sign in sign up
<<
Home , 35 Pegasi, Exoplanetology, Gemini Planet Imager, Hot Jupiter, Hot Neptune, HR 8799

Structure of

David S. Spiegela,1, Jonathan J. Fortneyb, and Christophe Sotinc

aSchool of Natural Sciences, Department, Institute for Advanced Study, Princeton, NJ 08540; bDepartment of and Astrophysics, University of California, Santa Cruz, CA 95064; and cJet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109

Edited by Adam S. Burrows, Princeton University, Princeton, NJ, and accepted by the Editorial Board December 4, 2013 (received for review July 24, 2013) The hundreds of exoplanets that have been discovered in the entire radius of the in which opacities are high enough past two decades offer a new perspective on planetary structure. that heat must be transported via convection; this region is Instead of being the archetypal examples of , those of our presumably well-mixed in chemical composition and specific are merely possible outcomes of entropy. Some gas giants have heavy-element cores at their centers, formation and evolution, and conceivably not even especially although it is not known whether all such planets have cores. common outcomes (although this remains an open question). Gas-giant planets of roughly ’s occupy a special Here, we review the diverse range of interior structures that are region of the mass/radius plane. At low , liquid or rocky both known and speculated to exist in exoplanetary systems—from planetary objects have roughly constant and suffer mostly degenerate objects that are more than 10× as massive as little compression from the overlying material. In such cases, 1=3 Jupiter, to intermediate-mass -like objects with large cores Rp ∝ Mp , where Rp and Mp are the planet’s radius mass. At high and moderate / envelopes, to rocky objects with masses, for objects that have had time to cool, electron de- roughly the mass of . generacy pressure becomes significant, and the mass/radius re- −1=3 lation changes such that the radius scales as Rp ∝ MP . Note gas giants | hot | super- that the H/He comp. curve in Fig. 1 does not display this be- havior at low masses, because this curve is calculated for highly ow can we, from many light away, learn about the irradiated objects that are a mere 0.045 a.u. from their -like Hinterior structure of exoplanets? observa- , which prevents them from reaching their zero-temperature tions provide minimum masses of exoplanets; observa- radius in 3 (3 Gyr). For cold spheres of H/He, it has tions provide planet radii, and taken singly, neither is especially long been appreciated that the maximum in the mass/radius re- informative about planet structure. However, when we know lation occurs near 4× Jupiter’s mass (9, 10), with a broad peak at both the mass and the radius of a planet we may learn much just over 1 RJ (where RJ is Jupiter’s radius), extending from ∼1/ more about the interior structure. The Kepler satellite (1, 2) is 10th of Jupiter’s mass to ∼100× Jupiter’s mass. For a planet to be a space-based, transit-detecting mission that, as of early 2013, smaller than the H/He minimum radius requires the presence of has identified ∼100 planets and ∼3,000 planet candidates, of a significant heavy-element component, in the form of a core in which the vast majority are almost certainly real (3, 4). Thanks to the object’s center or high- material well mixed Kepler and ground-based efforts such as the Hungarian Auto- throughout the envelope (where “metals” is taken to mean ele- mated (HAT) and Super Wide Angle Search for ments heavier than helium). For a planet to be larger than the Planets (SuperWASP) transit surveys (5, 6), there are now more zero-temperature radius (∼1 RJ) requires it to have a high enough than 200 known planets with measured masses and radii, span- temperature that pressure from ions becomes a significant frac- ning a range of irradiation conditions. tion of that due to electrons. Fig. 1 portrays how planet radii relate to masses and incident Before the discovery of transiting planets, it was assumed that fluxes, among the known planets (including the solar system approximately billion--old planets—even ones on close-in Kepler Kepler planets) and the candidates [ objects of interest around their (so-called “hot Jupiters,” with orbits (KOI)] (7, 8). Several trends are apparent in the data: planets with the largest radii tend to be near Jupiter’s mass and highly Significance irradiated, and Kepler seems to find low-radius planets at a wide range of orbital separations, including some small planets that are extremely highly irradiated. Planets around other stars, or exoplanets, are now known to In the remainder of this paper, we discuss what is known of the be common in our galaxy. Exoplanets span a much wider range structure of the massive planets (gas giants), of intermediate- of physical conditions than the planets in our solar system, and mass planets (), and of low-mass planets (terrestrial include extremely puffy gas giants to compact rocky planets and ocean planets). We conclude by considering how our that can have as high as that of . The diversity of knowledge of structure might improve over course exoplanets allows us to place the planets of our solar system in of the next decade. a broader perspective, and invites further research in the na- scent field of comparative . Here we review the Gas Giants continuum of detected planets, from the gas giants, which are Jupiter and are essentially giant spheres of hydrogen and composed mostly of hydrogen, to smaller worlds where helium (H/He) with smaller contributions from heavier elements may provide more of their mass, to rock-iron worlds in some and complex molecules. Many of the known exoplanets have ways similar to Earth. similar mass and similar radius to our local gas giants, and Author contributions: D.S.S., J.J.F., and C.S. designed research, performed research, con- probably have roughly similar bulk structure. The , or tributed new reagents/analytic tools, analyzed data, and wrote the paper. weather layer, of such an object is a thin outer region that is of The authors declare no conflict of interest. roughly the same relative depth as the skin of a grapefruit, and is – This article is a PNAS Direct Submission. A.S.B. is a guest editor invited by the Editorial typically defined as the region above the radiative convective Board. boundary,* which can be at pressures on the order of kilobars 1To whom correspondence should be addressed. E-mail: [email protected]. (kbar) for the most strongly irradiated planets and which occurs – in the vicinity of ∼1 kbar in gas giants subjected to lower irra- *The radiative convective boundary is the region bounding the convective interior of – a planet in which heat transport is dominated by convective eddies. This boundary diation, such as Jupiter and Saturn. Below the radiative con- occurs essentially where the vertical temperature gradient becomes subadiabatic and vective boundary, there is a deep envelope extending almost the therefore stable against convection.

