≡ Menu

A Gas Giant Ejected from our System?

Free-floating planets — planets moving through interstellar space without stars — may not be unusual. If solar systems in their epoch of formation go through chaotic periods when the orbits of their giant planets are affected by dynamical instability, then ejecting a gas giant from the system entirely is a plausible outcome. David Nesvorny (SwRI) has been studying the possibilities for such ejections in our Solar System, using computer simulations of the era when the system was no more than 600 million years old. Clues from the Kuiper Belt and the lunar cratering record had already suggested a scattering of giant planets and smaller bodies then.

An ejected planet makes sense. Studies of giant planets interacting with the protoplanetary disk show that they tend to migrate and wind up in a configuration where pairs of neighboring planets are locked in a mean motion resonance. Such a resonance occurs when two planets exert a regular, periodic gravitational influence on each other (there is a 2:3 resonance, for example, between Pluto and Neptune, with Pluto completing 2 solar orbits for every 3 of Neptune). Current work suggests that these resonant systems then become dynamically unstable once the gas of the protoplanetary disk disappears. Just how this happens is what the new work is all about.

From Nesvorny’s paper:

To stretch to the present, more relaxed state, the outer solar system most likely underwent a violent phase when planets scattered off of each other and acquired eccentric orbits… The system was subsequently stabilized by damping the excess orbital energy into the transplanetary disk, whose remains survived to this time in the Kuiper belt. Finally, as evidenced by dynamical structures observed in the present Kuiper belt, planets radially migrated to their current orbits by scattering planetesimals…

The scattering outlined here suggests that Jupiter moved inward in its orbit and scattered smaller bodies both outward and inward, some to take up residence in the Kuiper Belt, others to cause impacts on the inner planets and Earth’s Moon. The problem: A slow change to Jupiter’s orbit based on interaction with small bodies would have thoroughly disrupted the inner system.

A faster change of orbit due to interactions with Uranus and Neptune would have caused both the latter two planets to be ejected from the system. At this point Nesvorny added to the model an additional giant planet, working with an initial state where all the giant planets reached resonant orbits in the protoplanetary disk within a range of some 15 AU from the Sun. As the gas disk dispersed, Uranus and Neptune would have been scattered by the gas giants, reaching their current orbits and in turn scattering the planetesimals in that region into the Kuiper Belt.

The final consequence is the ejection of the fifth gas giant. An additional planet between Saturn and the ice giants — a world that was ultimately ejected from the system entirely — leaves us with a simulation that models the four giant planets we see today. Nesvorny’s simulations put a range of different masses for the planetesimal disk into play, with a total of 6000 scattering simulations following each system for 100 million years, when the planetesimal disk was depleted and planetary migration at an end. And it turns out that you are roughly ten times more likely to wind up with an analog to our Solar System if you start with five giant planets rather than four.

This scenario solves a variety of problems. Understanding how Uranus and Neptune formed is difficult because at their present distances of roughly 20 and 30 AU, accretion would have required too long a timescale. Nesvorny points out that the ice giants form readily at 15 AU or less, and the five planet resonant system his work discovered accounts for their movement outward. The results shift depending on whether one assumes an initial 3:2 resonance between Jupiter and Saturn or a 2:1 resonance, with the latter pushing the outer ice giant to a problematic 18-20 AU for formation. This makes the 3:2 resonance the most likely, but the scientist notes the need for more work on the question.

In the meantime, we’re left with the vision of an even more interesting early Solar System than we thought, and the possibility that the ejection of that fifth giant planet may be what spared the inner system — and our Earth — from complete disruption. We’re also given another look at the processes that produce dark, interstellar wanderers, planets with no star to light them, as Nesvorny notes:

“The possibility that the solar system had more than four giant planets initially, and ejected some, appears to be conceivable in view of the recent discovery of a large number of free-floating planets in interstellar space, indicating the planet ejection process could be a common occurrence.”

The paper is Nesvorny, “Young Solar System’s Fifth Giant Planet?” Preprint available.


