Stranger than fiction: The search for habitable exomoons


An artist's rendition of a habitable moon orbiting a gas-giant planet. / David A. Aguilar, Harvard-Smithsonian Center for Astrophysics

Nevermind the Ewoks. For astrobiologists, the best part of Return of the Jedi was probably the gas-giant planet Endor and its accompanying forest moon. This bizarre concept—a habitable, Earth-like world orbiting a massive planet like Jupiter or Saturn—has proved so captivating that it has inspired not only Avatar, the highest-grossing movie of all time, but also a canonical 1997 peer-reviewed research paper published in Nature.

Besides the idea's pure novelty, there are sound reasons for scientific interest in habitable "exomoons." The growing consensus is that after the Earth, the moons of giant planets like Jupiter and Saturn appear most likely to harbor some sort of life.

Europa has a vast liquid-water ocean beneath its icy crust that may be enriched with nutrients from the moon's rocky deep interior. Enceladus is a water-ice slushball that seems to have pockets of liquid water beneath its surface, which betray their presence in vaporous plumes jetting from the moon's southern polar regions. And Titan boasts not only a subsurface water ocean but also a thick atmosphere, complex organic chemistry, and a global methanological cycle that mirrors the aqueous rhythms of life on Earth. These are only the most notable and scrutinized potentially habitable moons in our solar system—there are others even more mysterious.

Even generously including Venus and Mars in the inner solar system's tally of habitable places, the outer solar system still offers more worlds where life could conceivably exist. When it comes to the search for alien life in our own backyard, moons are the next great frontier, even though they are quite different from the environments we're used to on Earth. And, according to David Kipping, an astronomer at University College London, moons may also be the next big thing in the search for life beyond the solar system.

Kipping has developed a way to look for habitable exomoons using NASA's Kepler mission, and says that we could conceivably discover one before finding another convincingly Earth-like exoplanet. It's even possible, he says, that the homeworlds for the majority of life in our galaxy more resemble Endor or Pandora than our familiar planet. As you read this, Kipping and his colleagues are sifting through the new Kepler data, looking for the telltale signatures of Earth-sized exomoons circling giant planets in the habitable zones of distant stars.

Is all this too outlandish to be true? Is Kipping on to something, or has he been watching too much Star Wars? I chatted with him about his ideas and research to give you the information you need to decide.

Lee Billings: How can someone detect an exomoon?

David Kipping: Detecting exomoons is very similar to how astronomers detect exoplanets. The oldest way to try to find exoplanets is known as astrometry, which is basically just looking at stars very carefully and watching to see if they periodically wiggle and wobble about on the sky. If a star exhibits that periodic motion, that's an indication that there's a planet in orbit and tugging on the star. To look for a moon, we do the same thing, but we look at a planet instead of a star.

The way we can measure these wobbles is through the transits that we see when planets are fortuitously aligned with our line of sight. A transit lets us measure both the position and the velocity of a planet at that moment in time. So by repeating this measurement over and over, we can see if the planet's position or velocity is changing. If it is, then that means something is tugging at it, which could be a moon or an unseen perturbing planet.

LB: I'm guessing you can tell the difference?

DK: Oh yeah. There are two dovetailing measurements, transit-timing variation (TTV) and transit-duration variation (TDV). TTV just measures the instant that a transit occurs, which means it's very sensitive to the position of the transiting planet in its orbit. TDV, on the other hand, measures the length of the transit, which can give you the planet's velocity. If the planet is moving faster, its transit duration will be shorter, and vice versa. For moons, the velocity shift is always out of phase with the position shift. So we look for that phase shift by measuring both TTV and TDV.

It's just like a swinging pendulum: When the pendulum hits its lowest point, it's moving at its maximum velocity, and when it swings to the top, its highest position, it has a momentary velocity of zero. If we detect that pendulum-like short-period phase shift for a transiting planet, we know it's definitely caused by a moon and not something else.

LB: How did you come up with this idea?

DK: I had this idea in 2008, when I was working out models for the shapes of the light curves we might see from Kepler. I was studying this very carefully, thinking about the different properties we could measure for a transit and how reliable they were. And I realized you could see TTV without knowing necessarily what was causing it. I started imagining the Earth going around the Sun as if it were a transiting planet seen from far away, and it struck me that our Moon would have a big effect there, that the Moon's orbit would periodically pull our planet around and change its velocity, therefore changing the transit's duration. That led to the two papers I wrote where I proposed how to find exomoons.

LB: How important is Kepler for looking for exomoons?

DK: Kepler is the most precise instrument we have for this at the moment. I worked on a feasibility study for Kepler, to see how well it could do, how small of a moon we could detect. And we found that in the best-case scenario we can detect habitable-zone Earth-like moons down to about a fifth of the Earth's mass. A more realistic scenario might be targeting exomoons of about one Earth-mass. In that case, there are about 25,000 stars in Kepler's field of view that are bright enough to give us sufficient signal-to-noise so that we can look for such objects. If these sorts of large, Earth-like moons exist, we should be able to find some of them in the next year or two with Kepler.

LB: How do you assemble a target list for this? What assumptions do you make?

