sometimes geophysicists call this the “late veneer.”
How much late accretion was there? We can figure that out with “highly-siderophile elements,” the most iron-loving of them all. Most are elements we don’t think of often in every day life, like osmium, rhenium, and iridium (although a global iridium-rich layer was an early indicator of the dinosaur-killing K/Pg impact). But some highly-siderophile elements are very precious, like platinum and gold!
The abundance of highly-siderophile elements (HSEs) is extremely low in the crust and mantle. Compared with primitive meteorites, HSEs show up at about 1 part in 200. This means, in simple terms, that only about the last half of one percent of Earth’s mass came after the Moon-forming impact. Nonetheless, that last half percent is responsible for all of humanity’s gold necklaces!
There are several instances where measurements of highly-siderophile elements feed back into models of terrestrial planet formation. One is simply the amount of HSEs. If a simulation of Earth’s growth ends up with too much mass colliding with the Earth after the last giant impact, it is inconsistent with reality because it predicts too high an abundance of HSEs than are actually found in Earth’s crust and mantle.
Yet the measured abundance of HSEs can be used as a clock for the Moon-forming impact. Imagine the Earth forming within a sea of planetesimals and planetary embryos. The earlier Earth’s final giant impact takes place, the larger the number of planetesimals remaining in the inner Solar System, a fraction of which will collide with Earth and deliver HSEs. In contrast, a very late Moon-forming impact implies that very few planetesimals should remain to deliver HSEs to Earth’s mantle and crust. Using a large suite of simulations to calibrate their results, one study found a best-fit age of the Moon-forming impact of about 95 million years after the start of planet formation. This matches the Hafnium-Tungsten estimate nicely.
The abundance of HSEs in the Moon and Mars are quite a bit lower than on Earth, even when accounting for their smaller masses. This is hard to understand, since the same population of planetesimals should have been colliding with all bodies in the inner Solar System. There are a couple of possible solutions to this problem. One is that perhaps the leftover planetesimals were very top-heavy, meaning that most of their mass was in a small number of very large bodies. In that case, late accretion has a significant random component and Earth usually ends up with more collisions than the Moon or Mars because it is more massive. Another possible solution is that the Moon’s crust took a long time to cool such that impacts before a certain time were not recorded. Yet another is that Earth simply held on to its HSEs more efficiently than other, smaller objects because the probability of non-accretionary collisions increases strongly at smaller sizes.
To conclude, the Moon-forming impact and late accretion are great examples of advances being made by combining cosmochemistry and numerical simulations. The future of planet formation will definitely include a lot of connection between disciplines.
Wrap-up
The TL;DR version of this post: The Moon is thought to have formed in Earth’s last giant impact with a planetary embryo about 100 million years after the start of planet formation. Highly-siderophile elements indicate that ~0.5% of Earth’s mass was accreted after that point.
Additional Resources
- The Solar System’s story (link to all posts so far)
- A cosmic crash may have made the Moon (Nautilus article about explaining the Earth-Moon compositional similarity)
- The Moon is underrated (Nautilus article article about the likelihood of forming large Moons around other stars)
- Can moons have moon? (A common question)
- Finally, a drawing of the Moon-forming impact by my son Owen (from our astronomy poem book):
PLANETPLANET 
You write “canonical” within quotation marks. Modern science does not have a “canon” as the term is usually understood. Does your use of “canonical” mean it is the most common explanation? This is the closest thing to a “canon” science has.