≡ Menu

Exomoons: Rare in Inner Stellar Systems?

Exomoons — moons around planets in other star systems — are an exhilarating and at the same time seemingly inevitable prospect. There is little reason to assume our Solar System is unique in its menagerie of moons, with the gas giants favoring us particularly with interesting mission targets, and then there’s that fascinating double system at Pluto/Charon. If we visualize what we expect to find in any given stellar system, surely moons are part of the mix, and investigations like the Hunt for Exomoons with Kepler will doubtless find them.

An actual exomoon detection would be a triumph for exoplanet science, especially given how recently it was that we nailed down the first confirmed exoplanet, 51 Pegasi b, in 1995 (or, if you prefer, the 1992 detection of terrestrial-mass planets orbiting the pulsar PSR B1257+12). We’re new at this, and what huge strides we’ve made! Given the small size of the transit signal and its changing relation to the body it orbits, exomoons offer a particularly difficult challenge, although David Kipping’s team at HEK has plenty of Kepler data to work with.

Image: A star with a transiting planet and its moon. The angled area shows the inclination of the moon orbit. Orbit positions beyond the dashed line are not undergoing transit, and are thus not observable. Credit: Michael Hippke.

With all this in mind, every paper that comes out of HEK gets my attention. Kipping (Columbia University), working with graduate student Alex Teachey and citizen scientist Allan Schmitt, has now produced a paper that takes a significant step as the investigation proceeds. We have no detection yet — more about that in a moment — but we do have a broader result showing that exomoons are unusual in the inner regions of the systems surveyed.

Kipping and Teachey looked at 284 viable moon-hosting Kepler planetary candidates to search for moons around planets from Earth to Jupiter in size and distances from their stars of 0.1 to 1 AU. This finding seems to be getting less attention in the press than it deserves, so let’s dig into the paper on it:

Our results place new upper limits on the exomoon population for planets orbiting within about 1 AU of their host star, upper limits that are remarkably low. We have also analyzed subsets of the ensemble to test the effect of various data cuts, and we have identified the regime in which the OSE model presented in Heller (2014) breaks down, which we call the “Callisto Effect” — beyond 20 planetary radii, discrepancies appear in the results.

OSE stands for Orbital Sampling Effect, developed by René Heller in 2014 and described by Michael Hippke in Exomoons: A Data Search for the Orbital Sampling Effect and the Scatter Peak. OSE stacks multiple planet transits to search for an exomoon signature. What the paper is referring to as the ‘Callisto effect’ is the disagreement between OSE predictions and moons like Callisto. Even so, the authors continue to see OSE as a useful tool, and learning about an area in which it breaks down is helpful as we fine-tune our capabilities.

Back to the paper:

Our analysis suggests that exomoons may be quite rare around planets at small semi-major axes, a finding that supports theoretical work suggesting moons may be lost as planets migrate inward. On the other hand, if the dearth of exomoons can be read as a reliable indicator of migration, our results suggest a large fraction of the planets in the ensemble have migrated to their present location.

And that is a pointer to which we need to pay attention. Is a lack of exomoons a marker for planetary migration? If further analysis determines that it is, then we’ve found an extremely handy tool for studying the formation history of other stellar systems.

The Kepler data did yield one exomoon candidate in the Kepler-1625 system for which the authors have set up plans for follow-up observations with Hubble this fall. There is no way to know at this point whether we’ve got a genuine exomoon here or not. And I much appreciate the thorough job that Alex Teachey did in getting this point across to the public in his article Are Astronomers on the Verge of Finding an Exomoon? We learn here that the authors put their paper online earlier than intended because a media outlet was going to publish news about the upcoming Hubble study (Hubble proposals are publicly posted online).

And Teachey’s point is sound at a time when ideas whip around the Internet at lightspeed:

Peer review is a critical part of the scientific process, and we are not terribly comfortable putting out our results before they have been examined by a qualified referee. Unfortunately, we feel the circumstances have forced us to make our results freely available to the public before such a review, so that everyone may see for themselves what we are claiming and what we are not. While David and I are both big proponents of engaging with the public and boosting interest in the incredible things happening every day in astronomy, we have serious concerns about the potential for sensational headlines misleading the public into thinking a discovery has been made when it is really too early to say that for sure.

It’s a solid point. But I also want to emphasize that this paper’s findings about the apparent rarity of exomoons in the inner systems of the stars being studied is quite significant. To my knowledge this is the first time we’ve developed a constraint on exomoon formation. We doubless have moons hiding in the data (recall that the authors are looking for analogs to the Galilean moons of Jupiter), and we can also suspect they are going to be much more common in outer stellar systems, which is certainly the case in our own Solar System.

