≡ Menu

Kepler 1625b: Orbited by an Exomoon?

8,000 light years from Earth in the constellation Cygnus, the star designated Kepler 1625 may be harboring a planet with a moon. The planet, Kepler 1625b, is a gas giant several times the mass of Jupiter. What David Kipping (Columbia University) and graduate student Alex Teachey have found is compelling though not definitive evidence of a moon orbiting the confirmed planet.

If we do indeed have a moon here, and upcoming work should be able to resolve the question, we are dealing, at least in part, with the intriguing scenario many scientists (and science fiction writers) have speculated about. Although a gas giant, Kepler 1625b orbits close to or within the habitable zone of its star. A large, rocky moon around it could be a venue for life, but the moon posited for this planet doesn‘t qualify. It’s quite large — roughly the size of Neptune — and like its putative parent, a gaseous body. If we can confirm the first exomoon, we’ll have made a major advance, but the quest for habitable exomoons does not begin around Kepler 1625b.

Image: Columbia’s Alex Teachey, lead author of the paper on the detection of a potential exomoon. Credit: Columbia University.

None of this should take away from the importance of the detection, for exploring moons around exoplanets will doubtless teach us a great deal about how such moons form. Unlike the Earth-Moon system, or the Pluto/Charon binary in our own Solar System, Kepler-1625b’s candidate moon would not have formed through a collision between two rocky bodies early in the history of planetary development. We’d like to learn how it got there, if indeed it is there. Far larger than any Solar System moon, it is estimated to be but 1.5% of its companion’s mass.

The methods David Kipping has long espoused through the Hunt for Exomoons with Kepler (HEK) project have come to fruition here. Working with data on 284 planets identified through the Kepler mission, each of them with orbital periods greater than 30 days, Kipping and Teachey found interesting anomalies at Kepler-1625b, where Kepler had recorded three transits. The lightcurves produced by these transits across the face of the star as seen from the spacecraft showed deviations that demanded explanation.

Image: Exomoon hunter David Kipping. Credit: Columbia University.

Their interest heightened, the researchers requested and were awarded 40 hours of time on the Hubble Space Telescope, whose larger aperture could produce data four times more precise than that available from Kepler. On October 28 and 29 of 2017, the scientists took data through 26 Hubble orbits. Examining the lightcurve of a 19-hour long transit of Kepler-1625b, they noted a second, much smaller decrease in the star’s light some 3.5 hours later.

Kipping refers to it as being consistent with “…a moon trailing the planet like a dog following its owner on a leash,” but adds that the Hubble observation window closed before the complete transit of the candidate moon could be measured. The paper addresses this second dip:

The most compelling piece of evidence for an exomoon would be an exomoon transit, in addition to the observed TTV [transit timing variation]. If Kepler-1625b’s early transit were indeed due to an exomoon, then we should expect the moon to transit late on the opposite side of the barycenter. The previously mentioned existence of an apparent flux decrease toward the end of our observations is therefore where we would expect it to be under this hypothesis. Although we have established that this dip is most likely astrophysical, we have yet to discuss its significance or its compatibility with a self-consistent moon model.

In and of itself, this is exciting information, but as noted above, we also learn in this morning’s paper in Science Advances that transit timing variations are apparent here. The planet itself began its transit 77.8 minutes earlier than predicted. One way to account for this is by the pull of a moon on the planet, resulting in their both orbiting a common center of gravity and thus throwing the transit calculation (based on an unaccompanied planet) off. What Kipping and Teachey will need to eliminate is the possibility that a second planet, yet undetected, could have caused the timing variation. There is thus far no evidence from Kepler of such a planet.

“A companion moon is the simplest and most natural explanation for the second dip in the light curve and the orbit-timing deviation,” said lead author Teachey. “It was a shocking moment to see that light curve, my heart started beating a little faster and I just kept looking at that signature. But we knew our job was to keep a level head testing every conceivable way in which the data could be tricking us until we were left with no other explanation.”

Image: This is Figure 4 from the paper. Caption: Moon solutions. The three transits in Kepler (top) and the October 2017 transit observed with HST (bottom) for the three trend model solutions. The three colored lines show the corresponding trend model solutions for model M, our favored transit model. The shape of the HST transit differs from that of the Kepler transits owing to limb darkening differences between the bandpasses. Credit: David Kipping, Alex Teachey.