Comments on this entry are closed.

  • Alex McLin November 11, 2011, 16:04

    If we gain confidence that Nesvorny’s model accurately describes the early Solar System would it be a reasonable possibility to determine where the missing gas giant went? Or even to locate it somewhere in the interstellar space?

  • Joe November 11, 2011, 16:56

    Certain isotopic signatures common to all bodies in our solar system but distinct from others (think of identifying interstellar dust grains) might help identify the lost planet. But this ejection would have occurred over four billion years ago. Even assuming the lost planet followed a galactic orbit similar to our own, that would make for a huge search space. The planet itself would be old and cold by now, practically invisible.

  • Astronist November 11, 2011, 17:35

    Don’t understand some of this. You say that resonant systems become dynamically unstable once the protoplanetary gas clears, but Neptune and Pluto are in such a resonance, and are clearlylong-term stable (perhaps an exceptional case?). Jupiter scatters smaller bodies outward, some of which “take up residence in the Kuiper Belt”, but that is impossible unless they can somehow circularise their orbits at aphelion. In general, surely gravitational scattering, because of its random nature, tends to put planets into eccentric orbits? But all the orbits of our Solar System’s giant planets are close to circular. Still, a fascinating hypothesis.

    Oxford, UK

  • LarryD November 11, 2011, 18:02

    It also depends on if “ejected” means “complete escape from the suns gravitational influence” or not. The Oort cloud has managed to follow the sun through its 21 orbits around the galactic center. And there is believed to be a sub-brown dwarf object in the Oort cloud, though that may more likely be a capture form one of our stellar nursery mates.

  • jkittle November 11, 2011, 20:41

    two thoughts
    First, this type of modeling should be able to produce and justify Bodes Law, which has always been a mystery.
    Second, does “ejection” mean reaching escape velocity for the solar system ( I doubt it, but I suppose it is possible) or just moving the planet to a higher and possibly eccentric orbit ( inner reach of orbit at 100 au for example, maybe the outer reaches of the orbit reaching , say 1000 AU ) this might account for the sparse population of Kuiper belt object past the “Kuiper cliff”
    at what point does the planet become susceptible to being dislodged by the passing of another star, ( it is likely that our Oort cloud has brushed the “oort cloud” of other star systems several times in the past.
    Lets’ find that missing planet!

  • Mark November 12, 2011, 0:37

    We need to get that planet back. If that planet should find itself on a trajectory to impact an inhabited planet in a remote corner of the galaxy, it’s not inconceivable that an advanced race could track it back (using methods which we can only guess at today with our primitive science) and determine that it came from our solar system. If they are a vengeful race, they may conclude that *we* are responsible for the ejection of that planet and imminent destruction of their world. The few very angry survivors of that doomed civilization may decide to wreak havoc upon us in retaliation. We need to get that planet back.

    Of course, returning this planet to its rightful place will totally upset the practice of Astrology. Small price to pay for setting the solar system right again.


  • Christopher Phoenix November 12, 2011, 1:24

    Interesting stuff!! It’s always interesting to see how scientific theory grows and changes since I first read about the evolution of the early solar system in my books.

    “The possibility that the solar system had more than four giant planets initially, and ejected some, appears to be conceivable in view of the recent discovery of a large number of free-floating planets in interstellar space, indicating the planet ejection process could be a common occurrence.”

    I’ve been wondering whether we might find a free-floating planet wandering near our solar system. How many such planets might exist? If there are many of them, perhaps one has wandered near our solar system.

    How might we detect a free-floating planet? This planet would be dark and cold, with now nearby sun to illuminate it. Most planet detection techniques could not detect it this planet- it isn’t near a sun, so techniques like Radial Velocity or the Transit Method. If this wandering planet passed very near our solar system, we could detect its gravitational effects on our solar system…

    If such a planet is discovered, might we mount a spaceflight to this wandering planet? It is possible microbial life could survive on a wandering planet or moon of a wandering planet by sheltering near volcanoes or black smokers under an ice sheet. This sort of world is an intriguing destination.

