Saturday, November 24, 2012

Dwarfs vs. Giants

Figure 1. In the early stages of system evolution, young planets may be engulfed by their host stars.
Credit: ESA / L. Calcada

To understand the potential architectures of a broad range of planetary systems – not to mention the likelihood that a given architecture might support habitable planets – we’ve found that mass matters. Low-mass exoplanets resembling Earth and Neptune are often found in the company of similar objects, whereas gas giants like Jupiter typically have no near neighbors (Steffen et al. 2012, Johansen et al. 2012).

A new study by Soko Matsumura and colleagues (2012) investigates the relationship between the orbital dynamics of extrasolar gas giants and the possibility of Earth-mass planets in the system habitable zone (the orbital space where temperatures permit surface bodies of water on rocky planets). Matsumura’s group builds on a pair of studies by Dimitri Veras and Philip Armitage (2005, 2006), whose models predict that systems with eccentric gas giants, even on relatively wide orbits, will be deficient in rocky planets on smaller orbits (2006).

battle of the giants

Matsumura and colleagues ran 40 simulations of young planetary systems containing 3 Jupiter-mass gas giants (1 Mjup each) orbiting between 3.8 and 6.5 AU. Each system also had 11 Earth-mass planets (1 Mea each) between 0.1 and 3 AU. All planets initially traveled on circular orbits.

In all runs, the gas giants rapidly experienced dynamical instabilities that reduced their number from three to two. Giant subtraction followed three possible pathways, in order of likelihood: the merger of two giants (43%), the ejection of at least one giant (37%), or the engulfment of at least one giant by the host star (20%). More than one pathway might be active in a single system. Except for runs resulting in mergers, the surviving giants had eccentric orbits. In two-thirds of systems, two giants remained, but in one-third there was only one.

collateral damage

Violent episodes of planet scattering typically wreak havoc on smaller, warmer planets, despite the latter’s status as innocent bystanders. Most destructive is a giant sweeping through the inner system to collide with the host star; in this case, all inner planets are likely to be annihilated. An ejected giant is almost as harmful. Even if the inner planets are initially untouched, long-term perturbations by surviving giants on eccentric orbits eventually drive most inner planets into the star or out of the system. Planets with the smallest semimajor axes (~0.1 AU) are the most likely to survive.

If the Solar System had suffered one of these catastrophes, we would have lost Mars, Venus, Earth, and maybe even Mercury.

reality check

Matsumura’s group compared the results of their simulations with current data on multiplanet systems that have retained both high- and low-mass planets. According to their models, these systems either never experienced violent instabilities or, if they did, managed to survive an epoch of planet scattering. They found that mixed-mass systems tend to have low eccentricities, consistent with peaceful dynamic histories.

They also note that this configuration is the exception rather than the rule among exoplanetary systems, since they calculate that just 5% to 10% of multiplanet systems (considering both HARPS and Kepler results) contain both gas giants and low-mass planets.

three-way slice

The findings of Matsumura’s group encouraged me to look at mixed-mass systems as a potential architectural category, specifically in contrast to multiplanet systems containing either low-mass planets only or gas giants only. Table 1 compares parameters for each of these three categories, using data on all confirmed, well-constrained multiplanet systems listed in the Extrasolar Planets Encyclopaedia at the end of October. For gas giants, the lower mass limit is 0.17 Mjup (55 Mea) and the lower size limit is 7 Earth radii (7 Rea). For low-mass planets, mass values derive from radial velocity or transit timing data, as available. For Kepler planets with neither type of data, maximum masses are estimated from planet radii.

Table 1. Summary of 104 multiplanet systems, binned by architectural category

* Values for orbital eccentricity and stellar metallicity are unavailable for most low-mass systems.

For many parameters, these three architectural categories appear to define a continuum. The progression is most obvious for semimajor axes, whose median values increase from one category to the next. Multiple systems with low-mass planets are the most compact, while those with high-mass planets are the most far-flung. Most low-mass and mixed-mass systems have one or more planets orbiting within 0.1 AU, but only 16% of the high-mass systems have such an object. Once again we see evidence that Hot Jupiters tend to be solitary planets, shunning even their own kind.

An analogous progression from least to most is visible in star mass, even if the contrasts are less striking. Low-mass stars tend to host low-mass planets, Solar-mass stars may host both low- and higher-mass planets, and high-mass stars typically host especially massive gas giants. No similar trend is apparent for star metallicity. Instead of a steadily rising enhancement in metals, we see a sharp cutoff between systems with and without gas giants, such that the former have super-Solar metallicities and the latter sub-Solar metallicities. High metallicity and large planet mass do not appear to be correlated.