Don’t expect an immediate result from the Hubble observations. According to this article in Nature, Kipping and team will take about six months to analyze the work before making any announcements. Steady, painstaking effort is how this job gets done.

The paper is Teachey, Kipping & Schmitt, “HEK VI: On the Dearth of Galilean Analogs in Kepler and the Exomoon Candidate Kepler-1625b I,” submitted to AAS journals and available as a preprint. For helpful background, check Kipping, “The Transits of Extrasolar Planets with Moons,” PhD thesis, University College London (March 14, 2011), available online.


Comments on this entry are closed.

  • Marshall Eubanks August 1, 2017, 10:38

    “Is a lack of exomoons a marker for planetary migration?”

    I would say, no. Basically, the argument is that the Hill spheres will shrink during migration, and thus moons would be lost. However, if these planets formed in situ, close to their stars, then they would already have the small Hill spheres, and, thus, no large exomoons.

    • Ashley Baldwin August 1, 2017, 16:03

      Ultimately it may depend on the environment of nascent stellar systems and the extent and nature of their circumstellar disks.

      The equation for the radius of sphere of planetary gravitational stability , the erstwhile Hill radius , r= a ( orbital semi major axis assuming no eccentricity ) X the cube root ( m/3M- where m is the mass of a moon and M the planet mass ) . This would atleast theoretically allow a small moon for close in Hot Jupiters or even smaller planets . However the question would be what and where the material any moon formed from comes from close into a star ( especially one with an active Pre- main sequence phase) . Unless as is postulated in this speculative article it is captured from a binary planetary system as an irregular moon or as a result of a collision as apparently created our own Moon.

      In terms of our own system , it ( to date ) is atypical in not possessing any Super Earths . So it’s possible that the currebt inner , small terrestrial planets formed after an earlier primordial population of such planets was shunted into the Sun possibly by a peripatetic Jupiter. This would fit with a planetary migration hypothesis too.

      Basically although the Kepler et al data has hinted at trends , it’s still far complete and not fully representative of planetary system architectures . Ironically the same precision transit work that may help confirm the presence of a moon in this case, will allow increased detection of increasingly small exoplanets too . In combination with more sensitive Doppler spectroscopic searches ( like 100 Earths ) and astrometry ( Gaia and hopefully Theia ) we should have a much more complete dataset within a decade or so, by which time exomoons will likely have been discovered frequently if not in abundance.

    • J. Jason Wentworth August 2, 2017, 8:41

      That was my thought, too. Our experience with orbiters–even close-in ones–around Mercury and Venus demonstrates how the gravity of the nearby Sun makes their orbits short-lived (they would be even shorter if the spacecraft didn’t periodically raise the periapses of their orbits using bursts of rocket power, something that no natural satellite could do), and:

      Perhaps some of the impact craters on both planets (maybe the Caloris Basin on Mercury is one of them) were formed by long-“decayed” moons (or their building blocks, which never had a chance to coalesce into one or more larger satellites) that fell down to the planets’ surfaces, as the various Mercury and Venus orbiters finally did after their propellants were exhausted?

  • DJ Kaplan August 1, 2017, 11:43

    I’d say that our science of exo-planet observation is still in its infancy. It’s best not to make any hard and fast conclusions yet.
    In other news, looks like complex chemistry at Titan is being confirmed.

  • Alex Tolley August 1, 2017, 11:52

    Are we to infer that Venus and Mercury, possibly Mars, also migrated, whilst Earth didn’t? I thought the density distributions were indicative of in situ planetary formation.

    • Rob Flores August 1, 2017, 12:54

      Indeed, but does that mean that a Large Moon was
      a very long shot for terrestrial inner planets solar system.

      If our moon was formed from a case of very EXACT conditions
      Then at most some exo-planets will have Mars type moons, small relics, full blown gravitationally compressed spheres.

      • Ashley Baldwin August 6, 2017, 15:44

        Good point. The short of it is that there just isn’t enough information for any simulations , based on what we know about the early solar system ,with or without additional data from exoplanet architecture. Think how inaccurate the early work was on the nature of exoplanets before Hot Jupiters began to turn up .

        Processing power may have improved by orders of magnitude but the software programmes have moved on much less so and can only be based on what is known plus yet more speculation . Just a few thousand exoplanets sounds a lot but is only scratching the surface and is far from a complete sample , most especially for smaller Earth like terrestrial bodies as well as more distant planets ( both Doppler spectroscopy and transit photometry still heavily favour big planets orbiting close to smaller stars ) . There is a long way to go in representative data capture before any solar system simulations have enough input to give a meaningful output.

