The Origin of Exoplanets

Article Updated: 24 Dec , 2015


We truly live in an amazing time for exoplanet research. It was only 18 years ago the first planet outside our solar system was discovered. Fifteen since the first confirmation of one around a main sequence star. Even more recently, direct images have begun to sprout up, as well as the first spectra of the atmospheres of such planets. So much data is becoming available, astronomers have even begun to be able to make inferences as to how these extra solar planets could have formed.

In general, there are two methods by which planets can form. The first is via coaccretion in which the star and the planet would form from gravitational collapse independently of one another, but in close enough proximity that their mutual gravity binds them together in orbit. The second, the method through which our solar system formed, is the disk method. In this, material from a thin disk around a proto-star collapses to form a planet. Each of these processes has a different set of parameters that may leave traces which could allow astronomers to uncover which method is dominant. A new paper from Helmut Abt of Kitt Peak National Observatory, looks at these characteristics and determines that, from our current sampling of exoplanets, our solar system may be an oddity.

The first parameter that distinguishes the two formation methods is that of eccentricity. To establish a baseline for comparison, Abt first plotted the distribution of eccentricities for 188 main-sequence binary stars and compared that to the same type of plot for the only known system to have formed via the disk method (our Solar System). This revealed that, while the majority of stars have orbits with low eccentricity, this percentage falls off slowly as the eccentricity increases. In our solar system, in which only one planet (Mercury) has an eccentricity greater than 0.2, the distribution falls off much more steeply. When Abt constructed the distribution for the 379 planets with known eccentricity, it was nearly identical to that for binary stars.

A similar plot was created for the semi major axis of binary stars and our solar system. Again, when this was plotted for the known extra solar planets the distribution was similar to that of binary star systems.

Abt also inspected the configuration of the systems. Star systems containing three stars generally contained a pair of stars in a tight binary orbit with a third in a much larger orbit. By comparing the ratios of such orbits, Abt quantified the orbital spacing. However, instead of simply comparing to the solar system, he considered the analogous situation of formation of stars around the central mass of the galaxy and built a similar distribution in this manner. In this case, the results were ambiguous; Both modes of formation produced similar results.

Lastly, Abt considered the amount of heavy elements in the more massive body. It is widely known that most extra-solar planets are found around metal-rich stars. While there’s no reason planets forming in a disk couldn’t be formed around high mass stars, having a metal-rich cloud from which to form stars and planets is a requirement for the coaccretion model because it tends to accelerate the collapse process, allowing giant planets to fully form before the cloud was dissipated as the star became active. Thus, the fact that the vast majority of extra-solar planets exist around metal-rich stars favors the coaccretion hypothesis.

Taken together, this provides four tests for formation models. In every case, current observations suggest that the majority of planets discovered thus far formed from coaccretion and not in a disc. However, Abt notes that this is most likely due to statistical biases imposed by the sensitivity limits of current instruments. As he notes, astronomers “do not yet have the radial velocity sensitivity to detect disk systems like the solar system, except for single large planets, like Jupiter at 5 AU.” As such, this view will likely change as new generations of instruments become available. Indeed, as instruments improve to the point that three dimensional mapping becomes available, and orbital inclinations can be directly observed, astronomers will be able to add another test to determine the modes of formation.

EDIT: Following some confusion and discussion in the comments, I wanted to add one further note. Keep in mind this is only the average of all systems currently known that looks like coaccreted systems. While there are undoubtedly some in there that did form from disks, their rarity in the current data makes them not stand out. Certainly, we know of at least one system that fits a strong test for the disk method. This recent discovery by Kepler, in which three planets have been observed transiting their host star demonstrates that all of these planets must lie in a disk which does not conform to expectations of independent condensation. As more systems like this are discovered, we expect that the distributions of the tests described above will become bimodal, having components that match each formation hypothesis.

16 Responses

  1. Jorge says:

    Allow me to nitpick.

    If you think there are only 8 planets in the Solar System, then there’s only one with an eccentricity greater than 0.2, as you wrote, but it’s not Pluto; it’s Mercury.

    If, on the contrary, you think dwarf planets are a subset of planets (as you should… ahem), then there’s more. Pluto, yes, but also Eris. Eris would then be the most eccentric of Solar System planets (but only while Sedna isn’t classified as a dwarf planet as well… for it will, sooner or later, and it would be hard to beat with its e=0.85)

  2. Torbjorn Larsson OM says:

    Jon, Jorge, the source of the discrepancy may be in the paper, where Abt somehow discuss the fit as if it was the observations, and the observations as if it was the fit: “The bottom panel shows the distribution … the remaining zeros show that the solar system has no eccentricities greater than 0.2? “.

    To the paper then: As a layman, this is the type of paper I wouldn’t ordinarily read, because it skips so many points that I wouldn’t know if they are feasible.

    Abt goes from formation models of brown dwarfs (accretionary objects) between 1-100 Jupiter masses to known brown dwarfs (still larger than 14 Jupiter masses or so) to systems, none known to be formed by accretion AFAIU. While there are many protoplanetary disks found out there, some with ongoing planet formation IIRC.

    Also, using eccentricities must be a source of uncertainty. Planet migration would likely introduce such AFAIU, even if remaining planetoids may dampen it down after. Our own system, with limited but still migration, is witness to that. But there is no mentioning of that.

    I haven’t read the paper thoroughly yet, but accepting it as valid, my main problem, again as a layman, is that he claims “most systems” will be aggregationary formed. While Kepler data suggest most planets will be terrestrials, which isn’t aggregationary formed what I know of. (Certainly Abt doesn’t give another impression, see above.)

