≡ Menu

New Planet Detected in Circumbinary System

The transit method has proven invaluable for exoplanet detection, as the runaway success of the Kepler/K2 mission demonstrates. But stars where planets have been detected with this method are still capable of revealing further secrets. Consider Kepler-47. Here we have a circumbinary system some 3340 light years away in the direction of the constellation Cygnus, and as we are now learning about circumbinaries — planets that orbit two stars — the alignment of the orbital plane of the planet is likely to change with time.

Let’s pause for a moment on the value of the detection method. Transits detected in the lightcurve have helped us identify 10 transiting circumbinary planets, with the benefit of allowing astronomers to measure the planets’ radius even as variations in the duration of transits and deviations from the expected timing of the transits establish the circumbinary orbit.

At Kepler-47, we’re looking at the only known multi-planet circumbinary system. Moreover, the orbital period of the binary stars is in the range of 7.5 days, making this the shortest known orbital period for any known circumbinary system. The first two Kepler planets were detected in 2012, but San Diego State University astronomers now find a middle planet between these, Kepler-47d, its strengthening transit signal the result of orbital plane adjustment. In fact, the transit depth for the hitherto undetectable world has become the deepest of the three planets.

Jerome Orosz (SDSU) is the paper’s lead author.

“We saw a hint of a third planet back in 2012, but with only one transit we needed more data to be sure. With an additional transit, the planet’s orbital period could be determined, and we were then able to uncover more transits that were hidden in the noise in the earlier data.”

And according to co-author and SDSU colleague William Welsh, this planet defied expectations by showing up not exterior to the previously known planets but between them. “We certainly didn’t expect it to be the largest planet in the system. This was almost shocking,” said Welsh.

Image: Artist’s impression of the third planet in the Kepler-47 circumbinary system. Credit: NASA/JPL Caltech/T. Pyle.

So what we see at Kepler-47, at least at this juncture, is an inner planet 3.1 times the size of Earth, in an orbit taking 49 days around this G-class star orbiting an M-dwarf. The other planets here are, respectively, 7 times Earth-size on an 187-day orbit (Kepler-47d), and 4.7 Earth-size with a 303 day orbit. Remember that we are talking about planets orbiting two stars, in a system compact enough to fit inside the Earth’s orbit of the Sun.

Kepler-47’s system may be telling us something interesting about planet formation. From the paper:

This is the first detection of a dynamically packed region in a circumbinary system, and it further confirms suspicions that planet formation and subsequent migration can proceed much like that around a single star, at least when far from the binary (Pierens & Nelson 2008, 2013; Kley & Haghighipour 2014, 2015). We also find that, although they are close to having integer commensurate periods, the middle and outer planets are not in a mean-motion resonance–and yet they are gravitationally interacting and exchanging angular momentum, as indicated by their anti-phased oscillations in inclination and eccentricity.

The authors find the planetary configuration dynamically stable for at least 100 million years, adding:

This nearly circular, co-planar, packed configuration is unlikely to have arisen as an outcome of strong gravitational scattering of the planets into their current orbits. Rather, the observations suggest that the planetary configuration is the result of relatively gentle migration in a circumbinary protoplanetary disk.

Image: This is Figure 28 from the paper. Caption: The conservative (dark green) and optimistic (light green) habitable zone regions are shown for the Kepler-47 system. The red circle shows the critical stability radius (Holman & Wiegert 1999), interior to which planetary orbits are most likely unstable. Credit: Jerome Orosz/William Welsh et al.

On the matter of habitability, there is little reason to expect life here. These are low density worlds, all three being less dense than Saturn, which implies substantial hydrogen and helium atmospheres. The outer planet receives an average insolation from its two stars that is 86.5 percent of what the Earth receives. But while that puts this world within the boundaries of the circumbinary habitable zone, the density implies a world somewhere in the range between Neptune and Saturn. The newly discovered middle planet skirts the inner edge of the habitable zone, but again its density makes life unlikely.