One problem with exomoon hunting is that the ideal candidate planets are those in wide orbits, but this makes for long periods between transits. Even so, the number of large planets in orbits farther from their star than 1 AU is growing, and such worlds should be useful targets for the upcoming James Webb Space Telescope. Although we still have to confirm Kepler-1625b’s moon, such a confirmation could prove only the beginning of a growing exomoon census.

We’ll know more as we make more detections, but for now, I think Kipping and Teachey’s caution is commendable. Noting that confirmation will involve long scrutiny, observations and skepticism from within the community, they point out:

…it is difficult to assign a precise probability to the reality of Kepler-1625b-i. Formally, the preference for the moon model over the planet-only model is very high, with a Bayes factor exceeding 400,000. On the other hand, this is a complicated and involved analysis where a minor effect unaccounted for, or an anomalous artifact, could potentially change our interpretation. In short, it is the unknown unknowns that we cannot quantify. These reservations exist because this would be a first-of-its-kind detection — the first exomoon.

A final thought: The paper points out that the original Kepler data that flagged Kepler-1625b as interesting in exomoon terms were actually stronger than the Kepler data Kipping and Teachey added into the mix for this work. They are now working with the most recent release, which had to be revisited for all factors that could affect the analysis. It turns out that this most recent release “only modestly favors that hypothesis when treated in isolation.” The HST data make the strongest case in strengthening the case for an exomoon. The authors believe that this shows the need to pursue similar Kepler planets for exomoons with HST and other facilities, even in cases where the Kepler data themselves do not show a large exomoon-like signature.

The paper is Teachey & Kipping, “Evidence for a large exomoon orbiting Kepler-1625b,” Science Advances 3 October 2016 (complete citation when I have it).


Comments on this entry are closed.

  • Laintal October 3, 2018, 15:51

    Here is the link to follow up on

    Another interesting topic to read up on. I’m going to be busy today

  • andy October 3, 2018, 17:51

    The main sceptical commentary I’ve seen is in this Twitter thread by Hugh Osborn. I suspect these data are going to be subject to a lot of follow-up analyses in the run-up to the next transit and beyond. If confirmed, this will be a really fun system for the theorists!

  • Michael C. Fidler October 3, 2018, 20:15

    This is great! David Kipping has been working on this for a long time and finally seeing some good results. Many more should result from the load of useful data now coming from TESS and the many ground observations. Using AI to machine learn what to look for in individual transits and timing variations would help find interesting objects…

  • Nicky October 3, 2018, 20:49

    This discovery, if confirmed, can be profoundly impactful to both astronomical and astrobiology community.
    The star Kepler-1625 has mass similar to Sun but with radius 80% larger, implying the star has evolved off main-sequence. Teachey and Kipping constrain the age of the star to be 8.7 ± 1.8 billion years. Since Kepler-1625 has the same mass as our sun and ages as the luminosity and radius increase, Kepler-1625b and Kepler-1625b-i should have spent the first 5.4 billion years orbiting within the conservative habitable zone, comparable with Earth’s 5.8 billion years.
    A super-Jupiter with a Neptune-size moon orbiting within the stellar habitable zone? It might seem gas/ice giants don’t support the life as we know and defies current moon-formation theories, but such moon would also mean that there might exist many other Earth-size moons orbiting around habitable zone gas/ice giants.
    Habitable exomoons? In fact, the tidal heating from the gravitational interaction with giant planet can well extend the lifetime of the moon’s geological activities and cycles, such as plate tectonics, carbonate-silicate cycle and volcanism.

    Teachey and Kipping’s transit data was gathered from Hubble observation during the October 29, 2017 transit of Kepler-1625b, and the latest transit of this planet just happened three weeks ago. Unfortunately, they did not schedule another observation for the latest transit but for the next transit on May 26, 2019. Considering the time of data processing, interpreting, writing and publishing, the existence of Kepler-1625b-i should be confirmed/discarded during the second half of 2020.

    • andy October 4, 2018, 12:48

      A super-Jupiter with a Neptune-size moon orbiting within the stellar habitable zone? It might seem gas/ice giants don’t support the life as we know and defies current moon-formation theories, but such moon would also mean that there might exist many other Earth-size moons orbiting around habitable zone gas/ice giants.

