Discussion and news about the modern effort to understand the nature of life on Earth, finding planets around other stars, and the search for life elsewhere in the universe

Sunday, February 6, 2011


In the continued wake of the Kepler results that indicate a likely wealth of planets in our galaxy I thought I'd post a rather more personal note about an intimately related area that I think in many respects parallels where exoplanetary science was in the early 1990's.

Our solar system plays host to an extraordinary array of natural satellites, or moons. Many of these are entirely comparable in size, composition, and even chemical and geophysical activity to bona-fide planets. The only real difference is that these worlds reside deeper in the orbital hierarchy. Nine regular satellites in our system have diameters greater than 1500 km, the largest (Ganymede and Titan) are over 5000 km in diameter - larger than the planet Mercury. Io around Jupiter has extensive and active silicate and sulfur-rich volcanism. Titan has a frigid atmosphere that is somewhat denser than the Earth's, and a diverse and global hydrocarbon cycle from gas to liquid to solid. Many moons have signs of active and quiescent cryo-volcanism - from Enceladus to Triton and Europa. They also show good evidence for subsurface liquid water oceans that readily exceed the total volume of Earth's oceans. It is little wonder than many of the current concepts for future solar system exploration missions focus on these objects - they are tremendously interesting.

It is also true that our models of how moon systems form are even less well developed than our models of planet formation. It seems that moons around giant planets probably form out of circumplanetary disks of gas and dust much like a scaled-down version of planet formation itself, but there are many caveats. There's another sneaky truth; simulations of forming planetary systems are not typically set up in ways that allow us to track satellite formation, capture, or loss (embarrassed cough). In this sense we're even further behind than the equivalent situation for planets two decades ago.

Intriguingly though the prospects for detecting moons around exoplanets may not be too bad. It may even be on a par with the situation in 1994, on the cusp of the first radial velocity exoplanet discoveries. Lurking already in the bounty of Kepler data there could be evidence for exomoons as transit duration and transit timing variations. Moons make their planets wobble just as planets make their stars wobble by offsetting the system center-of-mass.

There are some new rules though. Stellar tides can be very bad for moons. The same forces that operate to eventually bring a planet into spin-synchronicity or tidal lock with a star also perturb satellite orbits and can pump their orbital ellipticity to a point where the moon just sails off. Additionally, once a planet becomes tidally-locked to its star then there are in fact no stable moon orbits and any such objects will over time spiral inwards due to moon-planet tides. The upshot of all this is that within about 0.6 astronomical units of a solar-mass star then in all but the youngest systems you might not expect to find any moons - assuming of course that they formed in the first place. So this recent Kepler data release of planets within about 0.5 AU of their stars may not be the ideal place to look. Kepler release 3.0 may be another story when we begin to confirm planets on longer orbits.

My own interest in exomoons was in part stimulated by what is perhaps the modern classic paper on the subject, by Williams, Kasting and Wade in 1997. By Jim Kasting's own admission the inspiration for this paper titled 'Habitable moons around extrasolar giant planets' came from a viewing of a certain episode of a certain sci-fi franchise depicting a place called Endor. It's a lovely paper. A key point in it was that gravitational tides in moon systems due to moon-moon interactions could be pivotal in dissipating enough energy to make up for a moon being well outside the classical habitable zone of a star. Instead of stellar heating you'd have more geophysical heating. In 2005 I attempted a bit of a followup of my own and with some funding from NASA made a small study of the potential for 'habitable' moons around the then known exoplanets. The idea was simple, we knew the stellar input for these planets and any moons they might have, so what kind of tidal forces would be needed to push them to temperatures that could sustain liquid surface water? I was surprised to find that it could all work out pretty well. Although there are several caveats then tidal heating in a plausible range could effectively double the size of the habitable zone in these systems if we were willing to consider moons as well as planets. The lovely thing about it all was that the energy for this all geophysical warmth came from the spin and (ultimately) orbital energy of the giant planet. Life powered by angular momentum? Perhaps so.

Five years later and I was sitting on a tediously long flight watching a movie about blue-skinned aliens romping around on a lush tropical moon orbiting a gas giant planet in the Alpha Centauri system. It occurred to me how funny it was that two epic Hollywood productions framed the interim works on exomoons, obviously we should listen to scriptwriters more often. It also occurred to me that exomoons might just be ready to fully emerge from the astrophysical subconscious. A few recent publications seem to have confirmed that.

We may talk about finding the first 'Earth-like' planet (once we figure out what that actually means). What if we're more likely to find an 'Earth-like' moon around an ice or gas giant? The odds quite conceivably favor such a situation. There may be a few million rocky planets in habitable zones in the galaxy, but there could be as many or more rocky, watery moons in the extended habitable zones around giant worlds. 

I'm not for a moment suggesting that we divert attention from hunting exoplanets. I also hope that some of the pioneers who devoted themselves prior to 1995 to what was seen as a fringe pursuit are the ones to find that Earth-twin, they deserve to. However, if we find barren world after barren world it will be time to turn our gaze on those strange and fantastic places that are held in thrall of giant planets.


kurt9 said...

We have this discussion on the Centauri Dreams website about 2-3 times per year.

The problem with gas giant moons is that the gas giants in our own solar system have rather nasty van allen radiation belts. The bigger the gas giant, the nastier the radiation belts. Jupiter's is so bad that humans probably can't even come near the place, let alone settle on any of the Galilean moons.