    A manned spaceflight to a wandering planet will be very similar to an interstellar journey. The astronauts will travel so far from Earth, the sun will appear to be just another star. I can imagine this being a precursor to an actual interstellar journey.

    Some science fiction stories have had wandering planets come near our solar system. These wandering planets cause trouble if they gravitationally disrupt the inner solar system or bring hostile alien civilizations with them. Sometimes, human space programs land on them and found colonies. I can see the appeal for science fiction writers- if the planets that are already in our solar system are becoming boring, than just add a new one!! Maybe this planet will become a candidate for colonization as it approaches the sun and warms up.

    It also depends on if “ejected” means “complete escape from the suns gravitational influence” or not. The Oort cloud has managed to follow the sun through its 21 orbits around the galactic center. And there is believed to be a sub-brown dwarf object in the Oort cloud, though that may more likely be a capture form one of our stellar nursery mates.

    That reminds me of the Cities in Flight novels written by James Blish. In the Cities in Flight novels, astronomers discover a trans-neptunian tenth planet that is christened Proserpina. Later, space missions are sent to Proserpina to found a base for exploration and scientific research. The U.S. space program had already explored and founded bases on the Moon, Mars, the moons of Jupiter, Titan (orbiting Saturn), and even further out. I suppose the astronauts were getting bored in the time between colonizing the solar system and inventing faster-than-light starship drives…

  • Martyn Fogg November 12, 2011, 4:16

    One possible scenario of many. Planetary accretion is a chaotic affair and I am skeptical of being able to recreate a detailed system formation scenario billions of years after the event. Nevertheless, it is a plausible idea. Models currently have great difficulty forming the giant planet cores — of sufficient mass to be able to accumulate gas — from a ‘minimum mass solar nebula’ starting point. That is, starting from a disk of solids that is already accounted for in the estimated amount of heavy elements currently within the giant planets. You have to boost the amount of mass available in the outer solar nebula in order to form the giant planets quickly enough before the gas component of the nebula disperses. Hence, ejection of mass into interstellar space becomes a necessity. Fortunately, models also show that once some planets grow to the size of Jupiter, they can be quite effective at slinging ‘failed cores’ out of the solar system.

  • Martyn Fogg November 12, 2011, 4:40

    To answer the problems posed by Astronist.

    (1) Orbital resonances can increase the stability of orbits or decrease it, depending on how the orbits are phased with respect to each other. In the Neptune-Pluto example you cite, each time Pluto returns to perihelion, Neptune is alternatively 50 degrees ahead or behind in its orbit. So the resonance is stable.
    In another example, if one imagines a 2:1 resonance in aligned orbits where conjunctions are at apocentre, this too tends to promote stability (closest approach is at the maximum possible distance), whereas conjunctions at pericentre do the opposite.

    (2) On the orbital circularization issue concerning planets scattered into non-hyperbolic orbits to the outer reaches of a system…
    This is thought to occur via “dynamical friction.” Gravitational interactions with smaller bodies can draw up the perihelion and circularise the orbit at a larger distance from that originally. This may have been what happened to Uranus and Neptune. Accretion seems to be too slow at their present remote locations and it is thought that they may have formed closer in, in the region where Saturn and Jupiter are now. As losers in the feeding frenzy for gas, they were slung out into a more unevolved planetesimal disk further from the sun. By scattering planetesimals in toward Jupiter, which then ejected them, Uranus and Neptune would have migrated outward, their orbits circularising during this process.

  • Astronist November 12, 2011, 13:37

    Martyn, thanks. (1) seems clear now, but still not sure about (2). But this sort of complex dynamics needs to be modelled on a computer (your speciality) as it’s not going to be very intuitive.