Another unexpected result is the similarity between mixed-mass and high-mass systems in terms of orbital eccentricity. The conclusions of Matsumura’s group might suggest a more striking divide. Even so, the high-mass systems typically feature more eccentric orbits than the other two types. Since high-mass systems also support planets at much wider semimajor axes, their formation processes were probably most active in regions beyond 1 AU.

Table 1 was arbitrarily constructed in terms of planet mass, so the rising trend for this variable may simply reflect the initial setup. Nevertheless, the median planet mass for each architectural category suggests that the trend is significant. For low-mass systems, the median is about 10 Mea, near the dividing line between primarily rocky planets and planets with substantial fractions of hydrogen and volatiles. For mixed-mass systems, the median is about 55 Mea, near the boundary between gas dwarfs like Neptune and gas giants like Jupiter. Planets of Jupiter mass or more are rare in this type of system. But for high-mass systems, the median mass is 1.82 Mjup, well above the median for all known extrasolar gas giants (1.5 Mjup in a sample of 574 with mass > 0.16 Mjup). 

A key difference between mixed-mass and high-mass systems may be that the former tend to have lightweight gas giants on smaller orbits, while the latter generally have far more massive giants on wider, more eccentric orbits.

A key similarity across all three architectural types is the tendency for lower-mass planets to follow smaller orbits than higher-mass planets. Thus, in all multiplanet systems, we see a strong trend for the innermost planet to be the most lightweight and the outermost to be the most massive. The Solar System also follows this trend within a semimajor axis of 6 AU (which incidentally exceeds the widest orbits in all but 5% of the systems sampled in Table 1).

We might expect the mixed-mass systems to be progressively mass-boosted versions of the low-mass systems, containing varying proportions of gas giant and low-mass planets, but what we see is a clear domination of the dwarfs by the giants. Fifteen out of 18 mixed-mass systems contain only one low-mass planet, which always has the shortest period in the system. Two others (GJ 876, Kepler-30) feature a low-mass planet orbiting inside one or two gas giants with a second low-mass planet orbiting outside the giant or giants. Just one system – HD 10180 – contains a procession of low-mass planets (at least 6 in all) climaxing in a single small gas giant at a larger semimajor axis.


It will be interesting to see if the mass-based architectonics implied by my rather arbitrary lump-and-split approach is borne out by future theory and observation. Will we begin to find gas giants in the outer regions of systems that now look like compact collections of telluric and gas dwarf planets? Will we begin to detect short- or long-period Neptune-mass planets in systems where we now see only eccentric gas giants? Will we find reason to believe that truly Earth-like planets (rocky objects with surface water) are possible only in the rare subset of systems that support analogs of Jupiter and Saturn? If our governments keep funding space-based astronomical missions, we may start to get answers soon.

Meanwhile, the divide between low-mass and high-mass planets shows all signs of being a fundamental split, as transit data in particular continue to confirm. Figure 2 below updates the graph I first posted in April, using data from the Extrasolar Planets Encyclopaedia. The sample of well-constrained low-mass planets continues to grow rapidly, and the tendency for these objects to support hydrogen atmospheres is robust and unmistakable, given the preponderance of planets with radii between 2 and 5 Rea.

Figure 2. All transiting planets smaller than 15 Earth radii (1.34 Jupiter radii) as of October 16, 2012

Abbreviations: E = Earth, U = Uranus, S = Saturn, J = Jupiter


Johansen A, Davies MB, Church RP, Holmelin V. (2012) Can planetary instability explain the Kepler dichotomy? Astrophysical Journal 758, 39. Abstract: 
Matsumura S, Ida S, Nagasawa M. (2012) Effects of dynamical evolution of giant planets on survival of terrestrial planets. In press; abstract:
Steffen JH, Ragozzine D, Fabrycky DC, Carter JA, Ford EB, Holman MJ, Jason F. Rowe JF, Welsh WF, Borucki WJ, Boss AP, Ciardi DR, Quinn SN. (2012) Kepler constraints on planets near hot Jupiters. Proceedings of the National Academy of Sciences 109, 7982-7987. Abstract:
Veras D, Armitage PJ. (2005) The influence of massive planet scattering on nascent terrestrial planets. Astrophysical Journal 620, L111-L114. Abstract:
Veras D, Armitage PJ. (2006) Predictions for the correlation between giant and terrestrial extrasolar planets in dynamically evolved systems. Astrophysical Journal 645, 1509-1515. Abstract: