≡ Menu

Learning More about Outer System Planets

What kind of planets are most common in the outer reaches of a planetary system? It’s a tricky question because most of the data we’ve gathered on exoplanets has to do with the inner regions. Both transit and radial velocity studies work best with large planets near their stars. But a new gravitational microlensing study looks hard at outer system planets, finding that planets of Neptune’s mass are those most likely to be found in these icy regions.

It should be no surprise that gravitational microlensing has produced few planets, about 50 so far, compared to the thousands detected through transit studies and radial velocity methods. After all, microlensing relies upon alignments that are far more unusual than even the transit method, in which a planet crosses the face of its star as seen from Earth. In microlensing, astronomers look for rare alignments between a distant star and one much nearer.

Given the right alignment, the ‘bending’ of spacetime caused by the nearer star’s mass allows researchers to study changes in the brightness of the background star, which can be clues to the existence of a planet. Microlensing can see not just planets close to their host stars but those far distant from the primary. Moreover, as the new work points out, we can use microlensing to figure out the mass ratio of the planet to the host star, and in about 40 percent of events, we can measure the mass of the host star and planet themselves.

A team led by Daisuke Suzuki (NASA GSFC) identified 1474 microlensing events between 2007 and 2012, drawing on data from the Microlensing Observations in Astrophysics (MOA) project, a collaboration between Japanese and New Zealand researchers that alerted astronomers to 3300 potential microlensing events in this time period. The analysis also incorporates data from the Optical Gravitational Lensing Experiment (OGLE).

Suzuki and colleagues homed in on the frequency of planets compared to the mass ratio of planet and star and the distances between them. A typical planet-hosting star is about 60 percent of the mass of the Sun. Its typical planet is between 10 and 40 times the mass of the Earth. By comparison, Neptune is about 17 Earth masses, while Jupiter is 318 times as massive as the Earth. Cold Neptune-mass worlds are thus identified as the most common kinds of planets beyond the ‘snow line,’ the point in a planetary system beyond which water remained frozen during planetary formation. In our Solar System, the snow line is at about 2.7 AU, roughly the middle of the main asteroid belt.

The paper surveys stars toward the galactic bulge, where the chances of a microlensing alignment are highest. Says Suzuki:

“We’ve found the apparent sweet spot in the sizes of cold planets. Contrary to some theoretical predictions, we infer from current detections that the most numerous have masses similar to Neptune, and there doesn’t seem to be the expected increase in number at lower masses. We conclude that Neptune-mass planets in these outer orbits are about 10 times more common than Jupiter-mass planets in Jupiter-like orbits.”


Image: This graph plots 4,769 exoplanets and planet candidates according to their masses and relative distances from the snow line, the point where water and other materials freeze solid (vertical cyan line). Gravitational microlensing is particularly sensitive to planets in this region. Planets are shaded according to the discovery technique listed at right. Masses for unconfirmed planetary candidates from NASA’s Kepler mission are calculated based on their sizes. For comparison, the graph also includes the planets of our solar system. Credit: NASA’s Goddard Space Flight Center

So based on the MOA data, planets forming in the outer reaches of a planetary system are likely to be Neptunes. But remember the limitations of the data here — we have relatively few detected exoplanets, and in fact, only 22 planets (with a possible 23rd) show up in the 1474 MOA events. What’s heartening is how we are going to go about expanding that dataset.

Tightening up the constraints on mass and distance to the lens systems will ultimately allow us to measure what the paper calls the ‘microlensing parallax effect,’ determining the distance of the system with the help of space telescopes far from the Earth. From the paper:

The ultimate word on the statistical properties of planetary systems will be achieved from the space based exoplanet survey (Bennett & Rhie 2002) of the WFIRST (Spergel et al. 2015) mission, and hopefully also the Euclid (Penny et al. 2013) mission. The high angular resolution of these space telescopes will allow mass and distance determinations of thousands of exoplanets because it will be possible to detect the lens star and measure the lens-source relative proper motion with the high resolution survey data itself. This will give us the same comprehensive picture of the properties of cold exoplanets that Kepler is providing for hot planets.

WFIRST (Wide Field Infrared Survey Telescope) was formally designated as a NASA mission at the beginning of this year. To be launched in the mid-2020s, it will carry a 288 megapixel multi-band near-infrared camera and a coronagraph for the suppression of starlight. ESA’s Euclid mission, like WFIRST, has a gravitational microlensing component, with a launch date in late 2020. If we can use space-based resources like these to enrich our microlensing catalog, our understanding of the outer precincts of exoplanetary systems will surge.

The paper is Suzuki et al., “The Exoplanet Mass-Ratio Function from the MOA-II Survey: Discovery of a Break and Likely Peak at a Neptune Mass,” Astrophysical Journal Vol. 833, No. 2 (13 December 2016). Abstract / preprint.


Comments on this entry are closed.

  • Alex Tolley December 19, 2016, 14:17

    Without looking at the numbers, visually the distribution of microlensed planets (pale blue) looks rather similar to Kepler transit (light green), just shifted in distance from te star due to the technique’s limitations.

    I’m wondering if the observed distributions are simply a result of the sensitivities of the different techniques, all of which are biased strongly towards larger planets.

    If Neptune-sized planets are truly the most common, is this predicted by planetary formation simulations, or is it a parameter to be tweaked to make the models fits the observations?

  • spaceman December 19, 2016, 14:30

    Paul, thank you for posting on this interesting study. I have a few questions about the main conclusion of the paper:

    1). To what extent are the current microlensing surveys biased in favor of finding planets in the mass range of 10 to 40 Earth masses as opposed to, say, planets below 10 Earth masses?

    2). What percentage of stars, according to this study, have outer planets in the 10 to 40 Earth masses range?

    3). If the most common type of outer planet is similar in mass to Neptune, what might be the underlying theoretical reason(s) from a planet formation perspective as to why planets this size are the most prevalent? In other words, it is easy to understand from core accretion models why cold Neptune mass planets are more common than Jupiter-sized worlds, but why would we expect them to be more common than cold Earth mass or cold Mars mass planets?

    Oh, one more: Do both the Euclid and WFIRST missions have planned microlensing surveys or just WFIRST?

    • Paul Gilster December 19, 2016, 14:55

      On the matter of Euclid and WFIRST, both have microlensing components. Euclid’s ESA site refers to using microlensing “to map the dark matter and measure dark energy by measuring the distortions of galaxy images by mass inhomogeneities along the line-of-sight.” Whether it will be used for the kind of exoplanet work Suzuki hopes for is not clear, and his paper refers to this only as a possibility.

      • Ashley Baldwin December 27, 2016, 13:25

        The Exoplanet Euclid Legacy Survey, ExELS hopes to discover several thousand down to Earth mass exoplanet’s via micro lensing of the galactic bulge , supplemented by a similar number of close in down to hot Neptune class planets via traditional transit photometry . That’s now , before the techniques required will be improved . It will want to compete with what is essentially its rival in WFIRST .(I understand there is some residual bad feeling between the two teams despite a public face of cooperation given the big cross over of the two surveys )

        • Ashley Baldwin December 27, 2016, 13:40

          So essentially between TESS, PLATO , Gaia , Euclid , ground based surveys and WFIRST (plus whatever Kepler K2 can squeeze out in its remaining time ) that’s one monstrous amount of representative exoplanets and exoplanetary systems. I would expect well in excess of 100 K by the end of the next decade to be a reasonable estimate assuming even just a modest degree of mission extensions. About time there were dedicated spectroscopic characterisation observatories rather than the laudable and developmental but limited ad hoc efforts of Hubble, Spitzer and JWST.

  • Marshall Eubanks December 19, 2016, 15:28

    Well, here in the solar system we also don’t have any Earth-mass planets at distances > 1 AU, while we have 2 “Neptunes.” It may be that, if planets form at all in outer solar systems, they will grow beyond Earth size.

    The solar system analog does, however, indicate that the absence of Earths beyond the snow line doesn’t mean that Earth’s can’t be found inside it.

    • Spaceman December 19, 2016, 20:24

      True, however, one of the central insights of the past 20 years of exoplanet research is that our solar system may be atypical. For instance, here in the solar system we don’t have super-Earths either even though we know from the Kepler mission and RV surveys that super-Earths and mini-Neptunes are quite common around other stars.

    • Ashley Baldwin December 27, 2016, 13:48

      Yes. The range of masses of the erstwhile “planet nine” although averaging out at ten Earth masses does have a lower limit of just two Earth masses. Whatever it’s still likely either a mini or micro Neptune . ( you heard the latter term first here !) Remains to be seen whether that represents a terrestrial planet though as the extreme low temperature/energy at that distance allows even wee little Pluto to have its own non negligible atmosphere . ( and still modestly massed Titan to have a substantial atmosphere 2/3 nearer the Sun)

  • Bruce Mayfield December 20, 2016, 9:11

    The general trend in nature is that in objects the smaller out-number the larger, and often vastly so, such as with stars. Why wouldn’t this be the case with planets too? We likely are just overlooking great numbers of small planets due to detectablity bias.

  • tardigrade December 20, 2016, 11:07

    This study just adds to the conventional wisdom that our solar system is a very rare kind of solar system. With two planets much more massive than Neptune and just two (for now) neptunes, we arer definitely not the norm. What is extremely suprising, is that the system of the star nearest to us now appears to be nowhere near the norm either. A recent paper (Proxima Reloaded: Unraveling the Stellar Noise in Radial Velocities. ArXiv.org/abs/1612.03786, by M Damasso, F. Del Sordo) apparently eliminates the possibility of a neptune in the 50 to 500 day orbit range*. Previous studies have eliminated brown dwarfs, Jupiters, and Saturns, so, unless the two “mesolensing” observations turn up something, we appear to be stuck with just one planet. What makes it more unique, is it appears to be the only system known now with an Earth- sized eccentric (0.17% from the new paper) planet in the habitable zone. Having two extremely rare and completely different from each other systems so close to each other is completely astounding! Cavaet: This may change! The PALE RED DOT team posted this on their Twitter account recently: “yes! we saw it! #Proximab confirmed to be robustly detected on current data, phew! Further signal…we’ll see next year? Stay tunned.” I,for one do not plan to stay tunned. I plan to remain in my active state and watch this one like a hawk (us tardigrades are collectively trying to merge into an analog of a hawk “eye” but so far without any success.)

    • tardigrade December 20, 2016, 11:10

      I am assuming that “Further* signal” means an additional planet rather than a transit signal of Proxima b.

  • tardigrade December 20, 2016, 11:47

    Correction! The new study states that Proxima b’s eccentricity is 17%, not 0.17%.

    • spaceman December 20, 2016, 12:48

      Though mini-Neptune type planets may be ruled out around Proxima centauri does not necessarily mean that the system only has one planet. There is some emerging evidence that mid to late M dwarfs have fewer several Earth mass planets than larger stars, but that stars like Proxima (M5?) have no fewer Earth to sub Earth mass planets. So, perhaps hypothetical outer planets around Proxima may be Earth or Mars sized thereby eluding detection thus far. For example, if Proximab was at 100 days would we have even been able to detect it so far??

      • Ashley Baldwin December 27, 2016, 13:14

        I think it will be discovered by either further visual or possibly ever improving IR Doppler spectroscopy. If not then via direct imaging by the E-ELT and/or possibly astrometry from WFIRST ( a little known stretch goal of that mission- see below )

  • Harry R Ray December 21, 2016, 12:51

    ALMAhas just recently had a MAJOR UPGRADE with the installatiopn of Band 5 which can detect VERY FAINT TRACES OF WATER! Considering how close Proxima b is to us, does ALMA now have a chance to detect water at the planet’s surface or in its cloud-tops? If any reader knows the answer to this, please post it here. If not, could someone contact PALE RED DOT and ask THEM?

    • Ashley Baldwin December 28, 2016, 16:28

      ALMA doesn’t have anything like the resolution given it images in long millimetre/sub millimetre wavelengths .Where angular resolution in metres: :

      R = 1.22 (wavelength in metres) /Aperture ,D, in metres

      Here D is represented by the maximum “synthetic ” aperture for an interferometer like ALMA. So although ALMA’s aperture demonenator will be in the hundreds of metres its effect on the equation will be offset by the long 1/1000m , millimetre wavelength of the nominator (versus say just billionths of a metre for visible or near infrared) . That’s why radio telescopes and other related long wavelength imaging devices are arranged as multi baseline interferometers to avoid the requirement and cost of having a large “filled in” aperture .

  • Ashley Baldwin December 26, 2016, 14:51

    In addition to the several thousand ( down to Earth mass or less) planets imaged by WFIRST further out from their parent stars , Gaia is expected to finds tens of thousands of Neptune and larger sized planets via astrometry ( the exact number depending on whether its five year primary mission is extended- it certainly has consumables for approaching ten allowing for instrument degradation) that like micro lensing ( and unlike more traditional ) transit photometry and Doppler spectroscopy ) also favours planetary discovery further out , and with accurate mass and orbital inclination calculation . In terms of micro lensing , Scott Gaudi has published on arxiv extensively on the technique and related surveys and is also WFIRST PI for this element of its mission. WFIRST will also direct in age. Smaller number of planets is expected to see several thousand more planets by transit photometry . Some exoplanet observatory ! Between WFIRST , Gaia, Euclid and Trasit based TESS and PLATO we will have a near fully representative sample of stellar systems by 2030.

    The additional Proxima planetary signature was in the RV spectroscopy data but hasn’t been confirmed given it was close to a multiple of the stars rotational period and the extensive background noise . Watch that space….

    • Ashley Baldwin December 27, 2016, 13:11

      If all goes well and its sensor array and stability are up to scratch ( amongst many other tough variables ) it’s also hoped that WFIRST will also discover via astrometry all ,down to Earth mass, planets within 10 parsecs too. With till 2024 to perfect the hard and software necessary , and with practical experience gleaned from both Gaia and Euclid I’m optimistic. Combined with Project Blue and the ELTs that gives good coverage of any Alpha Centauri ( and Proxima ) planets .

      • Dominic Benford February 22, 2017, 10:49

        WFIRST is considering having a precision relative astrometry mode (akin to Hubble’s “spatial scanning” mode) that should be roughly comparable to Gaia. This approach demands very little of its infrared detector array or its pointing stability, and so is a likely capability. However, there are of order 300 suitable stars to survey within the nearest 10 parsecs, and so these measurements would take on the order of 1000 hours of observatory time. This would likely be done in the context of a Guest Observer project, and so would have to make a powerful case for its value to receive such a large allocation of time.

  • ljk June 2, 2017, 21:43

    Let’s give Neptune’s ring system their due:


    And get a probe or ten back to that world too.