    Somehow that discrepancy must be resolved. I suspect it can’t while retaining reasonable planet distributions. (Extreme case: Put all Jupiters and larger in one type of system, all terrestrials in the other. Now you will have “most systems” aggregationary, if they have ~ 1/10 as many planets as the other type. Hmm.)

  3. Dark Gnat says:

    I don’t think solar systems will fit neatly into our little catagories. There are millions of variables that we can’t even detect (drag, gravity interactions, collisions, etc that can make a system chaotic). It’s a wonder that our system is so neat and tidy.

  4. Jon Voisey says:

    Jorge: Good catch. I was curious as to whether or not the author was including Pluto as a planet or not. The data point on the graph lists a planet at eccentricity of what looks to be 0.25, which is Pluto’s eccentricity. At that point, I assumed he was including Pluto, listed it, and didn’t check further to see what Mercury’s eccentricity was even though I knew it was rather high too. On a more careful reading of the paper, the author clearly states “eight planets in the solar system”, which would indicate Pluto shouldn’t be included. In that case, the data point on the 4th panel of the author’s Fig 1 is incorrectly placed at 0.25 instead of 0.205. However, this doesn’t invalidate the point of the article. If anything, it would strengthen it as pulling the point more to lower eccentricities would make the fall off even steeper, making it even less like that of the observed systems.

    Torbjorn: Another good catch. Given the data point (whether incorrectly placed or not) flatly contradicts this, the statement you pulled out is just silly.

    As far as the discrepancy between this paper and the Kepler data, this study is only talking about the planets we’ve actually observed thus far. The point is that there likely are numerous systems out there that did form from the disk method, but we haven’t seen enough of them yet to get both representing themselves in the data.

  5. Jon Voisey says:

    After a bit more reflection, I see why the data point was apparently misplaced. The data was binned in ranges (ie, 0 – 0.1, 0.1 – 0.2, etc…). Hence the reason the data point was centered at 0.25 instead of 0.205 where I’d have expected it. What an annoying coincidence that that just happened to be the eccentricity of another (ex)planet.

  6. Jorge says:

    Well, truth be told, and without reading the paper itself, let me say that I find that comparing exoplanets with the whole Solar System planetary brotherhood is a bit like comparing onions with oranges. We don’t have extrasolar data points lower than some 1.4 Me (excluding pulsar planets). I think the currently known extrasolar population would be comparable with our 4 big boys and only with them. For now.

    When we start finding Earth-sized planets out there, then we can add Earth and Venus to the bunch, and later on Mercury and Mars could join the fun as well, but for now it feels odd.

  7. agmartin says:

    One difference between most exoplanet systems and the solar system is that Jupiter and the other planets avoided migrating inside the snowline. This may account for the difference in eccentricity as interactions with the smaller icy planetesimals dampens the eccentricities of the giant planets after they scattered off each other in the Nice model.

  8. Torbjorn Larsson OM says:

    Jon, right on on the not enough data yet, I didn’t realize that. (I’m running a nasty cold, and I seem to confuse easily, I suspect I don’t have the energy to think things through. It’s a cheap trip!)

    We have to wait for the number of terrestrials to be comparable to the number of gas giants in the database, when we would have a statistically fair sample as far as this question goes. Fair with the current knowledge that is, there may be confounders we don’t see yet.

    I see Jorge gives similar reasons, and AGMartin mentions confounders (which comes out as size of the system and placement of giants).

  9. Torbjorn Larsson OM says:

    “running a nasty cold” – having a nasty cold (as opposed to “running a fever”).

  10. Jon Voisey says:

    I’ve added an addendum to the bottom of the original article in hopes that it will prevent some confusion for anyone else that finds this.

  11. agmartin says:

    I remember a couple of planets that were found to be in retrograde orbits with respect to the stars rotation. That would fit easier into a co-accretion model than the disk model.

  12. IVAN3MAN_AT_LARGE says:


    You’re referring, of course, to the exoplanets WASP-17b and HAT-P-7b; the retrograde motion of both planets are hypothesized to be the result of either the gravitational interactions with other celestial bodies via the Kozai mechanism or a collision with another planet. Alternatively, it may be that the star itself flipped over early in their system’s formation due to interactions between the star’s magnetic field and the planet-forming disc.*

    *Source: Wikipedia — Retrograde motion.

  13. Jon Voisey says:

    IVAN: I can’t imagine a collision being able to create retrograde motion. It begs the qusetion, where would the hypothetical collider come from? It would have to be moving in a retrograde motion itself and with sufficiently high momentum to change the direction. If it too formed in a disk, it would be orbiting the same direction and thus would lack the opposite momentum to make it possible. If it came from outside the disk, the chances of collision drop, making it unlikely. Thus, for both reasons, I can’t see it happening.

  14. IVAN3MAN_AT_LARGE says:


    Hmm… on checking the source of that Wikipedia reference above, I found that the original New Scientist article, “Planet found orbiting its star backwards for first time“, states:

    It may have been thrown onto the strange path after a near-collision with another, as-yet-undetected, planet in the same system. “A near-collision with the right trajectories can make a gravitational slingshot that flings one of the planets into a retrograde orbit,” says team member Coel Hellier of Keele University. [Emphasis mine.]

    So, whoever wrote that Wikipedia entry must have missed that “near” prefix. As one of the established editors on Wikipedia, I will fix that little detail after posting this comment.

  15. IVAN3MAN_AT_LARGE says:

    Update: I’ve fixed it!

  16. IVAN3MAN_AT_LARGE says:


    Oh, and thanks for bringing that ‘little’ detail to my attention! 🙂

Comments are closed.