        Even allowing for various connotations of gas/ice giant migration AND inner solar system collisions, few if any current simulations come up with an architecture that remotely matches the reality seen both in terms of numbers of planets, mass , orbital position and presence or absence of moons . Via binary capture, accretion or collision. For our own system let alone any alien one.

        Give it a decade or so and TESS, WFIRST’s microlensing survey , Gaia gas giant astrometry and most especially PLATO’s extended observation time series photometry -on top of Kepler and more sophisticated RV searches – and the picture will be far better representative as well as likely possessing the precision to have found a few larger exomoons too.

  • Haxo Angmark August 1, 2017, 17:58

    and please let us know when someone discovers a planet actually “migrating inward”. The reason planets in close orbits around red dwarves (dwarfs?) don’t have moons is probably because their sun gobbles them up. Red dwarfs (dwarves?) are that way. Ill-behaved. (Spelled both ways to avoid offending one.)

    • Ashley Baldwin August 3, 2017, 13:25

      “Dwarves ” ( and the related adjective “dwarvish” ) was a bespoke narrative term employed by JRR Tolkien ( forcefully defined in his forward to a later edition of his “Lord of the Rings” in response to early uninformed editorial “corrections” to the spelling ) to describe the plural of a fictional , distinct race of “dwarf” hominids . This as distinct from the conventional English term “dwarfs ” ( and presumably dwarfish!) , traditionally representing the plural of small variants of Modern man.

  • James Franklin August 2, 2017, 7:38

    I do not believe the lack of planetary satellites orbiting their parent at less that 1 AU is evidence of planetary migration toward the parent star. Most stars we are looking at (less than 3 solar masses) will have passed through the T-Tauri phase prior to becoming a Main Sequence star of F,G,K,M spectral classes.

    We know that T-Tauri stars are highly variable, have abundant and rampant stellar winds as the star collapses under gravity to the point where it “ignites”. There has been a host of research on these stellar babies, we know that their stellar winds vary from many tens to several thousands of times the volume of our own stellar wind, and perhaps this is the controlling factor.

    Stars with low stellar winds perhaps can have decent sized planets close in that can either form natural satellites or perhaps capture proto-planets in the early days and thus hang on to them. If the star has a more violent stellar wind this then blows material out of the way faster and allows only smaller, rocky planets to form close by and devoid of natural satellites. This sort of make a degree of sense. Mars has captures two asteroids that in a few hundred thousand years will impact the planet due to orbital instability. Mercury has no natural satellite, Venus has no natural satellite, then we have Earth. Depending on your thoughts on the origin of the Moon it seems likely that the Moon is a fluke, either formed separately and captured or, as the darling theory is at this time, formed following a major impact during the early days or perhaps in the late heavy bombardment period.

    Natural satellites are very common, stars have them, either as stars, brown dwarves, planets or smaller minor bodies, planets have them, asteroid have them, and it seems even some comments have them, hell, even galaxies have them, it seems highly probable that where nature can form them, it will, where conditions and circumstances do not allow it, they won’t form, or last.

  • Michael Fidler August 2, 2017, 10:39

    What if the Earth was originally a binary system that eventual merged or collided creating the moon? Take a look at Mars, one whole hemisphere is from an impact or merger. Many other moons show signs of mergers so could the early solsr system have had many binary planets and moons that merged early in the solar system history. The three body problem would tend to cause these mergers and the reason we would see so many asteroids with moons. Just think if the binay earths had a water world and a much higher density rocky world, make for some very interesting merging dynamics!

  • andy August 2, 2017, 14:32

    The claimed satellite detection at Kepler-1625 is a rather unexpected configuration: a Neptune-sized satellite orbiting a superjovian. This is a substantially larger satellite than previous models indicated would be likely (though it may be possible to form such a system via impact capture). Given that, I wouldn’t be too surprised if it did turn out to be something other than an exomoon system.

    On the other hand, it wouldn’t be surprising to me if the first exomoon did turn out to be an extreme system: I’m not sure where the candidate sits in terms of ease of detection but I’d guess that the large size helps. The initial exoplanet detections were after all examples of rare configurations: in 25 years of exoplanet discoveries, no further systems resembling the PSR B1257+12 planets have been found despite the sensitivity of pulsar timing as an exoplanet detection method, while hot Jupiters like 51 Peg b are now known to be a fairly rare type of planet that happen to be overrepresented in the planet catalogues by virtue of their ease of discovery.

    • Ronald August 7, 2017, 10:30

      My question here is: is there any reasonable chance of a real double-planet (binary planet) system developing in situ?
      Meaning that the mass of both planets is too similar to be considered a planet-moon system.
      Are there any models known showing this?

