Saturday, September 17, 2016

A New Planet for Kepler-20

Figure 1. All announced planets of Kepler-20 at their relative sizes, with colors corresponding to the densities provided by Buchhave et al. 2016 (see Figure 2 for the color key; redder hues indicate higher densities). Since planet g does not transit, the radius and composition shown here are informed guesses, as indicated by the striped fill. Planets e and f are about the same size as Earth, but none of the planets in this system are cool enough to support Earth-like conditions.

Lars Buchhave and colleagues just reported new radial velocity data on Kepler-20, one of the best-known systems revealed by the Kepler Mission. The new results include more precise mass estimates for three of the system’s five known planets (Kepler-20b, c, d) and robust evidence for a previously unknown sixth candidate (Kepler-20g), whose orbit is apparently misaligned with the others. The new planet is notably more massive than its five companions, but despite its short orbital period (35 days), it was not observed in transit by the Kepler Telescope.

When preliminary Kepler data started circulating in 2011, we learned that a highly specific orbital architecture is more common than anyone ever dreamed: compact systems with three or more planets visible in transit (Lissauer et al. 2011b). Lissauer & colleagues reported 55 in 2011. By May 2016, the Extrasolar Planets Encyclopaedia listed 154.

Two systems have always stood out from the pack. One is Kepler-20 (Gautier et al. 2012), which harbors five transiting planets inside a semimajor axis of 0.4 astronomical units (AU). The other is Kepler-11 (Lissauer et al. 2011a), with six transiting planets inside a semimajor axis of 0.5 AU. Both systems center on mature G-type stars like our Sun, yet each one sustains a rich multiplanet architecture confined within a radius similar to the semimajor axis of Mercury (0.39 AU). The corresponding region of our Solar System, of course, is empty.

From the beginning, Kepler-11 has received the lion’s share of attention, for one simple reason. Five of its six planets exhibit transit timing variations (TTVs), such that their orbital motion periodically speeds up or slows down to avoid awkward encounters with neighboring planets. Although this behavior had previously been theorized, Kepler-11 and another early discovery, Kepler-9, were the first cases ever actually confirmed (Ford et al. 2011).

For Kepler-11, analysis of TTVs enabled estimates of the masses of the five inner planets, which presented yet another surprise (Lissauer et al. 2011a). Although their radii were originally estimated in the range of 2 to 4.5 Earth units (Rea), all five turned out to be substantially less massive than Neptune, whose radius of 3.9 Rea contains a mass of 17.2 Earth units (Mea). Like Neptune, Lissauer & colleagues concluded, the Kepler-11 planets must be enveloped in extended atmospheres of hydrogen and helium (H/He), even though their individual masses are smaller than 10 Mea. With that inference began the modern era of planetology.

The planets of Kepler-20 are packed almost as tightly as those of Kepler-11, but no TTVs are available to provide mass estimates. Fortunately, Kepler-20 is substantially closer to our Solar System than is Kepler-11. Its distance is estimated at 290 parsecs (945 light years), versus 613 parsecs (1998 light years) for Kepler-11. This proximity enabled the collection of ground-based radial velocity observations in 2009-2011 by the Keck/HIRES spectrograph, which placed rough constraints on the masses of Kepler-20b, c, and d (Gautier et al. 2012). These constraints indicated that, despite the broad similarities between Kepler-20 and Kepler-11, the planets of the former star are both denser and more massive than those of the latter.

Figure 2. Kepler-20 and Kepler-11 Compared

Planets are rendered at their relative sizes on the same orbital scale, with semimajor axes in astronomical units (AU). Numbers in red indicate approximate planet masses in Earth units, rounded to the nearest integer. Planet colors indicate approximate densities, with values taken from Buchhave et al. (2016) for Kepler-20 and from Lissauer et al. (2013) for Kepler-11. Buchhave et al. propose that Kepler-20e and -20f have Earth-like compositions, given their similarity in size to Venus and Earth. The radius and composition of Kepler-20g are informed guesses, as are the mass and composition of Kepler-11g.

Observations of these two benchmark systems have continued since their discovery. A follow-up study using additional quarters of Kepler data revised the masses and radii of the Kepler-11 planets, mostly downward, resulting in even puffier planets (Lissauer et al. 2013). Another follow-up study reported radial velocity data on Kepler-11 obtained by Keck/HIRES in 2014 (Weiss et al. 2015), placing an upper limit of twice the mass values derived by Lissauer & colleagues.

Now Buchhave & colleagues (2016) have refined the masses of the known planets around Kepler-20 and validated a sixth planet by analyzing archival HIRES data and new HARPS-N data. Figure 2, above, is a graphic comparison of the two systems based on current findings; Table 1, below, provides comparative numbers.