      Playing Devil’s advocate, there might be a set of processes that tend to produce moons with masses like the ones we see in our solar system, and another process that produces binary ice/gas giant planets, with neither producing much in-between. More data needed!

      • Nicky October 4, 2018, 15:42

        Indeed. A bimodal distribution of masses for satellites, surely possible!
        But one thing is that our current moon-formation models predict much smaller satellites. The recent population synthesis model finds the peak of largest satellite-to-planet ratio distribution has upper end at 10^-3, and Galileo moons have a ratio of 10^-4. For Earth-mass satellites, the host planet has to be at least several times more massive than Jupiter based on the population synthesis model.

        Kipping’s HEK project also find planets within 1 au are devoid of moons larger than the Galilean ones.

        A Neptune-size moon just simply means the *formation of large moon in habitable zone* is possible. Kipping’s software predicts that the moon-to-planet mass ratio of Kepler-1625 is around 0.015. If it is indeed true, Saturn-mass planets might well have Earth-size moons.

        • Brett October 4, 2018, 16:44

          I suppose a Neptunian/Ice giant could have an Earth-sized moon, although in that case it would essentially be a double planet system formed either by capture or them forming together (I’m assuming an impact big enough to blow off an Earth-sized moon from a Neptunian world would probably blow off its entire hydrogen atmosphere as well if it’s inside the snow line. ).

    • AlexT October 5, 2018, 5:38

      “This discovery, if confirmed, can be profoundly impactful to both astronomical and astrobiology community.”
      This discovery was confirmed long time ago – the example oir own solar system and it’s gas giants that have lot of moons, some f moons have the mass comparable with Earth mass… There is not any reason to suppose that other star systems cannot have something similar…
      So I really do not understand your excitement…

      • Nicky October 5, 2018, 14:00

        Not sure what you’re talking about. Kepler-1625b-i was never confirmed. It is still considered a candidate. The largest moon in our solar system is Ganymede, which by mass is only 2% of Earth-mass, even smaller than a protoplanet or planetary embryo.

        • AlexT October 6, 2018, 11:46

          I only want to tell, that I do not understand excitement of some commenters here about Exo-Moon finding, when such fact was discovered ~400 years ago by Galileo Galilei :-)
          Wandering why the fact about our own Solar System so underestimated by community here …
          The exact moon/Earth proportions are not changing much in this fact…
          The fact G-type main-sequence star can have Gaz giant planets and those planets can have a relatively big moons…
          So exo-moon finding can only give us the additional prove to already well known fact, it is not really the “new discovery”. This fact also give us evidence that our astronomical instruments and methods have better resolution than it was in the past.

  • Jean Schneider October 4, 2018, 5:45

    1/ The distance is not 8000 ly but 1213 parsec:

    2/ (very) large exomoons were predicted under the name
    of “binary planets” by Cabrera & Schneider in 2007:

    • Alex Tolley October 4, 2018, 10:18

      Your link shows the distance as 2460 pc. That calculates to 8000 ly.

    • andy October 4, 2018, 12:37

      I believe 1213 parsecs was the distance estimate pre-Gaia DR2. The new distance estimate of 8000 light years is from the Gaia DR2 parallax.

      It is unfortunate that the note on exoplanet.eu states that “The exo-moon Kepler-1625 b I is confirmed (Teachey & Kipping)” when Alex Teachey and David Kipping have been making it very clear both in the paper and on Twitter that they do not consider the object to be confirmed.

  • Harry R Ray October 4, 2018, 9:50

    The star is 7200 light-years from Earth, but it is Quite luminous(G?IV)and was bright enough to be included in the GAIA DR2, so, if the planet is several Jupiter mass as proposed, RV measurements are possible. A TTV that large caused by a second planet would mean that the second planet would also have to be VERY MASSIVE as well, and should be detectable. Unfortunately, ESPRESSO is too far south to observe it, so it could take years for HARPS North to detect a planet INTERIOR to Kepler 1625b, or over a decade to detect one EXTERIOR to it. This leaves Spitzer and MAYBE TESS to do IMMEDIATE follow-up observations.