  • Ashley Baldwin August 3, 2017, 13:37

    Nature likes binaries .

    I’m hoping that it is indeed a case of binary planet capture as if a Neptune mass body can be captured , so could an Earth mass terrestrial planet . Rene Heller’s models struggle to show accretion development of even Mars sized moons around Super Jovians ( though moon mass scales with increasing planetary mass) . So if there is to be Earth like , habitable exomoons then binary capture is likely to represent the best bet. Ideally by smaller Neptune mass planets to reduce closing velocities and resultant planetary disruption.

    In terms of alternative theories , could the Super Jovian in fact simply be a small brown dwarf with attendant planet ?

    • Michael August 4, 2017, 2:55

      I think as we move into super Jupiter masses there is an inclination towards larger rocky worlds. The heat of formation of the super Jupiter’s would tend to cause material to pile up at a certain distance and that material will most likely be more rock than ices.

    • Rafik Bourne August 5, 2017, 8:40

      I think it more likely that, if confirmed by Hubble, this is actually a binary system with a small Brown Dwarf orbited by a Neptune size planet.

      Formation as a separate stellar system followed by migration appears a more logical progenitor than a “novel” capture or “grazing impact” process.

      The lack of “Jovian” type moons in inner systems should not be unexpected. All major moons of the outer system planets, except Io and Europa, would undergo sublimation as their parent planet migrates inside the “snow” line.

      Sublimation of moons in other systems would result in either complete or major loss of mass for all moons migrating from beyond the “snow line” with a significant water content. Hill sphere limitations are a secondary problem if the moon survives sublimation.

      • andy August 5, 2017, 19:18

        If it is a brown dwarf (i.e. low-mass tail of the star formation process), then this is by a substantial factor the tightest known binary system to host an S-type planet. Plus the mass ratio is quite high: scaling down the TRAPPIST-1 system to a 10 Jupiter mass host gives less than an Earth mass of material, and inward migration would tend to eject planets.

        Capture of large moons on the other hand is not an entirely novel scenario, given that our solar system contains the example of Triton, though the preferred mechanism in Triton’s case is capture of a binary KBO. Nevertheless, it would be amusing if the first Triton-type (captured) exomoon turns out to be more like Triton’s host planet.

    • Ronald August 7, 2017, 10:40

      See my comment with question directly above (under andy’s comment): do any models allow for in situ *binary planet* formation, i.e. not a planet-moon mass difference, but something much more similar in mass, and well outside the Hill sphere.
      Probably not by accretion (pity), maybe as a result of planetary impact?
      If so, the number of habitable earth-size planets (around Neptunes in compact systems) could be much greater.

      • Michael August 7, 2017, 15:59

        The larger a planet gets mass wise the greater the amount of gas collected. When large amounts of gas collide they explosively expand more than likely preventing reforming of another planet in orbit. Perhaps two formed planets pushed into orbiting is possible but in order for that to happen momentum must be shed at the right time which seems unlikely.

  • ljk August 7, 2017, 22:12

    An artificial eclipse for imaging extrasolar planets

    August 7, 2017

    by Taylor Kubota

    In our hunt for Earth-like planets and extraterrestrial life, we’ve found thousands of exoplanets orbiting stars other than our sun. The caveat is that most of these planets have been detected using indirect methods. Similar to how a person can’t look at anything too close to the sun, current telescopes can’t observe potential Earth-like planets because they are too close to the stars they orbit, which are about 10 billion times brighter than the planets that surround them.

    A possible solution might be to create an artificial solar eclipse with two precisely positioned spacecraft, according to Simone D’Amico, assistant professor of aeronautics and astronautics at Stanford and director of the Space Rendezvous Laboratory.

    One craft – known as a starshade – would position itself like the moon in a solar eclipse, blocking out the light of a distant star, so a second spacecraft with a telescope could view the nearby exoplanets from within the shadow cast by the starshade.

    “With indirect measurements, you can detect objects near a star and figure out their orbit period and distance from the star,” said D’Amico, whose lab is working on this eclipsing system. “This is all important information, but with direct observation you could characterize the chemical composition of the planet and potentially observe signs of biological activity – life.”

    Full article here:


  • Harry R Ray August 8, 2017, 9:50

    According to http://www.solar-flux.forumandco.com, the HILL SPHERE of Beta Pectoris b is in MID-TRANSIT TODAY!!! It is unclear whether the cubesat intended to observe the Hill-sphere transit is in orbit or not. If not, other observatories should be able to detect a Neptune-sized exomoon if there is one to be detected, but an Earth-sized one would be just a little bit out of reach.