Making the Moon Anew

An important new dynamical effect has made it easier for scientists to simulate the Moon's formation following a giant collision with primordial Earth

During the 1970s, scientists proposed that an object the size of Mars could have collided with Earth and thrown enough matter into orbit to create the Moon.
Don Davis / The New Solar System
One of the solar system's most nagging problems literally stares into the collective faces of planetary scientists on many clear nights every month. It's the Moon — or, specifically, how it came to exist.

Prior to the Space Age, scientists kicked around three models of lunar origin. In the first, the Moon and Earth simply formed together, existing as paired worlds from the outset. A second concept envisioned that the Moon was a maverick world that strayed too near Earth and became trapped in orbit around it. In the third model, Earth somehow began to spin so fast that a huge blob of its crust and mantle tore away to create the Moon. However, this "fission hypothesis" would have left the Earth-Moon system with far more angular momentum than it has today.

Once researchers got their hands on actual samples of the Moon, it became clear that none of these models was a good fit. For example, lunar rocks lack the water, iron, and certain other metals that are common on Earth, so forget about side-by-side formation. Yet the relative abundances of oxygen's three isotopes in lunar and terrestrial rocks are identical, so the Moon and Earth must share some common genetic link — and that rules out the capture hypothesis.

About 30 years ago a radical new idea emerged. If a body roughly the size of Mars had struck the very young Earth, then superheated vapor and debris from this "big splat" could have formed an orbiting ring of matter that quickly coalesced into a single large object. Making the Moon this way would satisfy a host of physical and geochemical constraints — but not all of them.

All along researchers realized that this scenario only works if the impactor dealt Earth a glancing blow, in order to eject a Moon's worth of debris into orbit. But decades of computer simulations of a glancing blow have all yielded the same result: a lunar composition with too much impactor and too little proto-Earth.

Frames from a computer simulation show how debris from a giant impact with early Earth evolved into a encircling ring of superheated matter — derived from both objects — that later collected and cooled to form the Moon.
Matija Ćuk & Sarah Stewart
Now there's a breakthrough that sidesteps this geochemical dilemma. The new results were published and also presented at a big meeting of planetary scientists last month, and we noted them at the time, but I want to delve into the story a little more deeply here.

The key is what happened after the impact. Initially, the Moon would have been very close to Earth. Strong tidal forces would have transferred energy to the Moon's orbit, pushing it farther outward, at the expense of Earth's rotation, which gradually slowed. (This process continues slightly even today.) Soon the two bodies became coupled in what's called an evection resonance. Dynamicists Jihad Touma and Jack Wisdom had probed this effect back in 1998. They found that it could pump up the Moon's orbital eccentricity and inclination — neatly explaining why the lunar orbit is tipped so much with respect to Earth's equator — but leaving the system's total angular momentum virtually unchanged.

However, Matija Ćuk (SETI Institute) and Sarah Stewart (Harvard University) have taken another look at the evection resonance and discovered it that could have lasted long enough to transfer much or even most of the post-impact angular momentum to the Sun. As they report in Science Express for October 17th, this means Earth could have been spinning so rapidly when it was hit, rotating once in just 2½ hours, that it was close to flying apart on its own.

With angular momentum no longer a constraint, it's much easier to fashion a Moon with the right mass and with a composition identical to that of Earth's mantle. Ćuk and Stewart's simulations envision relatively small impactors with only 3% to 5% of Earth's mass. They get the best fit to lunar-sample constraints with head-on and slightly retrograde impacts at velocities from 10 to 30 km (6 to 20 miles) per second.

The collision of primordial Earth with a like-size body might have led to the Moon's formation. In this sequence from a computer simulation, the objects collide at a relatively slow 12.3 km (7.6 miles) per second. The result is an iron- and volatile-poor silicate disk orbiting Earth, with about three times the Moon's mass. Each frame is 25,000 miles (40,000 km) wide. Click here to watch an animation of the collision.
Science Express / M. Ćuk & S. Stewart
Robin Canup (Southwest Research Institute) uses the relaxed angular-momentum constraints to try a very different approach. In her computer simulations, the impactor and primordial Earth have essentially the same mass. The two strike obliquely, though more nearly head-on than in her prior runs. The wildly gyrating two-lobed blob that results eventually settles into a molten, iron-rich central mass surrounded by a white-hot silicate-rich disk. The planet and disk end up with identical isotopic ratios, Canup explains, because the impactor adds its substantial mass to the planet and the two bodies become compositionally blended.

Historical footnote: The fission model for the Moon's origin was originally proposed by George Darwin (Charles's son) in 1879. He calculated that if the primordial Earth had spun faster and faster, it would have assumed a bowling-pin shape just before a large chunk broke away to form the Moon. (He calculated correctly that the spin rate would have to be about 2½ hours for this to occur.)

With a few notable exceptions, latter-day scientists dismissed the idea because there was no obvious way to slow down Earth's spin enough after the Moon calved away. Now they realize it's quite doable.

There might still be a "gotcha" in all this. The Earth-Moon system ends up with the right amount of angular momentum only if the two bodies remain locked in the evection resonance long enough after the impact, and if each body has just the right tide-inducing characteristics during that time. But at least the isotopic mismatches have been resolved, and it's more certain than ever that the Moon had a very dramatic birth.

8 thoughts on “Making the Moon Anew

  1. Rod

    My observation – we have two models described for the giant impact featuring radically different initial rotation rates for the proto-earth. The R. Canup model is a slow rotating proto-earth spun up after the giant impact and Matija Ćuk and Sarah Stewart features a rapidly spinning proto-earth that slows down during an apparently tight window of time for the evection resonance after the impact. We also have the proto-earth continuing to accrete mass growing to its present size after the giant impact, something I did not see discussed. We also have variables related to the Moon forming after the giant impact and its present inclination to Earth’s equator. After the giant impact the proto-moon would continue to accrete mass to form its present size, the Moon we see today. The initial distance of the proto-moon from the proto-earth would be about 3 earth radii separation in both R. Canup and Ćuk/Setwart models vs. today’s mean of about 60 earth radii separation. It seems we have a large number of undisclosed variables that could dramatically alter the computer simulation output for these two giant impact scenarios when compared to the present angular momentum of the Earth-Moon system as well as the mass differences between the two bodies and orbital configuration (e.g. the Saros cycle for predicting lunar and solar eclipses). As the computer simulation research continues into the giant impact model for the origin of our Moon, it appears that more tight constraints are surfacing that will depend upon special conditions for various parameters in the computer simulations to avoid creating something quite different than what we see today in our night sky. The giant impact computer models we need to remember do not explain the rotation period of Mars or Venus.

  2. Bruce

    I follow everything in your comment Rod until I get to your last sentence. Since neither of our planetary neighbors have large moons, their formations would have progressed along differing lines, so no large impact Earth/Moon simulation should be expected to have anything to say about Mars or Venus, I would think. Am I missing something?

  3. Rod

    Bruce, good comment here. The giant impact model for the origin of our Moon is wrestling with the angular momentum observed in the Earth-Moon system. The model that features a rapid rotating proto-earth, apply this to proto-Mars or proto-Venus and see what happens with their angular momentum observed today and rotation periods. Same issue applies to a slow rotating proto-earth, apply to proto-Mars. Plenty of fun in the computer models I think :)

  4. Bruce

    I see, I think. You seem to be suggesting that the proto-planets in our region of the solar system should be expected to have had similar beginning rotation rates. I can see where there could have been similarities between Mars and Earth and between Venus and Mercury because of current rotational periods, but to have Mars, Earth AND Venus to have had the same rotation rates at some point seems very unlikely to me. It’s also an unnecessary requirement. The impact history that we can read on the surfaces of every solid body we can see shows that the planetary accretion process results in random collisions, with impactors apparently falling in from all directions over time. Some impacts would speed up rotation, others would slow it down, while some can even radically alter the obliquity; the equatorial inclination to the planet’s orbital plane. To my thinking the wide range of present conditions suggest that there would also have been a wide range of early conditions as well. I’m not trying to be argumentative Rod, just seeking a better understanding of this systemic creative process. I enjoy and welcome this interchange of ideas.

  5. Rod

    Bruce, some good points you raise. My main point-the computer models are using highly selected initial conditions which if applied to other bodies in the solar system results in quite a different solar system then we see today. Concerning the impact record observed today on Mercury, Venus, or Mars, this record has nothing to do with these bodies when they were proto-planets. The record of impacts we see today took place after the planets were already fully formed. Computer models do not show how dust grains can evolve into proto-planets or how a proto-earth just happens to stop growing at the right size and mass we have today so the giant impact model for the origin of the Moon does not create some strange configuration. I am confident as the computer models continue being tweaked here, more tight constraints and special conditions will emerge in these studies, enjoy.

  6. Bruce

    But I believe this does Perer, which is good to think about on this or any day (or night), taken from Psalms 136:1-9: “Give thanks to … the One making the heavens with understanding … to the One making the great lights … even the sun for domination by day … the moon and the stars for combined domination by night”

  7. Russ

    Rod and Bruce, interesting discussion. Most planetary rotations today are in the same direction as the sun and we can conclude that preservation of the angular momentum of the primordial cloud from which the solar system formed played a role. Loosely stated, the more massive the planet, the greater portion of angular momentum (of that cloud) is preserved by it today. The clear exceptions are Venus and Uranus. Impacts by much larger bodies than was ‘normal’ in the accretion process would be the best explanation for stopping Venus’ rotation and reversing that of Uranus. Then why did these planets not acquire large moons in the process?

    Robin Canup and other computer simulators would quickly and easily explain that only a tiny fraction of possible impact scenarios end in a binary planet such as the Earth/Moon system. Most possible impacts result in no satelite formation at all.

    The thing I like about Robin’s model is the much lower impact speed. Solar system formation was almost complete; orbits nearly circular. Bodies inhabiting the same neighbourhood would likely have low relative velocities.

    On the final angular momentum of the Earth/Moon system: it could have gone the other way depending on the collision parameters – if independently prograde before collision, could a retrograde impact yield a binary? Work needs to be done.

    On the rotation rates of Earth and Mars being similar: pure coincidence.