DK: We have to think very carefully about selecting targets. There are two competing tensions: Whether or not a moon could be present in the first place, and how easy a moon would be to detect. In terms of detecting a moon, you want shorter orbital periods that give you more transits per unit time, which allows you to see more events and build up better statistics. You also want puffy, low-density planets, planets that cast a big shadow but aren't too heavy. That combination would give you a big transit and also a bigger signal from any accompanying moon.

But in terms of a moon's probability of existence, short-period planets aren't where you'd prefer to look. For example, "hot Jupiters" are just no good; we wouldn't expect moons to be around them for dynamical reasons. It's like Mercury or Venus in our own solar system—we think they were too close to the Sun and its gravitational influence to keep any moons they may have had.

Planets with longer-period orbits, wider separations from their stars, should have better chances of harboring moons. We already know from radial-velocity searches that there are plenty of Jupiter- or Saturn-mass planets in the habitable zones of stars. These sorts of planets are out there. We just need to look at the ones that transit and try to find some moons.

For the new Kepler data release, the sweet spot would be big, low-density planets around lower-mass stars, K- or M-dwarfs, that have already displayed several transits. You'd get the enhanced perturbation from a big moon and you'd be getting enough data to feasibly work with, but since the stars don't have much mass, their gravitational influence is less, and the moons are less likely to have been ripped away before you can detect them. That's where we might find an exomoon in the near term.

It's looking good. There's a new interesting paper from Eric Ford, where he and his team have been doing transit-timing on all the new Kepler candidates. And they found that at least 12 percent of all planetary candidates for which transit-timing could reliably be detected are actually showing TTV. Which means that these candidates are being perturbed by something. Whether that's another planet, or a moon, we have to do more work to be sure.

LB: You're making it sound easy. It's not, right?

DK: There are lots of things that make this difficult. We prefer single, isolated candidates rather than ones in multi-planet systems, like the packed planets in Kepler-11, for example. The planet-planet gravitational interactions there make subsequent analysis harder, because you have to delete all those timing changes in search of any from a moon. We try to avoid candidates with eccentric orbits, because that can indicate disastrous early history filled with planet-scattering events that moons might not survive. Then there are variations in the co-planarity of the planet-moon system—how much they orbit in the same, shared plane can complicate your analysis. Some configurations will make it easier to discern the moon's gravitational tug, and others will hide it.

Analyzing these things takes time. A single candidate can take a month or more. We have to clean the data of spurious signals from cosmic-ray hits on Kepler's detectors and process it a few times to make sure the signals are robust. Then we generate a synthetic data set that doesn't have a moon in it, for example, and make sure we recover a null result. We have to do a lot of self-checking before we're sure anything we're seeing is real.

LB: Okay. So let's say you find a potentially habitable exomoon or two in the Kepler data. What then?

DK: Well, for most of the Earth-size planets we'll find in Kepler's field, it will be hard to really pin down their mass with radial-velocity measurements because their stars are so faint. It's not impossible but it's hard. And without their mass you don't know their density, which means we don't really know what these planets are made of. Moons, by contrast, automatically give you their mass and thus their density because they are detected through dynamic gravitational effects. So it's straightforward to say whether a moon is rocky or icy or gassy, whatever. That's important for quantifying habitability. And what it means is, we might learn more about small rocky bodies from studying Kepler's exomoons than from studying its exoplanets.

Unfortunately, as I said, almost all of Kepler's stars are quite faint, so it probably won't be feasible to look for atmospheric biosignatures like oxygen or methane or anything like that for Kepler's planets or moons. But if Kepler shows that plenty of moons are out there, the next-generation of transit surveys would be expected to find a certain fraction of moons orbiting transiting planets around nearby, bright stars. If a nearby exomoon has an orbital separation from its planet like the Earth and its Moon, it would be feasible to follow-up and look for biosignatures in that case using something like the James Webb Space Telescope.

LB: Right. Getting back to how you came up with this idea, imagining our solar system seen from afar, what planet-moon systems here could an alien astronomer see?

DK: The big two are the Earth-Moon system and Neptune's moon, Triton. For something like the Galilean moons of Jupiter, and even Saturn's moon Titan, the mass ratios are too small for us to currently detect, even when you add all the moons together. Kepler does have the photometric precision to detect Saturn's rings, though, and I think we'll find some ringed worlds with Kepler soon. But for the sorts of moons Jupiter and Saturn have, which are formed from the gaseous disk around their planets early in our solar system's history, that process seems to yield unfavorable planet-moon mass ratios.

Earth's moon is different. It's irregular. It formed from a Mars-sized impactor throwing debris around the primordial Earth. Triton is probably the remnant of a binary object that wandered too close to Neptune, where one of the objects got captured and the other got ejected from the solar system. For irregular cases like these, there's not really a mass limit, and you can imagine all sorts of odd configurations happening. The question is, are these moon-forming events rare or are they common? If they are common, I'm very hopeful that we'll find several exomoons with Kepler. But if it happens that the vast majority of moons form like the Galilean moons did, I think we'll need to wait for the next generation of space-based telescopes to find them.

The real point is, we still don't know how common rocky planets are in the habitable zone, let alone moons. It could conceivably be that there are more Earth-like moons in the universe than there are Earth-like planets, which, if true, gives a pretty different picture of what life might typically be like across the universe. This is so fascinating because no one has ever done it before. If we find an exomoon, that would open up an entirely new field of astronomy, just like the first detection of an exoplanet did.