12622–12627 | PNAS | September 2, 2014 | vol. 111 | no. 35 www.pnas.org/cgi/doi/10.1073/pnas.1304206111 Downloaded by guest on September 26, 2021 A 2.5 B 2.5 SPECIAL FEATURE 3e+10 25.00

1e+10 Flux (erg Radius vs. Mass 4.600 Mass (M Radius vs. Flux 3e+09 2 1e+09 2 0.855 KOI 3e+08 Solar System 0.157 Solar System / cm 1e+08 Planet Jupiter Planet 3e+07 0.029 ) ) 1.5 2 1.5 1e+07 / s) 0.005 ) 3e+06 Jupiter Jupiter 1e+06 0.001 R[M] for H/He comp. J 1 J

1 R (R R (R 3Gyr old, 0.045AU from Sun S S

0.5 0.5 R[M] for 100% water comp. UN R[M] for Earth−like comp. N U R[M] for 100% iron comp. V E M E V 0 0 M 3 4 5 6 7 8 9 10 11 12 0.001 0.01 0.1 1 10 10 10 10 10 10 10 10 10 10 10 M (M ) Jupiter F (erg/cm2/s)

Fig. 1. Radius vs. mass (A) and radius vs. incident-irradiating flux (B) for the confirmed exoplanets, the Kepler candidate planets (KOI), and the planets of our solar system. The ∼200 confirmed planets in this figure are represented with filled large circles; KOI (B) are represented with small white circles; solar system planets are represented with large pentagrams. For comparison, A shows the radius vs. mass relationship for Earth-like composition, which precisely matches the (mass, radius) values for and Earth (26). Planets span masses from less than Earth’stotensoftimesJupiter’s, radii from less than Earth’stomorethandouble − − Jupiter’s, and incident fluxes from nearly zero—in the case HR 8799b, at ∼70 a.u. (27)—to nearly 1012 erg·cm 2·s 1. The planets with the largest radii tend to be close to Jupiter’s mass and highly irradiated.

lasting ∼1 week or less)—would have cooled and shrunk to near essentially impossible to learn whether the extremely inflated their asymptotic radius (11, 12). However, the discovery of transiting planets have cores (although, if they do, the larger the core mass, ASTRONOMY hot Jupiters such as HD 209458b (13), with its &1.3 RJ radius, the greater the additional power that is required to explain their suggested that some planets are a good deal puffier than expected radii). Some of the known transiting gas giants, however, have (14). Many mechanisms have been suggested in the literature to smaller radii (at their known masses) than an H/He composition explain the apparently inflated radii of some of the hottest hot can produce. These planets must have a significant heavy-ele- Jupiters (15, 16); what these mechanisms all have in common is that ment component. The inferred metal fraction appears to be they increase the bulk entropy of the planet above what might have † correlated with the metallicity of the planet’s host star (20, 36), been expected from naive cooling models. This increase can come i suggesting that more metal-rich protoplanetary environments either from ( ) retarding the evolutionary cooling [via enhanced lead to more metal-rich planetary compositions, perhaps in the atmospheric opacity or modified atmospheric thermal profiles form of rocky cores. (19–21)] or (ii) from additional power added in the planet’s deep Some exotic objects orbiting other stars do not have direct interior [via tidal (22, 23) or ohmic (24, 25) dissipation]. structural analogs in our solar system. One planet that falls be- It remains uncertain whether a single inflation mechanism predominates. For the most highly inflated objects, however, tween our local archetypal categories is the enigmatic HD mechanisms of type i might not work. Close-in planets are pre- 149026b (37); although 20% more massive than Saturn, its radius sumably tidally locked (28) and have a permanent night side is 22% smaller, which suggests that its heavy-element content is through which they cool more efficiently (19, 29); accounting for greater than the entire mass of metals (outside the Sun) in the ∼ − -M the enhanced night-side cooling efficiency suggests that mecha- solar system, in the range of 60 110 ⊕ of metals (20, 38). In nisms that merely reduce the cooling rate might not be able to this respect, despite its greater-than-Saturn mass, this planet is explain the highly inflated planets, and some power must be perhaps more similar in structure to - and Neptune-like deposited in the deep interior (30). Despite the lack of a con- planets, the subject of the following section. sensus mechanism, it is clear that objects must either be quite Neptunes young (31) or very highly irradiated (32, 33) (Fig. 1) to have significantly inflated radii. The lowest irradiation experienced Giant planets where most of the planet’s mass is composed of by planet with a radius more than 1.5× Jupiter’sisthe∼109- heavy elements, rather than mostly H/He gas, begin our transi- − − erg·cm 2·s 1 incident on Kepler-12b, which still exceeds the tion to our next class of planets. These so-called “Neptune-class” solar flux upon Jupiter by a factor of 104. planets still have a thick H/He envelope, but the light-element One of the key structural uncertainties for many of the known envelope does not comprise the majority of the planet’s mass. In gas-giant planets is whether they have cores in their centers. (It is our solar system, Uranus (14:5 M⊕ and 4:0 R⊕), and Neptune also not known, at present, whether Jupiter has a heavy-element (17:1 M⊕ and 3:9 R⊕) are our examples of these planets. Fig. 2 core at its center, although Saturn must have one.) The presence portrays how the bulk structure of a Neptune-class planet differs or absence of cores is of interest both as it relates to our from that of either a Jupiter or an Earth (which we will address knowledge of the planets themselves and because it bears upon in the next section). — their formation mechanism whether by a runaway process of Uranus and Neptune are generally known for their bluish ∼ -M accreting gas onto 10 ⊕ cores (34) or via gravitational in- color and are often lumped together as two “ice giants” because stability of the protoplanetary gas disk (35). Unfortunately, it is most structure models find that the majority of the is in a deep fluid ionic probably consisting pre-