The paper is Orosz et al., “Discovery of a Third Transiting Planet in the Kepler-47 Circumbinary System,” Astronomical Journal Vol. 157, No. 5 (16 April 2019). Abstract / Preprint.


{ 15 comments… add one }
  • Harry R Ray April 25, 2019, 10:09

    I have some bad news and some good news. First the bad news. Kepler 1625b I did NOT transit Kepler 1625! It has now been proven to be just an artifact in the data. However Kepler 1625b I’s existence has not been proven wrong, because the TTV still seems to be real. Now for the good news. PROTON DECAY HAS BEEN DETECTED FOR THE FIRST TIME!!!!! The “embargoed”(lol)paper is set to appear in tomorrow’s “Nature”.

  • Gary Wilson April 25, 2019, 13:09

    Life that follows the Terran model that is. Can abiogenesis and evolution occur on gas giant planets or possibly their moons? We will need to study examples much closer to our own system to find out (as well as our own gas giants of course).

  • Bruce D. Mayfield April 25, 2019, 13:44

    “So what we see at Kepler-47, at least at this juncture, is an inner planet 3.1 times the size of Earth, in an orbit taking 49 days around this G-class star orbiting an M-dwarf.”

    Err, wouldn’t the M star be in orbit about the G?

    • Paul Gilster April 25, 2019, 14:50

      Good point. Maybe even better to say they’re both orbiting a common barycenter.

      • Bruce D. Mayfield April 26, 2019, 13:12

        Yes of course. You’re discussion here prompted me to drill down a little deeper into the properties of this interesting system. From wikipedia:

        “Kepler-47 is a binary star system composed of a G-type main sequence star (Kepler-47A) and a red dwarf star (Kepler-47B). The stars orbit each other around their center of mass between them, completing one full orbit every 7.45 days.[3] The stars orbit their barycenter from a distance of about 0.084 AU.[3] The stars have 104% and 35% of the Sun’s mass, and 96% and 35% of the Sun’s radius, respectively.[3] They have surface temperatures of 5636 K and 3357 K.[3] Based on the stellar characteristics and orbital dynamics, an estimated age of 4–5 billion years for the system is possible.[3] In comparison, the Sun is about 4.6 billion years old[16] and has a temperature of 5778 K.[17]

        The primary star is somewhat metal-poor, with a metallicity ([Fe/H]) of about –0.25, or about 56% of the amount of iron and other heavier metals found in the Sun.[3] Both of the stars’ luminosities are typical for their kind, with a luminosities of around 84% and 1% of that of the solar luminosity, respectively.[3]”

        Elsewhere in the above quoted article it states that these two stars are classified as being G6 and M4. I was a bit surprised to find that this G star of about the same age and with a little more mass than our sun is as low in luminosity as it is. The lower metallicity must account for this cooler temp in spite of its larger mass.

        The centers of the pair being only 0.084 AU (12.6 million km) apart along with masses of 1.04 and 0.36 Sols puts the system’s barycenter at about 3.24 million km from the primary’s center, several times outside the radius of the G star.

  • Bruce D. Mayfield April 26, 2019, 15:17

    There is one other thing that seems to be in error here: the illustration from the paper showing the obits and the HZs. As shown, the outer c planet is near the inner (therefore hotter) edge of the “optimistic HZ”, but that’s not the case. It really would be closer to the outer colder edge, as the planet only receives about 87% of the energy per m^2 as Earth. The planet’s average temperature is estimated to be a chilly -17c.

    • Bruce D. Mayfield April 26, 2019, 20:18

      Correction, that temp for Kepler-47c is an even colder -28C.

    • andy April 27, 2019, 5:51

      Actually that does put it fairly close to the inner edge. The reference for the habitable zone boundaries is Kopparapu et al. (2013), who calculate the boundaries for the Sun at 101% (inner) and 34% (outer) of Earth insolation for their conservative habitable zone boundaries. The Earth is right at the inner boundary.

      • Bruce D. Mayfield April 27, 2019, 21:11

        Andy, I like being agreeable, but cannot on this point. Earth’s mean temp is about 15C, While Kepler-47c’s is -28C. I don’t question the science, just the illustration. Mistakes can be made with the way graphics are drawn up, and I think this is such a case.