    • Thomas R Mazanec October 4, 2018, 13:53

      Harry, PLEASE fix the SHIFT key on your keyboard, it is getting ANNOYING.

  • Brett October 4, 2018, 12:30

    That’s wild if it can be confirmed. Some of the modeling suggests that a Neptune-mass moon could form around a Jovian, although it would be very unlikely. But a capture would be pretty unlikely as well, so go figure.

  • P October 4, 2018, 21:01

    So, we have many well characterised transits of giant planets from Kepler, but most are hot/warm giants that might have been stripped of any putative moons – hence no detections despite people looking.

    Someone needs to take a closer look at Kepler’s smaller set of long term mono transits to see if there’s any interesting light curve anomalies, then followup with HST and eventually ELTs!


  • Mike October 4, 2018, 22:29

    I’m looking forward to the next set of observations of the transit. If confirmed it would be such a large moon even considering it’s orbiting a Jovian sized planet. Is this a planet and its moon or a double planet? Is their common centre of mass exterior to the Jovian or below its surface?

  • ljk October 5, 2018, 15:07

    What are the chances that such a large exomoon might have a moon of its own?

  • Michael C. Fidler October 6, 2018, 10:45

    In The Hunt For Aliens, Satellites May Light The Way.

    “The Earth is expanding, satellite by satellite, every rocket launch carrying a piece of the planet’s crust into orbit. Should this incidental geoengineering venture continue, it will reshape our planet’s profile as seen across even interstellar distances—giving our smooth sphere a noticeable bulge.

    If we’re puffing up our planet, other civilizations could be doing the same to theirs, producing a ring of satellites that we might be able to spot with telescopes we have today. That’s according to Hector Socas-Navarro, an astrophysicist at the Instituto de Astrofísica de Canarias in Spain, who gave a talk on the topic at NASA’s Technosignatures Workshop in Houston last week. Scientists have long speculated that fantastical sun-sized structures might betray the presence of technological aliens, but while a mega-solar panel blocking a distant star is theoretically easy to spot, such notions remain squarely in the realm of science fiction. Thought experiments like Socas-Navarro’s, however, show that now, equipped with better telescopes than their predecessors, researchers are taking searches for planet-level changes more seriously.”


    Toward Detection of Exoplanetary Rings via Transit Photometry: Methodology and a Possible Candidate.


    Systematic Search for Rings around Kepler Planet Candidates: Constraints on Ring Size and Occurrence Rate.


    Possible Photometric Signatures of Moderately Advanced Civilizations: The Clarke Exobelt.


  • Harry R Ray October 7, 2018, 14:17

    One of my major concerns was that, even if the Kepler 1625 “companion” were to be confirmed, it might end up being merely a planet orbiting a brown dwarf, instead of a moon. Their original paper favored a ~10Mj mass for Kepler 1625b, but now the most favored mass is just a bit greater than 3Mj, so, for the time being, we can put that concern to rest. ALSO: “While a low-surface gravity planet will show very pronounced molecular features and a steep slope at short wavelengths due to Rayleigh scattering, a high surface gravity world will yield a substantially flatter transmission spectrum…”. Analysing this spectrum was a major factor in constraing Kepler 1625b’s mass. Which leads to the following speculation: The best mass fit for Kepler 1625b I is still >Uranus<Neptune. The authors were strangely silent in their lack of mentioning any spectrum obtained from Kepler 1625b I. Perhaps they saw something, but nothing conclusive. However, teasing something molecular out with some kind of machine learning algorithm, would immediately confirm the exomoon.

    • Harry R Ray October 11, 2018, 9:46

      Cavaet: The following is a purely “science fiction” scenario! Assume molecules(from my above comment)were detected. However, to emit enough photons to be detected, they had to be super-hot(near the point where they would break apart. Theorists conclude that the only way this could be happening is for Kepler 1625b I to actually be two objects of equal mass(~9Mearth)orbiting very close to each other, but not close enough to share a common atmospheric envelope(a la Rocheworld). One of these objects is a 1.5 Earth-radius “iron world” with an extremely powerful magnetic field. The other is a 3 Earth-radius “pure water” world with a negligable iron core, and thus, a very weak magnetic field. Because of this, the magnetic field of the “iron world” overwhelms the magnetic field of the “pure water” world, and its magnetic field lines extend deep into the mantle of the “pure water’ world, heating it up via ohmic dissipation to a temperature much greater than can be achieved by either instellation or tidal disruption. My question to any reader of this comment up to the task is: Could such a system be stable over 9 billion years?