The most striking difference between the planetary systems of Kepler-11 and Kepler-20 appears in the bulk compositions of their planets. The estimated density of Kepler-20b substantially exceeds that of Earth, the densest planet in our system. Its composition might be explained by an enhancement in iron. Likewise, despite their similarity in mass to Uranus, Kepler-20c and -20d are much denser, indicating a smaller bulk percentage of H/He and an enrichment in metal, rock, and potentially ice. Conclusive data are lacking for the other three planets of Kepler-20, but planet g is likely to be similar in composition to planets c and d, while planets e and f might resemble our system’s terrestrial planets.

All the planets of Kepler-11 are less massive than Uranus, but two of them (b, d) have densities comparable to Neptune and Uranus, respectively. The other three planets with estimated densities (c, e, f) are comparable to Saturn, the most rarefied planet in our system, despite masses in the range of 2 to 8 Mea. For context, Saturn is 95 Mea, and if our ringed planet fell into an ocean large enough, it would float. Half the Kepler-11 planets could float along with it, bobbing like rubber ducklings after a colossal rubber duck.

kepler-20: six-planet architecture

The new study by Buchhave & colleagues (hereafter B16) begins by reexamining the properties of the host star. For Kepler-20, they report a higher mass (0.948 Solar masses or Msol), a slightly higher metallicity (+0.07 ±0.08), and an earlier spectral type (G2) than did earlier sources. These new parameters imply an even closer resemblance among Kepler-20, Kepler-11, and our own Sun than previously indicated. Kepler-20 and Kepler-11 are now assigned virtually identical masses and metallicities, while both appear older and slightly less massive than our Sun.

The most significant contribution from B16 is their discovery of the new planet, Kepler-20g. Notably, Hansen & Murray (2013) previously found that a stable orbit might be available between planets f and d, and the HARPS-N radial velocity data have confirmed their hypothesis. In addition, B16 report a more precise mass for Kepler-20d, which remains the outermost of the known planets. Its new mass value (10 Mea) is just half the upper limit reported in the discovery paper (Gautier et al. 2012), although the precision of that estimate is still inferior to those for planets b and c.

Other major results from B16 support the basic picture unveiled by the discovery paper, including null results for TTVs. As before, we see a distinctive mass distribution in which smaller, more lightweight planets alternate with larger, more massive planets inside 0.25 AU. The two smallest planets (e, f) are remarkably similar in radius – and probably in mass and composition – to the terrestrial planets of our Solar System. Their diminutive profiles are inconsistent with H/He envelopes, whereas such envelopes are essential to explain the radii of the two largest planets (c, d). We are also safe in assuming an H/He atmosphere for the most massive planet (g).

The innermost planet (b), however, appears to be rocky, without any contribution from water or H/He. Kepler-20b is therefore the most massive rocky planet discovered to date. All other objects with similar radii and well-constrained masses – with the possible exception of 55 Cancri e – require substantial volatile content to explain their profiles.

Unfortunately, B16 did not present stability limits for any additional planets that might orbit outside 0.4 AU, in the cooler region where the system’s habitable zone is located. As a result, the potential of Kepler-20 to support habitable planets remains unexplored.

Table 1. Kepler-20 and Kepler-11 System Parameters

Masses and radii are expressed in Earth units. Mass is rounded to a single decimal place, with uncertainties omitted. Radius is rounded to two decimal places. a = semimajor axis, expressed in astronomical units (AU), where 1 AU = separation between Earth and Sun. Period = days. Teq = equilibrium temperature, expressed in Kelvin (K); for context, the Teq of Earth is 255 K. Density = grams/cc. Data on Kepler-20 are from Buchhave et al. 2016, except for Teq values in parentheses, which are older estimates from the Kepler Table based on a lower stellar effective temperature. Data on Kepler-11 are from Lissauer et al. 2013.

kepler-11 and kepler-20: formation scenarios

Although they do not discuss any specific formation models for the planets of Kepler-20, B16 opine that all six planets assembled shortly before the dissipation of the primordial circumstellar nebula. This timing enabled all six to accrete small amounts of H/He from the nebula without initiating a runaway process that would turn them into gas giants. After the nebula dispersed, stellar flux likely ablated the H/He envelopes from planets b, e, and f – the former because of its star-hugging orbit, the latter two because of their small masses. The cooler and heavier planets (c, d, g) were able to retain their lightweight atmospheres, consistent with current models of atmospheric evolution (Lopez & Fortney 2014, Erkaev et al. 2016).

The release of B16 was almost simultaneous with the publication of a new study of possible evolutionary scenarios for Kepler-11 (D’Angelo & Bodenheimer 2016). Given the similarities between the two host stars, a scenario that can explain one system architecture should also tell us something about the other.