† dominantly of water, and also containing ammonia and . There are some uncertainties in the proper equation of state for H/He/metals mixtures (17, 18), but the attendant uncertainties in radius are insufficient to explain the very Though the planets appear outwardly very similar, there is ample ≳ ’ large radii ( 1.5 RJ) occasionally seen. evidence that the planet s interiors are quite different. The diversity

Spiegel et al. PNAS | September 2, 2014 | vol. 111 | no. 35 | 12623 Downloaded by guest on September 26, 2021 Fig. 2. Pie-slice diagrams of a representative gas-giant (Jupiter), ice-giant (Neptune), and terrestrial (Earth) planets. Jupiter’s outer ∼20% in radius consists mostly of molecular hydrogen. Pressures and temperatures quickly increase inside the planet, and the bulk of Jupiter is made of liquid , a form of liquid hydrogen that is highly electrically conductive; at its center there may be a concentration of heavier elements, but constraints are not firm. The visible atmosphere, and perhaps the entire H/He envelope, is enhanced in metals by a factor of 3–5 compared with the Sun. Within Neptune, most or perhaps all of the hydrogen is found in molecular form, whereas the bulk of the planet’s mass, the “middle” gray layer, is composed of fluid-conducting

molecules and molecular fragments probably consisting of H2O, NH3, and CH4. Constraints on the interior rock-to-ice ratio are not firm, and it is not clear how distinct the layer boundaries actually are. The visible atmosphere is enhanced in by a factor of ∼50 compared with the Sun, but there is little constraint on other elements. Within the Earth, the is composed of silicates whose mineralogy differs between the upper and the lower mantle. The distinct core is mostly iron-nickel, with small admixtures of unknown lighter elements. The solid core is currently growing at the expense of the liquid core. Most ofthe interiors of all of the planets are thought to be convective.

within our two Neptune-class planets should be a clear reminder For instance, one could model the planet with only H/He and that this class of exoplanets should harbor tremendous diversity. rock, or with only H/He and water, ignoring a third component. First, both planets do not simply have homogeneous three- Another complication is that one needs to be able to model the layer structures with an H/He upper envelope, water-dominated thermal evolution of the planet to understand the deep interior middle envelope, and rocky core. Neither planet is as centrally temperatures and densities of the possible components of this condensed as this often-suggested but too-simple picture would several-billion-year-old planet. Degeneracy in inferred composi- imply. Uranus is more centrally condensed that Neptune; more tion for Neptune class exoplanets will always be the rule (46, 47). dramatically, it also has a heat flux from its deep interior that is Diversity in exo-Neptune interior structure can be seen from no more than 10% that of Neptune, which may be due to deep a comparison of GJ 436b to Kepler-30d. Models of GJ 436b composition gradients that suppress large-scale convection. generally suggest an interior that is 80–90% heavy elements Uranus is also flipped over on its side with its spin axis nearly in (similar to Uranus and Neptune). However, Kepler-30d has its orbital plane, which might imply that the stochastic of nearly the same mass ð23:1 M⊕Þ but a radius twice as large giant collisions near the end of the planet-formation era plays ð8:8 R⊕Þ (48), which suggests that the planet is only 30% heavy a major role in determining the structure of this class of planet. elements and 70% H/He gas, thereby already breaking our “rule” Even today, much of our knowledge of the structure and evolution that planets of this class (mass range) should not be made pre- of these planets remains uncertain and provisional (39–41). dominantly of hydrogen. Both planets have H/He that are strongly Kepler data show a dramatic increase in planetary occurrence enriched in metals. Only the carbon abundance (in methane) can from 6 R⊕ to 2 R⊕, indicating that “sub-Neptunes” from 2-3 R⊕ be measuredly fairly definitively via spectroscopy, and requires are a very common planetary type (4, 49). Such planets also gen- a carbon enhancement of ∼50× solar in each planet (42). Indirect erally need an H/He envelope, but one that comprises perhaps only evidence suggests that may be several hundred times solar ∼1–5% of the planetary mass. Several of these objects have masses in Neptune (43), which suggests a framework where the relatively in the 3-10 M⊕ range, indicating that even relatively small thin H/He envelopes of Neptune-class planets may well be ex- planets can accrete and maintain gaseous H/He envelopes, tremely enhanced in metals compared with their parent stars. despite ongoing evaporative mass loss (50). Whether Neptune-class exoplanets are true ice giants (mean- Though the majority of Kepler planets are around distant, faint ing that much of their mass is made up of water and other fluid stars, there is also a relatively nearby example of this class of sub- planetary ices) is an open question. If planets form beyond the Neptune planet, named GJ 1214b. This ∼500-K transiting planet ice line, where water has condensed (like all of our giant plan- orbits its cool star in a close-in , and its mass ð6:5 M⊕Þ and ets), then a substantial fraction of the mass of Neptune-class radius ð2:7 R⊕Þ are well determined (51) and seem to imply a planets is probably water. However, planets may also form within gaseous component atop a liquid/solid core (52, 53). A series of the ice line, in which case they would have rock/iron interiors observational campaigns have attempted to characterize the with an accreted gaseous envelope. visible H/He atmosphere of the planet, which would perhaps The first Neptune-class exoplanet with a measured mass and allow the composition of the entire H/He envelope to be con- radius was the transiting “” GJ 436b, which is strained. However, the observed transit–radius spectrum has 22:2 M⊕ and 4:3 R⊕ (44, 45). The planet’s location on the mass/ been nearly featureless (54, 55), and this suggests either that radius diagram implies that the planet must have a substantial cloud material is obscuring the atmosphere or that the atmo- H/He envelope and is not composed only of water, for instance. sphere is quite compact, so that it imprints little signal on the The relative mass fractions of H/He, water (and other icy com- stellar transmitted light. If the planet’s atmosphere is strongly ponents like ammonia and methane), rock, and iron cannot be enriched with metals, at a level even higher than Uranus and ascertained from mass and radius alone. Even with the as- Neptune, this could greatly increase the mean molecular mass sumption of a fixed rock:iron ratio, one can mix a wide array of and reduce the atmosphere’s vertical height, although such three-component compositions of rock/iron, water, and H/He. a high atmospheric mean-molecular weight is disfavored by the

