The example of the Solar System demonstrates that moons are a natural outcome of planetary evolution. Six of the system’s eight planets host among them more than 150 moons. Each moon has one of three possible histories:

1. Formation in an accretion disk orbiting the host planet, in a process that recapitulates planetary accretion in the primordial Solar nebula. As various investigators have shown, this process occurs at the end of planet formation, when the disk-shaped cloud of gas surrounding the parent star has largely dispersed (Canup & Ward 2006, Alibert & Mousis 2007, Dobbs-Dixon et al. 2007). With the exception of the Earth’s Moon and Neptune's principal moon, Triton, all of the largest satellites in the Solar System were formed in this way. Sometimes known as co-formation, this process tends to yield co-planar moons traveling in prograde orbits of low eccentricity, like the Galilean moons of Jupiter. In a given satellite system, the aggregate mass of the final assemblage of moons will be equal to about 0.0001-0.0002 of the mass of the primary planet. This ratio is predicted by Canup & Ward’s evolutionary model and observed in the Solar System (Canup & Ward 2006). Individual moons will be still smaller. For example, Ganymede, the most massive moon in the Solar System, has only 0.000075 the mass of its host planet, Jupiter.
   
   
2. Capture from a field of planetesimals by the host planet. This process has plainly occurred in the satellite systems of all four giant planets, and it might be the origin of Deimos and Phobos, the two tiny satellites of Mars, and of Triton, the largest moon of Neptune. The outcome of such a mechanism is likely to be a satellite on an eccentric, highly inclined, or retrograde orbit. With the exception of Triton, all known or suspected captured moons are small and irregular like asteroids, rather than spherical like planets or co-formed moons (see Agnor & Hamilton 2006). Triton betrays its violent past through its retrograde, highly inclined orbit – which is nevertheless almost a perfect circle.
   
   
3. Formation through collision of two protoplanets and subsequent accretion in the resulting high-mass disk around the surviving planet. Often called the “giant impact” scenario, this was the origin of the Earth’s moon, as demonstrated in a number of recent simulations (Canup 2004). None of the other satellites in the Solar System had such a genesis. Formation through giant impact is likely to occur only at relatively small semimajor axes (less than 2 AU), since it requires a population of protoplanets traveling in fast orbits. The mass of Luna is extremely large in relation to the mass of the Earth, at more than 1% (0.0123). This large value is probably a result of Luna’s unique origin, and it may define the upper limit for the mass ratio of satellites to their host planets.

Notably, the Solar System’s innermost planets, Mercury and Venus, lack any satellites whatsoever, hinting that the formation (or at least the retention) of moons may be inhibited in the vicinity of the primary star. In fact, Barnes & O’Brien (2002) have calculated that, on account of stellar tides and other gravitational effects, massive moons cannot survive in orbit around planets of Sun-like stars if the host planet’s semimajor axis is less than 0.6 AU. The survival of moons may be compromised at even larger distances, as suggested by the example of Venus (semimajor axis = 0.72 AU).

If we apply these observational and theoretical constraints to the existing sample of exoplanets, we can sketch a very general picture of the population of exomoons that awaits discovery, pending the application of appropriate technologies and resources.

ORBIT   Exomoons become more likely and potentially more massive as the periastron of the host planet increases. According to Barnes & O’Brien, Mars-mass moons may be possible beyond 0.1 AU, while moons of any mass whatsoever are possible beyond 0.6 AU (Barnes & O’Brien 2002). Thus, the potential population of large exomoons is substantial, since well over 100 extrasolar planets have semimajor axes wider than 0.6 AU.

MASS   The mass of the host planet imposes far more stringent constraints on potential satellites than does its orbital separation. For co-formed satellites, Canup & Ward arrive at a mass ratio of 1:10,000 (Canup & Ward 2006). Thus the largest moons of a Jupiter-mass planet will be similar in mass to Ganymede, Titan, and Callisto. A planet of 4 MJUP (e.g., the nearby exoplanets 55 Cancri d and Upsilon Andromedae d) might host exomoons as massive as Mars (0.00033 MJUP). Alternatively, if the entire potential mass of the satellite system were concentrated in a single object, we might find exomoons two to four times as massive as Mars around such planets; this would place them at the lower limit to sustain plate tectonics over billions of years, estimated at 0.3 MEA (Raymond et al. 2006). Only the very largest and least numerous exoplanets, with masses of 10 MJUP or more, would seem capable of sustaining accretion disks in which an Earth-mass object could form.

COMPOSITION   The co-formation of satellites begins only with the dispersal of the gaseous component of the protoplanetary disk (Alibert & Mousis 2007). This is also the evolutionary phase when the inward spiral of giant planets due to Type II migration is complete (Alibert et al. 2005a) Since most known exoplanets orbit within the ice lines of their systems, in regions where primordial volatiles were far less available than refractory constituents, a substantial proportion of exomoons must also form in this region. We can expect such satellites to be rocky like Luna, Mercury, and Mars, rather than icy like Ganymede and Callisto. However, orderly migration is not the only mechanism that delivers giant planets to small semimajor axes. Planet-planet scattering can also result in orbital radii as small as 1 AU, and such a history is especially likely for planets currently following eccentric orbits. Scattering might also disrupt the satellite system of a gas giant, ejecting some moons entirely and replacing others with captured asteroids or protoplanets. Thus gas giants orbiting at 1 AU might have either rocky or icy moons, or both, depending on their evolutionary history.

CAPTURE   Any giant planet that arrives at a small semimajor axis, whether through migration or scattering, has the potential to capture rocky objects formed in this region. Many such protoplanets may have attained at least the mass of Mars. If the mass ratio of Luna to Earth is a valid guideline, such captured satellites can possess as much as 1% of the mass of their hosts. A giant planet would then require a minimum mass of only 0.3 MJUP (equivalent to Saturn) to retain an Earth-mass moon. Since the minimum mass for a habitable body is even smaller (0.3 MEA), a planet at the appropriate semimajor axis would need only 0.1 MJUP to host such a satellite. Most known exoplanets satisfy this condition.

Considerations of habitability are often among the primary motivations for speculation about exomoons. None of the more than 300 exoplanets so far detected is a likely abode of life. As a result, those with astrobiological inclinations are forced to investigate the possible existence of Earthlike planets (as yet unknown) or of Earthlike moons within known extrasolar planetary systems.

Unfortunately, the number of large gas giants traveling on relatively circular orbits within the habitable zones of their host systems is quite small. Perhaps the best known example of such a planet is HD 218185 b, which is almost 6 times as massive as Jupiter, and which follows an orbit of low eccentricity at a semimajor axis of 1.03 AU around a G5 star (Santos et al. 2001). Most other high-mass, long-period planets have far more eccentric orbits (see Exoplanet Populations and Super Jupiters). Ultimately we may find that Earthlike planets are more common in the Milky Way than Earthlike moons.

Last update September 2008