    Mark: a wonderful speculation, which in reality is impossible in so many ways! We aren’t sure that planet really existed or how big it was, and have no way of guessing in which direction it departed. If it did exist, the probability of its hitting another planet is infinitesimal: it would have to find a corridor a few thousand km wide at a range on the order of at least 10 light-years (or, if you are thinking of a “remote corner of the Galaxy”, then a range of up to 50,000 light-years). In the nature of things, it would have to make a number of close passes of heavy bodies before finally impacting something (because the probability of a flyby is so much greater than for a hit), and a couple of flybys would be enough to magnify the inevitable errors in tracking back its trajectory until its calculated origin became smudged over the entire Galaxy. Finally, any aliens capable of tracking it back to our Solar System would also be capable of working out that it was ejected before life ever evolved on Earth, and so retaliation would be pointless. Plus, we have no way of moving a planet-sized body back to our system, and in any case it has no rightful place here (if it did, Earth would probably not exist). Still, an amusing scenario, and instructive to contemplate.

    Oxford, UK

  • Abelard Lindsey November 12, 2011, 13:52

    Maybe this 5th gas giant is Nemesis.

  • Rob Henry November 12, 2011, 18:44

    Astronist, I agree that Mark’s details are all wrong but, to me, it is just possible that the outline may hold up.

    I could see that if life developed in Sol much earlier that is currently believed and was liberally transferred around by lithopanspermia, then this body may contain evidence of life on a system with our suns spectral signature. Alarmed and forewarned that they are not alone by the advent of this rouge planet passing close enough for them to find subtle evidence of this exceedingly rare instance of abiogenesis, the only ETI in our galaxy tracks down its origin with such haste that they have not yet decided how drastic their course of action should be when they find its origin.

  • djlactin November 13, 2011, 7:39

    Powers that eject a Neptune-sized planet would also work for lesser-sized objects (in spades!). I wonder (a) how many lesser bodies were also ejected [no doubt a huge number] and (b) whether the total mass of these objects could account for the “missing” dark mass?

  • A. A. Jackson November 13, 2011, 10:47

    I would have to see Nesvorny paper to see who he cites, but this is not a new idea.
    The whole question started with Pierre-Simon Laplace in 1776, tho the question can be traced from Newton through the celestial mechanics community until Laplace thought he had proved the stability of the solar system.
    When Henri Poincaré showed the restricted three body problem was chaotic (the word ‘chaos’ was not used in his work, but he was the first to have seen it and even recoil from it)…. no one knew what to think until advent of the digital computer.
    Then the Jack Wisdom led Digital Orrery project in the late 1980’s and Lascar finally showing that the inner solar system is chaotic in the future:

    J. Laskar (1994). “Large-scale chaos in the solar system”. Astronomy and Astrophysics 287: L9–L12

    That encouraged many orbital dynamic experts and solar system formation people to go back and look at the Planetary Migration problem during the late stages of making the solar system.
    Consequences of that were how to make the Oort cloud, Jupiter (and the Galactic Tide) mainly responsible there, the Kuiper Belt and Migration of Uranus and Neptune.
    It’s still not a settled dynamical process.
    One consequence of all these studies was that the Solar System may have had ,at one time, more than ‘8’ planets (sorry Pluto) , it’s not clear just how many may have been ejected.

  • Ole Burde November 13, 2011, 14:16

    If there are great numbers of these “lonely planets” it seems likely that a few of them would have achieved great velocities . This might be usefull for saving a lot of fuel for a spacecraft ,if one of these planets was travelling in more or the less the disired direction . Not only for the slingshot effect , but also for aero braking through its atmosphere .

  • Dave Moore November 14, 2011, 21:21

    It looks like from planetary formation models that ejected bodies are common. The average planetary system seems to eject one giant and probably several Mars sized bodies not to mention a blizzard of planetoids.

    If the average planetary system ejects one giant then there should be one free floating giant planet for every star in our galaxy, which would mean there’s a 50% chance there is a giant free floating planet closer than Alpha Centauri. Smaller free floating bodies passing through our Oort cloud must be a fairly common occurrence. Does anybody know offhand any papers providing estimates of this?