12624 | www.pnas.org/cgi/doi/10.1073/pnas.1304206111 Spiegel et al. Downloaded by guest on September 26, 2021 observed mass and radius (53). Once the atmospheres of more of these planets have been probed in more detail, we can begin to SPECIAL FEATURE make firmer connections that could link the composition of the H/He envelope with planetary mass and orbital location. Below some mass ð.3 − 5 M⊕?Þ, cores are small enough that they either do not accrete nebular H/He gas or the gas that is accreted is quickly lost. Such objects are true terrestrial or water planets, often called super-Earths. We now turn our attention to these larger cousins of Earth and Venus. Terrestrial and Ocean Planets The search for Earth-like terrestrial planets is a major objective in the study of exoplanets. Terrestrial planets are objects such as , Venus, and that are composed predominantly of elements such as Si, Mg, Fe, O, and, perhaps in some cases, C Fig. 3. Schematic representation of the interior structure and dynamics of (56). Now that exoplanets have been discovered that are roughly the Earth (Left) and Venus (Right). On both planets, hot plumes form at the the size of Venus or Earth, one key question is whether a given core/mantle boundary, although the heat that comes out of core is a fraction ∼ terrestrial exoplanet is more similar to Venus or to Earth, as ( 25% for the Earth) of the total heat released at the surface. The miner- Terrestrial Planets alogical transformation at the upper/lower mantle interface may or may not discussed in below. However, other types of be a barrier to convection: (Left) a hot plume that would be stopped at this planets or satellites may be habitable and, in some sense, Earth- interface; (Right) how the plume moves up to the bottom of the conductive like. In the solar system, and share with the lid. The depth of this interface on Venus is larger than that of the Earth Earth the rare characteristic of possessing an ocean that is in because the gravity on Venus is lower. On Earth (Left), the oceanic litho- contact with a rocky interior. It is not known whether there is sphere (black) forms at the midocean ridges and is recycled in the mantle at on Enceladus or Europa, but their existence motivates the study subduction zones (adapted from refs. 63 and 72). of exoplanets that would have a large fraction of H2O, known as ocean planets (57). As discussed in Ocean Planets, it might be difficult to determine unambiguously that an exoplanet actually (67, 68). However, the yield strength depends on more than just is an ocean planet, because such worlds could occupy the same the size of the planet. Recent work has shown that the depth ASTRONOMY region of the ðMp=RpÞ plane as planets consisting of rocky cores dependence of the crustal yield strength, a poorly constrained surrounded by atmospheres of H/He (58). parameter, is critical in determining the convective regime and the likelihood of (65): the more the yield strength Terrestrial Planets. Venus and Earth have about the same mass increases with depth, the lower the probability of plate tectonics. and radius, and Venus’s slightly smaller density could be explained The yield strength and its variation with depth depend on surface by its smaller mass; that is, the uncompressed density is the same temperature, the presence of liquid water, the geological history for the two planets. However, Venus and Earth have evolved to of the planet (such as whether large impact craters have weakened present-day conditions so different as to invite the question of the lithosphere), and the presence of light, Earth-like continents whether distance to the Sun is the only parameter that drives such (Fig. 3), at whose border the lithosphere is weaker. Other geo- a different evolution. logical properties, such as the temperature dependence of , Among the major differences between Venus and Earth is the the amount of interior heating, and mineralogical transformations lack of current plate tectonics on Venus, as revealed by the geo- in the mantle, also influence the propensity for plate tectonics. logical analysis of radar images acquired by the Magellan mission Furthermore, episodic regimes in which stagnant-lid periods al- (59, 60). Plate tectonics plays an important role on Earth, providing ternate with active-lid periods may also exist (65). One hypothesis an efficient way of cooling the Earth’s interior and an exchange that has been proposed to explain Venus’s global resurfacing is mechanism between the interior, the surface, and the atmosphere: a transition from a plate-tectonics regime to a stagnant-lid regime at subduction zones, hydrated minerals and sediments are carried ∼1Gyrago(66). back to the mantle; at midocean ridges, new and Definitively determining whether a planet around another star contained in the mantle are released into the atmosphere (Fig. 3). undergoes plate tectonics will be extremely difficult. Still, as our Whether such a cycle (61) is required for life to form (62) and understanding of geology on Earth and elsewhere in the solar evolve on a is debated. Still, because the only system improves, we might be better able to estimate the likeli- world where life has been identified has plate tectonics, it is crucial hood of exoplanetary plate tectonics. to understand what controls tectonic dynamics. Plate tectonics occurs in response to convective motions in the Ocean Planets. Water is a key ingredient for the formation and ‡ mantle (Fig. 3). The main features are the formation of instabilities development of life. “Follow the water” has, therefore, been – at the thermal boundaries located at the core mantle transition and a motto for Mars exploration. Although H2O is known to exist in at the surface (Fig. 3). Hot plumes are common features of Venus ice and vapor form on Mars, subsurface liquid water has not yet “ ”— and Earth (63). Venus is in the so-called stagnant-lid regime been identified. However, geophysical observations (magnetic cold plumes form at a cold thermal-boundary layer, deep below the field and gravity field) by the Galileo mission and the Cassini surface, at the base of the stagnant lid through which heat is mission strongly suggest the presence of water under the icy crust transferred by conduction. The stagnant lid regime is not effi- of , , Europa, , and Enceladus, and this cient in removing heat from the interior (64). The tectonic re- has inspired several studies about the possibility of ocean- gime (plate tectonics or stagnant lid) depends on a number of dominated exoplanets. parameters, including the size of the planet, the surface temper- Ocean exoplanets are not known to exist, but some of their ature, the presence of water at the surface, and more generally the possible properties have been theoretically explored. For a given history of the planet (65). The reason why Venus and Earth have evolved so differently pathways remains uncertain (66).

Convective stresses scale with the vigor of convection. Some ‡ It is also possible that having too much water might hinder the development of life, authors argue that larger planets have larger convective stresses because it might both dilute important nutrients and disrupt geochemical thermal and, therefore, are more likely to experience plate tectonics regulation processes.

Spiegel et al. PNAS | September 2, 2014 | vol. 111 | no. 35 | 12625 Downloaded by guest on September 26, 2021 mass, a planet that is 50% (by mass) H2O and 50% Earth-like Conclusions composition would have a radius ∼25% larger than that of Observational campaigns in the next decade will help us refine a terrestrial exoplanet (57). As illustrated in Fig. 1, several our knowledge of the structure of planets both in our solar sys- exoplanets have ðMp; RpÞ values that lie in the vicinity of the tem and beyond. The Juno mission (73) will provide crucial new Rp½Mp relationship for Earth-like composition. Recent work has insights into Jupiter’s internal structure, and should help resolve examined the structure of ocean planets with a variety of water the long-standing question of how much water is in Jupiter. fractions and found, generically, that ocean planets have Rp½Mp Beyond the solar system, the recently approved Transiting Exo- relations that are somewhat larger at a given mass than Earth- planet Survey Satellite will identify many planets around stars like planets (26, 53, 69, 70), with the degree of increased radius close enough and bright enough that the planets are amenable to follow-up observations from the ground or with the James Webb depending on the water fraction. However, there are degeneracies: ðM ; R Þ Space Telescope. Learning about both the masses (via radial for instance, a planet whose p p location is consistent with velocity measurements) and about the atmospheres of these being an ocean planet could also have a silicate (terrestrial) core planets will inform our understanding of their bulk compositions veiled by an H/He-rich atmosphere (58). This degeneracy can be and structures. resolved if the atmospheric composition can be discerned (71). Finally, in some cases, giant exoplanets that are sufficiently far An ocean planet’s habitability could be affected by whether its from their stars and sufficiently self-luminous, such as those in liquid water is in contact with a rocky core. Planets with large the HR 8799 system (27), may be directly imaged with new, high- amounts of H2O might develop a high-pressure ice layer between contrast imaging techniques. These direct observations of young the core and the liquid layer (26), similar to the structure pro- objects are sensitive to planets’ initial conditions and, therefore, posed for Ganymede, Callisto, and Titan. Counterintuitively, might help us (i) to distinguish between formation mechanisms more massive planets might have thinner oceans, because they of widely separated, young Jovian objects (74–76), and (ii)to have greater pressure gradients and, therefore, their oceans learn about their interior structures. The ground-based programs more quickly enter the high-pressure ice-layer regime. To have that will undertake such surveys include the Gemini Planet Im- contact between the ocean and the rocky core, the fraction of ager (77), the Spectro-Polarimetric High-Contrast Exoplanet −4 Research Instrument on the (78), and H2O has to be small, as is the case on Earth ð2 × 10 Þ and ð × −2Þ more. Exoplanetary observations have revealed unanticipated Europa 7 10 . However, in some circumstances, even if the structures in planets both large and small, and so the prospect of planet has a large water fraction, there can still be contact be- a flood of upcoming data promises more surprises and new in- tween a fluid water layer and a silicate layer; this would occur if sight into comparative planetology. the temperature at the top of the H2O layer is high enough that the envelope’s temperature gradient yields a temperature at the ACKNOWLEDGMENTS. Support for this work was provided by National base of the water layer that is that is still in the fluid regime, and Science Foundation (NSF) Grant AST-0807444, a Keck Fellowship, Friends of the Institute, and Association of Members of the Institute for Advanced might be a natural outcome in Neptune-class planets with or Study (to D.S.S.); NSF Grant AST-1010017 (to J.F.); and NASA without the loss of their primordial H2–He atmospheres. Institute Icy Worlds (C.S.).

1. Borucki WJ, et al. (2011) Characteristics of planetary candidates observed by Kepler. II. 22. Bodenheimer P, Lin DNC, Mardling RA (2001) On the tidal inflation of short-period Analysis of the first four months of data. Astrophys J 736:19. extrasolar planets. Astrophys J 548:466–472. 2. Batalha NM, et al. (2013) Planetary candidates observed by Kepler. III. Analysis of the 23. Miller N, Fortney JJ, Jackson B (2009) Inflating and deflating hot Jupiters: Coupled first 16 months of data. arXiv:1202.5852. tidal and thermal evolution of known transiting planets. Astrophys J 702:1413–1427. 3. Morton TD, Johnson JA (2011) On the low false positive probabilities of Kepler planet 24. Batygin K, Stevenson DJ (2010) Inflating hot Jupiters with ohmic dissipation. As- candidates. Astrophys J 738:170–182. trophys J 714:L238–L243. 4. Fressin F, et al. (2013) The false positive rate of Kepler and the occurrence of planets. 25. Wu Y, Lithwick Y (2013) Ohmic heating suspends, not reverses, the cooling contrac- Astrophys J 766(2):81–100. tion of hot Jupiters. Astrophys J 763:13. 5. Bakos G, et al. (2004) Wide-field millimagnitude photometry with the HAT: A tool for 26. Grasset O, Schneider J, Sotin C (2009) A study of the accuracy of M-R relationships for – extrasolar planet detection. Publ Astron Soc Pac 116:266 277. Si-rich and H2O-rich planets up to 100 earth masses. Astrophys J 693:722–733. 6. Pollacco DL, et al. (2006) The WASP project and the SuperWASP cameras. PASP 118: 27. Marois C, et al. (2008) Direct imaging of multiple planets orbiting the star HR 8799. – 1407 1418. Science 322(5906):1348–1352. – 7. Wright JT, et al. (2011) The Exoplanet Orbit Database. PASP 123:412 422. 28. Arras P, Socrates A (2010) Thermal tides in fluid extrasolar planets. Astrophys J 714: 8. Schneider J, Dedieu C, Le Sidaner P, Savalle R, Zolotukhin I (2011) Defining and 1–12. – cataloging exoplanets: The exoplanet.eu database. Astron Astrophys 532 A79. 29. Budaj J, Hubeny I, Burrows A (2012) Day and night side core cooling of a strongly 9. Zapolsky HS, Salpeter EE (1969) The mass-radius relation for cold spheres of low mass. irradiated . Astron Astrophys 537:A115. Astrophys J 158:809. 30. Spiegel DS, Burrows A (2013) Thermal processes governing hot-Jupiter radii. As- 10. Burrows A, Hubbard WB, Lunine JI, Liebert J (2001) The theory of brown dwarfs and trophys J 772:76. extrasolar giant planets. Rev Mod Phys 73:719–765. 31. Baraffe I, Chabrier G, Barman TS, Allard F, Hauschildt PH (2003) Evolutionary models 11. Guillot T, Burrows A, Hubbard WB, Lunine JI, Saumon D (1996) Giant planets at small for cool brown dwarfs and extrasolar giant planets. The case of HD 209458. Astron orbital distances. Astrophys J 459:L35––L38. Astrophys 402:701–712. 12. Saumon D, et al. (1996) A theory of extrasolar giant planets. Ap J 460:993–1018. 32. Fortney JJ, et al. (2011) Discovery and atmospheric characterization of giant planet 13. Charbonneau D, Brown TM, Latham DW, Mayor M (2000) Detection of planetary Kepler-12b: An inflated radius outlier. Astrophys J Suppl 197:9. transits across a sun-like star. Astrophys J 529(1):L45–L48. 33. Weiss LM, et al. (2013) The mass of KOI-94d and a relation for planet radius, mass, and 14. Burrows A, et al. (2000) On the radii of close-in giant planets. Astrophys J 534(1): incident flux. Astrophys J 768:14. L97–L100. 34. Mizuno H, Nakazawa K, Hayashi C (1978) Instability of a gaseous envelope surrounding 15. Baraffe I, Chabrier G, Barman T (2010) The physical properties of extra-solar planets. – Rep Prog Phys 73:016901. a planetary core and formation of giant planets. Prog Theor Phys 60:699 710. 16. Fortney JJ, Nettelmann N (2010) The interior structure, composition, and evolution of 35. Boss AP (2010) Giant planet formation by disk instability in low mass disks? Astrophys – giant planets. Space Sci Rev 152:423–447. J 725:L145 L149. 17. Saumon D, Chabrier G, van Horn HM (1995) An equation of state for low-mass stars 36. Guillot T, et al. (2006) A correlation between the heavy element content of transiting and giant planets. Astrophys J Suppl 99:713–741. extrasolar planets and the metallicity of their parent stars. Astron Astrophys 453: 18. Militzer B, Hubbard WB (2013) Ab initio equation of state for hydrogen-helium L21–L24. mixtures with recalibration of the giant-planet M-R relation. Astrophys J 774(2):148. 37. Sato B, et al. (2005) The . II. A transiting hot Saturn around HD 19. Guillot T, Showman AP (2002) Evolution of “51 b-like” planets. Astron As- 149026 with a large dense core. Astrophys J 633:465–473. trophys 385:156–165. 38. Fortney JJ, Saumon D, Marley MS, Lodders K, Freedman RS (2006) Atmosphere, in- 20. Burrows A, Hubeny I, Budaj J, Hubbard WB (2007) Possible solutions to the radius terior, and evolution of the metal-rich transiting planet HD 149026b. Astrophys J 642: anomalies of transiting giant planets. Astrophys J 661:502–514. 495–504. 21. Perna R, Menou K, Rauscher E (2010) Ohmic dissipation in the atmospheres of hot 39. Fortney JJ, Ikoma M, Nettelmann N, Guillot T, Marley MS (2011) Self-consistent model Jupiters. Astrophys J 724:313–317. atmospheres and the cooling of the solar system’s giant planets. Astrophys J 729:32.

12626 | www.pnas.org/cgi/doi/10.1073/pnas.1304206111 Spiegel et al. Downloaded by guest on September 26, 2021 40. Podolak M, Helled R (2012) What do we really know about Uranus and Neptune? 61. Walker JCG, Hays PB, Kasting JF (1981) A negative feedback mechanism for the long-

