Every model here is a giant impact. They disagree on the impactor, on how fast Earth spun, and on what the debris became. A 2023 review and a 2023 survey both find that no single model yet fits every constraint, and a 2024 follow-up agrees.
| Model | Impactor, then what it leaves (model) | Explains the isotopes? |
|---|
| Canonical (2001, 2014) | Mars-sized, about 10 to 15% of the total mass, near escape speed, about 45°. A mostly molten disk, roughly 10 to 30% vapour, up to 2 Moon masses | No: over 80% of the disk comes from Theia |
|---|
| Hit-and-run (2012) | At least 20% of the total mass, 1.2 to 1.4 times escape speed, 30 to 40°. A disk 40 to 60% impactor | Partly |
|---|
| Fast-spinning Earth (2012) | 2.6 to 10% of Earth's mass, 1.5 to 3 times escape speed, into an Earth whose day was about 2.5 hours. Mostly vapour | Yes: the disk comes mainly from Earth's mantle |
|---|
| Half-Earth (2012) | Two bodies of 40 to 45% and 55 to 60% of the total. A disk of 1.7 to 5.6 Moon masses, 50 to 90% vapour, and a day of 2.0 to 2.7 hours | Yes, by symmetry |
|---|
| Synestia (2017, 2018) | High-energy, high-spin impacts. Earth becomes a spinning doughnut of rock vapour, and the Moon grows inside it | Yes, if the vapour mixes well, which is not established |
|---|
| Many impacts (2017) | About 20 to 30 impacts, each 1 to 10% of Earth's mass. Small moons that must merge | Better odds, but only if the small moons merge, and later work finds they often do not |
|---|
| Immediate satellite (2022) | 13% of Earth's mass, about 45°, about 9 km/s. A body of 0.69 Moon masses on a wide orbit, within hours | Partly: its outer layers are about 60% Earth |
|---|
Too much spin. Several of these models leave the Earth and Moon with up to 2.7 times today's angular momentum, so something had to brake them. Two brakes have been proposed, a resonance with the Sun and a tilt instability as the Moon moved out, and both are debated.
What is drawn here is the canonical impact and its disk, the case the reviews measure the others against. Its known problem, the isotopes, is in the table. The half-Earth and synestia cases would look very different, so they are listed here rather than drawn.
When. Published dates for the impact run from about 4.35 to 4.51 billion years ago. Among them: 4.51 billion years or earlier (zircons, 2017), 4.425 ± 0.025 billion (a model of the young Moon's magma ocean, 2020), and at least 4.46 billion (zircons, dated in 2021 and checked in 2023). A 2024 study reads the many rock ages near 4.35 billion as a later remelting, so the Moon may be older. The screen says about 4.5 billion years ago, an estimate.
The impact. In canonical runs Theia carries about 10 to 15% of the total mass, strikes at about 45°, and below 1.1 times escape speed, and almost all of its iron core joins Earth's. The 2022 model's impactor: 13% of Earth's mass at about 9 km/s.
The disk. Canonical runs put it at 3,000 to 4,000 K; hotter variants reach 6,000 to 7,000 K. It held about 1 to 5 Moon masses across the models, much of it inside the Roche limit, about 2.9 Earth radii (about 18,500 km, our arithmetic). The Moon gathers at about 3.8 to 4.8 Earth radii (about 24,000 to 31,000 km, our arithmetic): about 40% of it in months and the rest over about a century, in one model. Other models keep a disk for up to about 1,000 years. Earth's surface then takes around 3,000 years to cool. If the Earth and Moon have lost no spin since, Earth's day just after the impact was about 5 hours.
Measured today. The Moon is 384,400 km away on average. Its titanium matches Earth's within about 4 parts per million, 1/150 of the range in meteorites. Oxygen: 0.2 ± 1.6 parts per million apart (2024); the same group had reported a 12 part per million difference in 2014 and did not find it again. The Moon's iron core is small, about 1 to 3% of its mass.
Theia. Iron isotopes and a mass balance suggest it formed in the inner Solar System, perhaps closer to the Sun than Earth (2025). Its colour, surface and orbit are unknown, so it is drawn plain grey. Earth before the impact is drawn plain grey too, since the research behind this story does not describe its surface.