  • A. A. Jackson November 15, 2011, 9:08

    “Dave Moore”
    There was a paper in Nature , this year , estimating 400 billion of them!
    Here is the news article:


    I know that gravitational micro-lensing was proposed quite some time ago as a method of find them. I don’t know what the estimates of the numbers where then.
    At one the strange dynamics of galaxies had a proposal that galactic ‘free floating’ planets might be explained with them, but even at the number given about I don’t think they compete with ‘dark matter’.

  • Ronald November 15, 2011, 13:29

    @AA Jackson: no, indeed, even if there are 400 billion free floating gas giants in our MW galaxy, and ‘only’ 200 billion stars, and the average ff planet is Jupiter sized, and the average star is o.4 solar mass, then the total mass of these planets is only 0.05% of total stellar mass.

  • Ronald November 15, 2011, 13:30

    Oops, I meant 0.5%, half of one percent.

  • LarryD November 16, 2011, 15:29

    jkittle, your question “at what point does the planet become susceptible to being dislodged by the passing of another star” is basically asking “what is the radius of the Hill sphere (AKA Roche sphere) of our sun. But that can’t be answered without knowing the distance of the closest approach (and the mass) of the most massive perturbing star.

    In short, “it depends.”

  • Rob Henry November 19, 2011, 22:01

    Something in the way that that Nature article linked by A A Jackson treats its subject really grates me. Like all other treatments on the subject it assumes that the ratio of stars to planets near the core would not be drastically different than elsewhere – but how do you get around the following factor.

    The equilibrium average relative velocity of stars with respect to their neighbours is set by their own random perturbations of each other. Their current expected velocity is set by the history of those perturbing events, and for a lighter body the change of velocity is the greater in each encounter. The net effect should be to give the same distribution of velocities as that of gas particles, where the typical velocity of a particle with respect to the average motion of that system of particles scales inversely to the root of the mass of that particle. So Jupiter sized bodies should be moving about thirty times faster from the typical orbital velocity than other stars if they are confined there. But their not confined, and are thus free to migrate to the outskirts of the galaxy – so wouldn’t microlensing near the core give a vast underestimate of the number of such planets in our galaxy? I can’t see where I could have made a mistake here, but would appreciate it if anyone could see a flaw in my thinking.

  • kzb November 22, 2011, 8:40

    Rob Henry
    You’ve got to figure in the mean free path between significant interactions. Even in the galactic bulge this could be a long distance, maybe so long that a free planet could escape the bulge without any interactions, let alone the large number that would be required to “thermalise” the velocity distribution.

    Going against that, you need the initial ejection velocity to be great enough to escape the bulge gravitational potential.

    If your theory is true, there should be some selection effect on bulge stellar populations, not just planets. Small stars should be observed to have higher dispersion velocities, and perhaps be selected against versus high mass stars in dense environments. I’ve not seen anything that would back this up?

  • Rob Henry November 22, 2011, 15:45

    Kzb said
    “If your theory is true, there should be some selection effect on bulge stellar populations, not just planets. Small stars should be observed to have higher dispersion velocities, and perhaps be selected against versus high mass stars in dense environments. I’ve not seen anything that would back this up?”

    I agree completely kzb, but star formation rates are important also, so perhaps the analysis is too hard for Rv measurements except for the high visibility, and short life stars where this factor would overwhelm. I also suspect that the very complex models used to find the galactic equilibrium might obscure simple facts. In the disc things should be far simpler where you can just look at the z velocities (these do not get a chance to “cool off by migration”) and these should be apparent just from the distribution. Unfortunately these are obscured by the thin disc – thick disc situation, and its implications that there are too few interactions in these outer galactic suburbs to randomise star velocities in just a billion years or so (at least I think that’s the implication, but I should find out more about it).

    So how can I possibly think I’m right? It’s simple. I’m not imaginative enough to see how you can beat thermodynamics, but then you astronomers are awfully clever!