Astrophys J 759:L32. term stabilization of Earth’s surface temperature. J Geophys Res 86:9776–9782. SPECIAL FEATURE 41. Nettelmann N, Helled R, Fortney JJ, Redmer R (2013) New indication for a dichotomy 62. Spiegel DS, Turner EL (2012) Bayesian analysis of the astrobiological implications of in the interior structure of Uranus and Neptune from the application of modified life’s early emergence on Earth. Proc Natl Acad Sci USA 109(2):395–400. shape and rotation data. Planet Space Sci 77:143–151. 63. Smrekar SE, Sotin C (2012) Constraints on mantle plumes on Venus: Implications for ’ 42. Fletcher LN, Drossart P, Burgdorf M, Orton GS, Encrenaz T (2010) Neptune s atmo- volatile history. Icarus 217:510–523. spheric composition from AKARI spectroscopy. Astron Astrophys 514:A17. 64. Reese CC, Solomatov VS, Moresi L-N (1998) Heat transport efficiency for stagnant lid ’ 43. Luszcz-Cook SH, de Pater I (2013) Constraining the origins of Neptune s carbon mon- convection with dislocation viscosity: Application to Mars and Venus. J Geophys Res oxide abundance with CARMA millimeter-wave observations. Icarus 222:379–400. 103:13643–13658. 44. Butler RP, et al. (2004) A Neptune-mass planet orbiting the nearby M dwarf GJ 436. 65. Lenardic A, Crowley JW (2012) On the notion of well-defined tectonic regimes for Astrophys J 617:580–588. terrestrial planets in this solar system and others. Astrophys J 755:132. 45. Gillon M, et al. (2007) Accurate Spitzer infrared radius measurement for the hot 66. Stevenson D (2003) Les styles de convection du manteau et leur influence sur l’évo- Neptune GJ 436b. Astron Astrophys 471:L51–L54. lution des planètes [Styles of mantle convection and their influence on planetary 46. Adams ER, Seager S, Elkins-Tanton L (2008) Ocean planet or thick atmosphere: On the – mass-radius relationship for solid exoplanets with massive atmospheres. Astrophys J evolution]. C R Geosci 335:99 111, French. ’ 673:1160–1164. 67. Valencia D, Sasselov DD, O Connell RJ (2007) Detailed models of super-Earths: How – 47. Nettelmann N, Kramm U, Redmer R, Neuhäuser R (2010) Interior structure models of well can we infer bulk properties? Astrophys J 665:1413 1420. GJ 436b. Astron Astrophys 523:A26–A35. 68. Valencia D, Ikoma M, Guillot T, Nettelmann N (2010) Composition and fate of short- 48. Sanchis-Ojeda R, et al. (2012) Alignment of the stellar spin with the orbits of a three- period super-Earths. The case of CoRoT-7b. Astron Astrophys 516:A20. planet system. Nature 487(7408):449–453. 69. Rogers LA, Seager S (2010) A framework for quantifying the degeneracies of exo- 49. Dong S, Zhu Z (2013) Fast rise of Neptune-size planets from 10 to 250 days—statistics planet interior compositions. Astrophys J 712:974–991. of Kepler planet candidates up to 0.75 AU. Astrophys J 778:53. 70. Zeng L, Sasselov D (2013) A detailed model grid for solid planets from 0.1 through 100 50. Lopez ED, Fortney JJ, Miller N (2012) How thermal evolution and mass-loss sculpt Earth masses. Publ Astron Soc Pac 125:227–239. populations of super-earths and sub-Neptunes: Application to the Kepler-11 system 71. Miller-Ricci E, Seager S, Sasselov D (2009) The atmospheric signatures of super-Earths: and beyond. Astrophys J 761:59. How to distinguish between hydrogen-rich and hydrogen-poor atmospheres. As- 51. Charbonneau D, et al. (2009) A super-Earth transiting a nearby low-mass star. Nature trophys J 690:1056–1067. 462(7275):891–894. 72. Sotin C, Jackson JM, Seager S (2011). Terrestrial planet interiors. Exoplanets, ed Se- 52. Rogers LA, Seager S (2010) Three possible origins for the gas layer on GJ 1214b. As- ager S (Univ of Arizona Press, Tucson, AZ), pp 375–395. – trophys J 716:1208 1216. 73. Bolton SJ (2010) The Juno mission. Proceedings of the International Astronomical 53. Kipping DM, Spiegel DS, Sasselov DD (2013) A simple, quantitative method to infer Union IAU Symposium, ed Barbieri C, et al. (Cambridge Univ Press, Cambridge, UK), the minimum atmospheric height of small exoplanets. Mon Not R Astron Soc 434: Vol 269, pp 92–100. 1883–1888. 74. Marley MS, Fortney JJ, Hubickyj O, Bodenheimer P, Lissauer JJ (2007) On the lumi- 54. Bean JL, Kempton EM, Homeier D (2010) A ground-based transmission spectrum of nosity of young Jupiters. Astrophys J 655:541–549. the super-Earth exoplanet GJ 1214b. Nature 468(7324):669–672.

75. Fortney JJ, Marley MS, Saumon D, Lodders K (2008) Synthetic spectra and colors of ASTRONOMY 55. Berta ZK, et al. (2012) The flat transmission spectrum of the super-Earth GJ1214b from young giant planet atmospheres: Effects of initial conditions and atmospheric met- wide field camera 3 on the . Astrophys J 747:35. – 56. Sotin C, Grasset O, Mocquet A (2007) Mass radius curve for extrasolar Earth-like allicity. Astrophys J 683:1104 1116. planets and ocean planets. Icarus 191:337–351. 76. Spiegel DS, Burrows A (2012) Spectral and photometric diagnostics of giant planet 57. Léger A, et al. (2004) A new family of planets? Ocean planets. Icarus 169:499–504. formation scenarios. Astrophys J 745:174. 58. Seager S, Deming D (2010) Exoplanet atmospheres. ARAA 48:631–672. 77. Macintosh B, et al. (2006) The . Society of Photo-Optical In- 59. Schaber GG, et al. (1992) Geology and distribution of impact craters on Venus: What strumentation Engineers (SPIE) Conference Series. Proc SPIE 6272, 10.1117/12.672430. are they telling us? J Geophys Res 97:13257–13301. 78. Beuzit J-L, et al. (2008) SPHERE: A ‘planet finder’ instrument for the VLT. Society of Photo- 60. Basilevsky AT, Head JW (2002) Venus: Timing and rates of geologic activity. Geology Optical Instrumentation Engineers (SPIE) Conference Series. Proc SPIE 7014, 10.1117/ 30:1015. 12.790120.

Spiegel et al. PNAS | September 2, 2014 | vol. 111 | no. 35 | 12627 Downloaded by guest on September 26, 2021