The first day. It follows the published runs hour by hour. At contact the rock between the two bodies is shocked to over 9,000 K, and the hottest of it jets ahead at up to about 14 km/s. The half of Theia that missed Earth shears off: after about 50 minutes it is a bar 3 to 3.5 Earth radii long, and by about 2 hours a spiral arm, while one bulge runs about two-thirds of the way round Earth. The arm beads into clumps. The inner clump, holding Theia's iron core, hits Earth again after about 6 hours (about 8 in the 2022 model's animation). The outer clump swings out to about 50,000 km, then on a close pass is torn into a second arm that breaks into small clumps by about 27 hours. In about a quarter of such runs that clump survives as a moon-sized body instead. In a grid code the vapour is a smooth, lopsided cloud over 40,000 km across. The disk settles after 20 to 30 hours, thin and flat inside a thick, hot cloud. The runs disagree on how many clumps form and where the core lands, so the shapes and counts drawn are illustrative.
Drawn here. Earth, Theia and the Moon at about their true sizes and at the models' distances, lit from one side as if by the Sun. Earth is drawn at today's size, though in the 2022 model it had about 0.89 of today's mass before the impact. No run shows what the impact looked like to the eye: the simulations colour their particles by temperature, energy or material. So nothing here is a flash of light. The colours are black-body hints at the runs' temperatures, deepened so they show, over 9,000 K at the contact, along the arm from about 3,000 K at its tip to over 6,000 K at its root in the first hour, from about 2,000 K at the disk's outer edge to 4,000 K inside. How bright each part glows, the shapes of the debris, the disk's thickness and the paths between the sourced moments are illustrative, and the orbits are slowed. The 2022 simulation by Kegerreis and colleagues can be watched in NASA's release (video: NASA, Durham University, Jacob Kegerreis).
- Canup and Asphaug 2001, Nature 412: the canonical impact, the small core, the 5 hour day
- Canup 2014, Phil. Trans. R. Soc. A 372: the impactor, the angle, the core, the disk and its heat
- Canup et al. 2023, Reviews in Mineralogy and Geochemistry 89: the review: how fast the Moon gathers, the open questions, Venus
- Kegerreis et al. 2022, ApJL 937, L40: the Moon in hours, 9 km/s
- NASA, 4 October 2022: the simulation video, the Artemis samples
- Salmon and Canup 2012, ApJ 760, 83: the Roche limit, where the Moon gathers, a century
- Nakajima and Stevenson 2014, Icarus 233: hotter disks, disk masses
- Pahlevan and Stevenson 2007, EPSL 262: the disk's lifetime, Earth's magma ocean
- Zhang et al. 2012, Nature Geoscience 5: titanium
- Fischer et al. 2024, PNAS 121: oxygen
- Hopp et al. 2025, Science 390: where Theia formed (Max Planck release)
- Nimmo, Kleine and Morbidelli 2024, Nature 636: the date range
- Barboni et al. 2017, Science Advances 3: 4.51 billion years or earlier
- Maurice et al. 2020, Science Advances 6: the young Moon's magma ocean, 4.425 billion years
- Greer et al. 2023, Geochemical Perspectives Letters 27: zircons at least 4.46 billion years old
- Yuan et al. 2023, Nature 623: buried Theia, a hypothesis
- NASA, Earth's Moon: Facts: 384,400 km
- Timpe et al. 2023, ApJ 959: a survey of impacts
- Meier et al. 2024, ApJ 978: the survey with spinning bodies
- Ćuk and Stewart 2012, Science 338: a fast-spinning Earth
- Canup 2012, Science 338: the half-Earth case
- Lock and Stewart 2017, JGR Planets 122: synestias
- Lock et al. 2018, JGR Planets 123: the Moon inside a synestia
- Rufu, Aharonson and Perets 2017, Nature Geoscience 10: many impacts
- Rufu and Canup 2020, JGR Planets 125: the resonance brake
- Ćuk et al. 2016, Nature 539: the tilt brake
- Ćuk et al. 2021, Planetary Science Journal 2: the tilt brake, revisited
- Canup 2004, Icarus 168: the first day hour by hour: the jet, the arm, the core's return, the close pass
- Canup, Barr and Crawford 2013, Icarus 222: the same impact at three resolutions and in a grid code: the vapour cloud, when the disk settles
- Durham ICC, giant impact animations: the 2022 model hour by hour, the core's return at about 8 hours