Another problem is the scaling in the size of the moon with the gas giant. Saturn has one large moon. Jupiter has four. However, Jupiter's 4 moons are not much larger than Saturn's one moon. This suggests an upper limit on the size of the moons that is significantly smaller than that of the Earth. Any moons orbiting a gas giant in the HZ will have to be Earth-sized in order to hang on to its atmosphere. Perhaps you need a gas giant that is comparable to a brown dwarf in order to have Earth-sized moons. Since Kepler has found mostly Neptune sized planets, is it likely that they could have Earth-sized moons? Perhaps, but probably unlikely.

The third problem is not really a problem, but would make any habitable moon very different from the Earth. This is that any such moon will be rotationally locked with its primary. Its day-night cycle will be the same as the orbital period around the gas giant. This could be anywhere from, say, 60 hours up to several weeks. This would not make for a very Earth-like climate. The day-night temperature swing would be quite impressive on such a moon.

Anonymous said...

All good points Kurt9 but surely its much too early for anything but idle (but very interesting) speculation?

For starters I dont think the radiation environment at Callisto is too horrible. Ive seen the BOTE calcualtions about the maximum likely size of exomoons but would suggest its much too early to make generalisations - even if large exomoons (greater than Mars mass?)are at the tail
of a distribution there still could be a hell of a lot of them out there.

Finally, although its understandable that our search for Hzs is based on liquid water and conditions that we understand in our current sample of one there might be all sorts of exotic processes (like Caleb's geoheating) that we might eventually need to consider.

We live in interesting times


Caleb Scharf said...

kurt9, thanks for the comments. All excellent points. I do have some counterpoints though. First, the issue of giant planet radiation belts. Giant planet magnetic field strength does seem to scale up with planet mass and so we'd expect exoplanets to have significant fields - but that does not necessarily imply showstopping radiation environments. In the Jovian system there are 2 particular issues that are important for moon environments. One is that a lot of the particle radiation/plasma trapped in Jupiter's magnetosphere originates from Io (in fact from all the major moons) - due to its volcanism and lack of atmosphere to trap ejected particles. Take away Io and the radiation issue would be lessened. A caveat in my discussion of tidally heated 'habitable' moons is that such objects probably need to be at least 0.1 Earth masses in order to have significant and sustained magnetic fields due to geodynamo action and to hold on to an atmosphere (see the Williams et al paper). As we see with Ganymede - the only moon with an intrinsic magnetic dipole - this magnetic field not only produces a protective cavity in Jupiter's magnetosphere, but would also dramatically reduce the erosion of a moon's atmosphere - which is the biggest hurdle to overcome in such an environment.

This is connected to the 2nd point. The total satellite masses of Jupiter and Saturn are about 0.0002-0.0003 times the planet mass. The Canup and Ward paper that I link to in the post ('form') indicates that this might be a universal ratio for a particular model of moon formation (admittedly very uncertain). In that case planets just a few times the mass of Jupiter certainly have the potential to have moons of masses equal to or greater than 0.1 Earth masses.

Finally, tidal-lock is actually (contrary to popular belief) not the only end-point for spin-orbit interaction. More likely in many cases is actually a pseudo-synchronous situation, with for example 2 moon spins per moon orbit etc etc. Week long moon orbits may occur for outer, irregular, moons but large moons are most likely to be regular and occupy inner orbits with periods in the ~10 day regime.
Unlike a planet tidally locked to a star the day/night terminator on a moon will (except for highly inclined planet spins) be continually, albeit slowly, moving. While it's certainly not Earth-like I think it's actually far less extreme than a tidally locked planet and thermal inertia in an atmosphere/ocean would help mitigate some of the variation. So as you say, this may not be a problem, but certainly different.

So, certainly issues for moons, but as a moon enthusiast I don't actually think these issues are worse than those facing 'normal' planets.

Dave Spiegel said...

Excellent post, excellent comments.

Just a couple points.

(1) Definitively diagnosing transit timing variations (TTVs) and transit duration variations (TDVs) as due to the presence of moons might be quite difficult. Kepler data will surely reveal TTVs and TDVs, but discerning that they are from moons and not, e.g., the influence of other planets, will not be trivial.

(2) Unless giant planets host multiple habitable moons per planet, probably most habitable "Earth-like" objects will be planets, not moons, simply because the mass function seems to increase so much toward lower masses.

Nevertheless, this is an interesting discussion, and we certainly do live in interesting times (and not in the Chinese curse sense of the phrase, at least not regarding exoplanet science).

Caleb Scharf said...

Thanks for the comments Dave. Great points. Yes, TTV/TDV measurements are going to be tough - also because estimates on sensitivity for moons (e.g. Kipping's work) are generally for rms values then I think dealing with limited data is going to be very challenging. My own feeling is that one may also have to factor in the additional photometric effects of any satellites, while small (tiny) they would both be a potential bias for timing measurements and a potential way to discriminate (I think Schneider et al looked into this).

Agreed on the frequency of moons. I was wondering whether the potential for an 'extended' HZ would mitigate some of the mass function skewness, but you're right, probably doesn't compensate so much.

Anonymous said...

Very cool.

Would Kepler be able to detect a double planet system, for example if two roughly Earth sized planets were orbiting around their center of gravity would Kepler notice that or just read it as a single large planet?

Caleb Scharf said...

Double planets - I think Kepler could certainly see this, although getting the right interpretation might take some time. It already found a system with 2 low mass stars in a 'double' in orbit around a 3rd star. However, the pathways by which planets form and how their orbital dynamics evolve probably make double Earth massed planets extremely unlikely.