In their new study, Gennaro D’Angelo and Peter Bodenheimer (hereafter DB16) test the two most popular formation processes for close-in planets: 1) in situ accretion of solids locally available in the inner nebula, and 2) accretion of solids over a broad radial distance by a planetary core migrating from the outer nebula to the vicinity of the central star. For brevity they call the latter process “ex situ formation.” For each scenario they seek initial conditions that permit the formation of the known planets within currently observed parameters, and in both cases they achieve some degree of success. Accordingly, they conclude that “it is not possible to distinguish between the two modes of formation from [the planets’] final properties.”

However, a review of their models suggests that ex situ is more plausible than in situ, given two major flaws in the in situ scenario. First, DB16 find that a protoplanetary nebula massing 0.18 Msol inside 70 AU would be required to achieve the concentration of solid mass needed for in situ formation of the Kepler-11 planets inside 0.5 AU. This total is equivalent to the mass of a mid to late M dwarf star. Yet observations of protoplanetary nebulae in nearby star-forming regions indicate that most have masses in the range of 0.002 to 0.01 Msol (Williams & Cieza 2011, Andrews et al. 2013). Such findings argue that the in situ nebula invoked by DB16 is unrealistic.

In the second place, even assuming the existence of a protoplanetary nebula more massive than Proxima Centauri, the in situ model still could not produce a good analog of Kepler-11b. Since the inner nebula is completely dry in this model, the six planets originally form as rock/metal cores surrounded by H/He atmospheres. No water or other volatile materials are available to enhance their composition. Given the low core mass and tight semimajor axis of Kepler-11b, DB16 find that the planet’s primordial atmosphere would be stripped by stellar flux within 40 million years after the evaporation of the ambient nebula. To achieve the puffy radius observed today, some 8 billion years later, the planet would have to outgas very large quantities of H/He from its interior over a period lasting a few hundred million years – first to replace its original envelope, and later to replenish the outgassed atmosphere, which would remain vulnerable to stripping during the time it takes for a G star to settle into maturity (Erkaev et al. 2016). DB16 acknowledge the difficulties involved in this outcome.

Their ex situ or migratory model avoids both pitfalls. They begin with a protoplanetary nebula of 0.03 Msol inside a radius of 60 AU. Although this value is larger than a typical nebular mass, it still falls within an order of magnitude of observations, and is therefore substantially more realistic than the in situ nebula. Within this structure DB16 insert newly formed planetary cores of approximately Martian mass (0.1 Mea) at orbital radii ranging from 2.1 AU to 5.35 AU, in the region of the nebula where ices are abundant. All cores begin accreting mass and migrating inward. The timing of their insertion in the nebula is just as important as their radial placement, since cores that achieve smaller final masses need to begin accreting later than those that achieve larger masses in order to avoid orbit crossings during migration.

Because they originate in a well-hydrated region, all six forming planets accrete abundant water along with their H/He envelopes, and all are subject to loss of atmosphere once they reach their final destinations and the nebula dissipates. With this model, DB16 succeed in reproducing the masses and radii of all six planets of Kepler-11. In particular, they find that Kepler-11b retains a steam atmosphere after the loss of its lightweight H/He envelope, and that this heavier atmosphere can survive until the present age of the system, consistent with the planet’s current radius.

DB16 underscore the difference in planetary compositions produced by their two contrasting approaches to system evolution. Although they do not apply the term, the water-rich bodies resulting from their ex situ model are consistent with the Ocean Planets predicted by Leger et al. (2004).

The approach taken by DB16 is paralleled in some of the studies that accompanied the recent announcement of Proxima Centauri b (Barnes et al. 2016, Coleman et al. 2016), as discussed in my previous post. Like DB16, Coleman & colleagues explored several scenarios for planet formation, including in situ accretion in an implausibly massive protoplanetary disk, as well as long-distance migration in a more realistic disk. As with DB16, the migration scenario provided a better fit with observations than did in situ formation. Notably, Coleman’s group explicitly invoked Ocean Planets to describe the objects produced by migration.

I look forward to a study that applies similar models to the dual evolutionary histories of Kepler-20 and Kepler-11. I suspect that in situ scenarios will continue to face difficult challenges.

the high-multiplicity sample

With the confirmation of Kepler-20g, Kepler-20 moves from one exclusive club – systems with at least five known planets – to the even smaller elite – systems with at least six planets. The current exoplanetary census offers only four examples of this rare architecture: HD 10180, Kepler-11, Kepler-90, and now Kepler-20. Their orbital arrangements provide invaluable data that will continue to inform our understanding of planet formation and the distribution of specific planetary types (including temperate rocky planets) in our region of the Galaxy.


Andrews SM, Rosenfeld KA, Kraus AL, Wilner DJ. (2013) The mass dependence between protoplanetary disks and their stellar hosts. Astrophysical Journal 771, 129.
Barnes R, Deitrick R, Luger R, Driscoll PE, Quinn TR, Fleming DP, Guyer B, McDonald DV, Meadows VS, Arney G, Crisp D, Domagal-Goldman SD, Lincowski A, Lustig-Yaeger J, Schwieterman E. (2016) The habitability of Proxima Centauri b I: Evolutionary scenarios. In press. Abstract: 2016arXiv160806919B
Buchhave LA, Dressing CD, Dumusque X, Rice K, Vanderburg A, Mortier A, Lopez-Morales M, Lopez E, et al. (2016) A 1.9 Earth radius rocky planet and the discovery of a non-transiting planet in the Kepler-20 system. In press. 2016arXiv160806836B
Coleman GAL, Nelson RP, Paardekooper SJ, Dreizler S, Giesers B, Anglada-Escude G. (2016) Exploring plausible formation scenarios for the planet candidate orbiting Proxima Centauri. Monthly Notices of the Royal Astronomical Society, in press. Abstract: 2016arXiv160806908C
D’Angelo G, Bodenheimer P. (2016) In situ and ex situ formation models of Kepler 11 planets. Astrophysical Journal 828, 33. Abstract: 2016ApJ...828...33D
Erkaev NV, Lammer H, Odert P, Kislyakova KG, Johnstone CP, Gudel M, Khodachenko ML. (2016) Thermal mass loss of protoplanetary cores with hydrogen-dominated atmospheres: The influences of ionization and orbital distance. Monthly Notices of the Royal Astronomical Society 460, 1300-1309.
Ford EB, Rowe JF, Fabrycky DC, Carter JA, Holman MJ, Lissauer JJ, et al. (2011) Transit timing observations from Kepler. I. Statistical analysis of the first four months. Astrophysical Journal Supplement Series 197, 2.
Fressin F, Torres G, Rowe JF, Charbonneau D, Rogers LA, Ballard S, Batalha NM, Borucki WJ, Bryson ST, Buchhave LA, et al. (2012) Two Earth-sized planets orbiting Kepler-20. Nature 482, 195-198. Abstract: 2012Natur.482..195F
Gautier TN, Charbonneau D, Rowe JF, Marcy GW, Isaacson H, Torres G, Fressin F, Rogers LA, Desert J-M, Buchhave LA, et al. (2012) Kepler-20: A Sun-like star with three sub-Neptune exoplanets and two Earth-size candidates. Astrophysical Journal 749, 15. Abstract: 2012ApJ...749...15G
Hansen B, Murray N. (2013) Testing in situ assembly with the Kepler planet candidate sample. Astrophysical Journal 775, 53. Abstract: 2013ApJ...775...53H
Leger A, Selsis F, Sotin C,  Guillot T, Despois D, Mawet D, Ollivier M, Labeque A, Valette C, Brachet F, Chazelas B, Lammer H. (2004b) A new family of planets? “Ocean Planets.” Icarus 169, 499-504.
Lissauer JJ, Fabrycky DC, Ford EB, Borucki WJ, Fressin F, Marcy GW, et al. (2011a) A closely packed system of low-mass, low-density planets transiting Kepler-11. Nature 470, 53-58. Abstract: 2011Natur.470...53L
Lissauer JJ, Ragozzine D, Fabrycky DC, Steffen JH, Ford EB, Jenkins JM, et al. (2011b) Architecture and dynamics of Kepler’s candidate multiple transiting planet systems. Astrophysical Journal Supplement Series 197, 8.
Lissauer JJ, Jontof-Hutter D, Rowe JF, Fabrycky DC, Lopez ED, Agol E, et al. (2013) All six planets known to orbit Kepler-11 have low densities. Astrophysical Journal 770, 131. Abstract: 2013ApJ...770..131L
Lopez E, Fortney J. (2014) Understanding the mass-radius relation for sub-Neptunes: Radius as a proxy for composition. Astrophysical Journal 792, 1.
Weiss LM, Marcy GW, Isaacson H, Deck KM, Lissauer JJ, Jontof-Hutter D. (2015) Constraining the masses of the Kepler-11 planets with radial velocities. Abstract presented at Physics of Exoplanets: From Earth-sized to Mini-Neptunes, February 23-27, 2015.
Williams JP, Cieza LC. (2011) Protoplanetary disks and their evolution. Annual Review of Astronomy and Astrophysics 49, 67-117. Abstract: 2011ARA&A..49...67W