        • andy April 30, 2019, 14:12

          Nope, sorry, that’s not a like-for-like comparison. The -28°C temperature for Kepler-47c is an equilibrium temperature assuming an albedo of 0.34 and full redistribution of energy across the entire surface of the planet. The value for Earth calculated using the same assumptions is -22°C. The actual Earth is significantly warmer than this because the atmosphere creates a greenhouse effect. It is also less reflective than the assumed value for Kepler-47c: the value of 0.34 chosen by the authors is appropriate for a Jupiter/Saturn-like planet.

          This goes to show that comparisons using temperatures can be utterly misleading when considering habitability – it is far better to work in terms of incident flux.

          • Bruce D. Mayfield May 3, 2019, 5:43

            Ok then, with that further explanation my understanding is improved. Thanks for your help andy.


  • wdk April 26, 2019, 23:49

    The recurring issue of “density” of planets detected by Kepler gives rise to
    some questions. How many of these density estimates are based on spectroscopic doppler determinations of stellar lines ( receding and advancing velocities)? What’s their accuracy? Now, what if all the information we have about a transiting planet is based on luminosity drops of the primary star? …
    Suppose we were observing the Earth and the moon from afar and observed that the occulting of the sun gave 1 and 1/16th the value of
    Earth’s surface area as the reason for luminosity loss? Would an observer argue that there was a moon of Earth with a quarter of its radius based on solar doppler? It wouldn’t be of any help. The mass of the Earth Moon system would be 82 times the mass of the moon or 1 + 1/81st the mass of the earth.

    Absent sensitive information about atmospheres, it could be that some large binary planetary systems ( e.g. Earth=sized and another Earth-sized) could be hiding in low density transit data. Two earths in a locked binary would come across as 2 x earth mass but with twice the surface area. A two x mass of Earth with same density would have cube root of two increase in radius and root 2^(2/3) increase in surface area.

    Another argument for this is the proliferation of binary star systems and sometimes their embedding in other binaries. So far, with several thousand exo-planets, we have only one or two moon candidates. And those are based on drop outs. A planet and moon along the line of sight reducing the shadow. With significant separation, or out of plane motions, this might be a rare event. So in effect, we have a binary planet desert in comparison to stellar binaries – or perhaps even asteroids.

  • wdk April 27, 2019, 14:32

    A little further consideration of the binary planet vs. low density planet
    arguments – coming to mind after I hit the send button.
    If the Earth -moon system were viewed as a transiting system by someone else’s ( another star system’s) Kepler telescope, detection of the moon as distinct from the Earth would require someone out there to notice a plus or minus 1/16th variation in the loss of luminosity due to the Earth’s transit. And the separation of about 400,000 km between
    the Earth and moon at max elongation against the solar diameter of about 1.4 million kilometers. The transit isn’t always going to be at 400,000 kilometers at the edge. Each month, it could range in separation from 0 to 400,000, average about half.

    With this example, I think the diameter of the moon would be more of a problem than the distance separation against the background star. But I can imagine that other systems (Jupiter and I with about the same radial separation) would be more problematic due to surface area ratios.
    Binaries could be tighter with high mass ratios than Earth Moon, but
    there would be the Roche limit(s) to contend with too.

    Still, I submit that there might be a few binaries lurking in the data base
    of “low density” planets.

    • Bruce D. Mayfield April 28, 2019, 10:58

      Those are some excellent points about binary planets and exomoons wdk. No doubt the reason so few have been found is just how difficult they are to detect with current technology. But the current tools also introduce a bias against finding moons because what is easiest to find has been all these tightly packed systems in which having large moons is unlikely due to orbital instability issues.

      Widely spread planetary systems (like ours) provide the needed extra room for moons and double planets to exist in stable orbits. The question is then, how common is Sol’s type of spread out system? The fact that this type of system is hard to find is obscuring their true rate of occurance.

Leave a Comment