  • Rob Henry November 22, 2011, 18:43

    Kzb, I might have misunderstood your earlier paragraph. If “injection speed” means “typical hyperbolic excess velocity of the ejected planet from its system” then this should be only one or two kilometers per second if they are generated about Jupiter’s distance from their star. On the other hand if this means “initial velocity that such a planet may have due to the typical high velocity of a star with a protoplanetary disc in comparison to its neighbours” this might be 30 km/s.

    I understand the latter case as you would know typical galactic core “birth velocities” much better than me, but then such factors should equally apply to all stars born there and surely that would have a noticeable effect.

    If I take you concern to be due to the former case, then this would mean that most such ejections typically occur very very close to their star, in fact so close that we must hypothesise that the region of the closest in hot Jupiters (which according to Wikipedia are at 0.015 AU) are where the majority of such planets are ejected, just to make these velocities of equal importance to the typical relative stellar velocities of neighboring stars (and even here my relative stellar velocities are taken from our neighbourhood not the core).

  • Eniac November 22, 2011, 23:22

    @Rob: I like your theory. I once considered a similar one to explain cosmic rays. If you “thermalized” an ion against the stars, as could happen by electromagnetic field interactions, its energy would be truly astronomical. A little far out, I know, but who is to say?

  • kzb November 24, 2011, 8:35

    Rob Henry it’s very gratifying that you think me some kind of professional in this field -I assure you I am not !

    I don’t think that most ejections occur close to the star however. I think it is the reverse.

    And if you posit that ejection velocities are small, not large, you then must have a mechanism for the planet to escape the inner galaxy gravitational well. That is your thermalisation mechanism, but for that to operate there has to be a significant number of planet-star interactions over time. So you would need to prove that this is statistically plausible.

    I was looking at a recent paper the other day, about brown dwarf populations in a stellar cluster. The authors were at pains to show that neighbouring areas did not have an excess of BD’s compared to the cluster centre, showing that their population statistics for the cluster were valid. If lower mass members of clusters are selectively removed, then all these studies of the IMF in clusters are probably invalidated.

  • Rob Henry November 24, 2011, 18:06

    Very interesting kzb, but how old and dense are these open clusters being investigated. I know that they are so short lived that even O stars barely betray their origins through O associations. That would put the lifespan of clusters at << 10 million years. I will return to that fact soon.

    Your “mean free path” objection is a very good one, but let’s model the situation. There would be very few encounters between stars that would be so close as to produce significantly hyperbolic interactions. This makes the effective frequency of interactions with passing stars in the region of radius R directly proportional to R. Now the strength of these interactions falls off with R^2, but it effectively takes R times as much time to pass through the region of closest approach to these stars. Taking all these into account, stars at all distances make an equal contribution to the thermalisation process. Now it would be easy to calculate further, but there is a problem. We need the typical speed at which one star moves through the background of others, and this starts off much less than the equilibrium figure. This would accelerate the early history of this process by the factor by which neighbouring velocities are typically lower that their equilibrium values.

    Also, we must input the initial stellar densities, so how can we proceed? It’s simple! We know that open clusters last much less than 10 million years, and thermalisation is the only way to do this, so, at those densities, this process has time scales of a million years or less.

    All this gives me the feeling that those measurements of BD cluster compositions against their neighbouring values were to check that the cluster was very young, and thus not thermalised, and thus valid for survey purposes.

  • Rob Henry November 24, 2011, 21:06

    I now realise that above I should have stated that stars at all radii R less than the average separation for stars in that that region, should add an equal contribution to the thermalisation process. Of cause, perturbations by stars further out than that would be balanced by the effects of stars on the “other side” of the path of each star.

  • kzb November 28, 2011, 8:41

    OK, a better laboratory for the idea is globular clusters. They are old and a lot of them are dense. There is definitely a paper somewhere on the lifetime of delocalised planets in a certain globular cluster, because I have it somewhere!

  • Rob Henry November 28, 2011, 16:05

    Globular clusters form a perfect test here. If you can find a relevant paper on them please post a link or reference.