  • Michael C. Fidler October 8, 2018, 1:14

    Evidence for a Large Exomoon Orbiting Kepler-1625b.
    Alex Teachey, David M. Kipping
    (Submitted on 4 Oct 2018)


    The habitable zone for Earthlike exomoons orbiting Kepler-1625b
    Duncan Forgan.
    (Submitted on 4 Oct 2018)

    “The recent announcement of a Neptune-sized exomoon candidate orbiting the Jupiter-sized object Kepler-1625b has forced us to rethink our assumptions regarding both exomoons and their host exoplanets. In this paper I describe calculations of the habitable zone for Earthlike exomoons in orbit of Kepler-1625b under a variety of assumptions. I find that the candidate exomoon, Kepler-1625b-i, does not currently reside within the exomoon habitable zone, but may have done so when Kepler-1625 occupied the main sequence. If it were to possess its own moon (a “moon-moon”) that was Earthlike, this could potentially have been a habitable world. If other exomoons orbit Kepler-1625b, then there are a range of possible semimajor axes/eccentricities that would permit a habitable surface during the main sequence phase, while remaining dynamically stable under the perturbations of Kepler-1625b-i. This is however contingent on effective atmospheric CO2 regulation.”


  • Michael C. Fidler October 9, 2018, 10:27
  • Michael C. Fidler October 10, 2018, 10:31

    10 October 2018
    Moons can have moons and they are called moonmoons
    By Leah Crane

    “Stars are orbited by planets, which are orbited by moons, but what comes next? More moons, according to a new analysis.
    Juna Kollmeier at Carnegie Observatories in Pasadena, California, and Sean Raymond at the University of Bordeaux, France calculated whether a moon orbiting a planet could have a moon of its own.
    A moon of a moon has no formal name, perhaps because we have never spotted one, but both submoon and moonmoon have been suggested.”


    Now I know how the cow jump over the moooonmoooon….

    • Harry R Ray October 12, 2018, 9:25

      I propose another name. These objects are probably not stable over billions of years. If, when they are finally ejected from the lunar orbit, they make a very close pass to a massive super-Jupiter, two things will happen. ONE: Tidal stresses will elongate the former moonmoon, and; TWO: It will be ejected from the solar system and wander the galaxy forever. Therefore, the more appropriate name for these objects should be(are you ready for the punchline?)`Oumoonamoonas(lol)!

  • Harry R Ray October 14, 2018, 16:47

    Just a thought: Jupiter’s magnetopause is 50-100 Rj in length. Kepler 1625 b’ s magnetopause should be similar in length. Thus, if it does exist, Kepler 1625b I should orbit entirely within Kepler 1625b’s megnetosphere. My question to any reader who knows: Could our most powerful currently operating radio telescopes(VLBI-radioAstron, Janski VLA, LOFAR) be able to detect any magnetic reconnection events?

  • Harry R Ray October 15, 2018, 10:00

    Good news and bad(but good for kipping et al.)news! First the bad news(and the link to exomoons): Analysis of old DASCHE plates suggests strongly that J 1407b is not an exoplanet, but instead,is a brown dwarf either in an extremely long period orbit or not even bound to J1407. So, in case it takes several years for Kipping et al to obtain the “extrordinary proof” necessary to confirm Kepler 1625b I as an exomoon, they cannot be “scooped” by Mamajek et al. Now the good news: The first TRAPPIST-1 paper that incorporates the July revised masses for the TRAPPIST-1 planets! ArXiv: 1810.05210. “Limits on clouds and Hazes for the TRAPPIST-1 planets.” by Sarah E Moran, Sara M Horst, Natasha E. Batalha, Nicole K Lewis, Hannah R Wakefield. My take: Looks like TRAPPIST-1d may be the first “sub-Venus”, whereas TRAPPIST-1e may very well be an Earth analog.

  • ljk October 30, 2018, 10:15

    The astronomical satellite that analyzed this system may have gone to that big exoplanet in the… bigger sky: