Radial velocity observations remain the most reliable and consistent method of detecting extrasolar planets (see Detection Methods). Even under the most favorable circumstances, however, they yield very limited data: orbital period and eccentricity, minimum mass, approximate semimajor axis. Moving from this stark list of numerical values to a three-dimensional view of a planet requires considerable extrapolation, much of which is informed by theoretical speculations that remain in flux.

Early on, it was recognized that extrasolar planets vary widely in terms of the shapes and sizes of their orbits. In this regard they bear little resemblance to those in the Solar System. Seven out of eight planets circling our Sun, with semimajor axes ranging from 0.7 to 30 AU, have orbital eccentricities smaller than 0.1. Even the most elliptical orbit, that of Mercury, has an eccentricity of only 0.206. Although Mercury's semimajor axis is 0.39 AU, it approaches the Sun as close as 0.31 AU and retreats as far as 0.47 AU over the course of a single orbit.

Extrasolar planets tend to follow orbits that are either much tighter or much more elliptical. More than 40% of all detected exoplanets have semimajor axes smaller than Mercury's, while a similar percentage have eccentricities more extreme. Eccentricity varies by distance from the host star. Among planets with semimajor axes smaller than 0.05 AU, almost 90% have orbital eccentricities (e) smaller than 0.1, and none have e > 0.3. The likelihood of elliptical orbits increases along with semimajor axis, such that only 19% of planets orbiting beyond 1 AU have e < 0.1, while 45% have e > 0.3. A small group – amounting to 4% of the total sample – has extreme eccentricities of 0.7 or higher, like the orbits of comets rather than those of planets as we usually imagine them.

By themselves, orbital eccentricity and semimajor axis can tell us very little about an exoplanet. They must be interpreted in the context of the planet's mass as well as of the mass, luminosity, spectral type, and history of its parent star. From the various permutations of these values has emerged an ad hoc taxonomy of exoplanets, including such species as Hot Jupiters, Hot Neptunes, Super Earths, Super Jupiters (or Megajovians), and Jupiter analogs. All but the last were unknown and largely unsuspected until the modern era of exoplanetary studies began in the 1990s.

hot jupiters

More than 30% of the current population of exoplanets have semimajor axes smaller than 0.1 AU (Extrasolar Planets Encyclopaedia, 9/08). Most of this population consists of gas giants. The orbital periods of this group of massive, close-in planets are usually shorter than one week. The tightness of their orbits makes them hotter than Venus, the hottest planet in the Solar System. As a result they are typically known in English as Hot Jupiters. In French they may be called pegasides or Pegasids, referring to 51 Pegasi, the first such object to be detected, and also recalling a well-known meteor swarm.

Astronomical consensus favors a cooler and more distant location for their birthplace, probably in the vicinity of their home system’s ice line (see also Evolution and System Architectures). Especially around highly metallic stars, planetary accretion in this region can proceed very quickly, encouraging newborn gas giants to spiral inward through the protoplanetary nebula while capturing massive atmospheres of hydrogen. Their migration necessarily halts when they reach the inner edge of the nebula, which is truncated by the magnetosphere of the central star at an approximate radius of 0.05 AU. Once migration ends, stellar tides modify the shape of their final orbits, so that most “star-grazing” or epistellar planets have orbital eccentricities approaching zero.

Our understanding of Hot Jupiters remains sketchy. At minimum we can be sure that, despite their nickname, they will not resemble the Solar System’s giant planets. Their epistellar orbits will prevent the formation of moons and stall their rotation through tidal drag. Thus they always turn the same hemisphere to their host stars, as the Moon does around the Earth. Hot Jupiters will lack the colorful cloud bands that we see on our own Jupiter and Saturn, since such features are a direct consequence of rapid rotation (Jupiter and Saturn each rotate in about 10 hours). Nor are they likely to preserve rings in the face of stellar tides.

Intense stellar irradiation causes these gaseous bodies to expand in diameter and become increasingly tenuous, as confirmed by photometric transit observations (Charbonneau et al. 2007). I. Baraffe has argued that the ultimate effect of this irradiation is “strong atmospheric evaporation and mass loss,” such that a planet originally as massive as Jupiter will shed 90% of its bulk within 2 to 5 billion years (Baraffe et al. 2005, 2006). However, observations do not support this conclusion (Melo et al. 2006). Rather, Hot Jupiters seem to retain their large masses and distended hydrogen envelopes over billions of years (Hubbard et al. 2007).

David Sudarsky and colleagues have generated theoretical spectra for gas giant planets at a range of temperatures, which correspond to various combinations of semimajor axis and mass (Sudarsky et al. 2003). For orbits within a few tenths of an AU (corresponding to Hot and “Warm” Jupiters) they suggest more than one possibility. The very closest planets (in their terminology, “roasters”) would be covered by dark clouds of silicate and iron compounds. At somewhat cooler temperatures (but still in the vicinity of 500-1000 K) no clouds at all would be able to form, and planets in this range would present clear spheres of indigo or deep blue (“Gaseous Class III and Class II”).

transiting hot jupiters

More concrete data on hot giant planets have resulted from studies of the select group that are observed in transits of their host stars. By September 2008, more than 50 transiting exoplanets had been characterized through a combination of radial velocity and photometric studies (Torres et al. 2008, EPE). Transit observations establish that Hot Jupiters (and by inference other exoplanets in the gas giant range) vary widely in terms of composition and structure. On the basis of mass, radius, and inferred density, it seems clear that some are composed almost entirely of hydrogen, like stars, while others have large percentages of heavy elements, more like terrestrial planets.

An example of the former is TrES-4, which orbits an F8 star located about 440 parsecs away. The planet's semimajor axis of 0.05 AU corresponds to a period of less than 4 days. The mass of TrES-4 is unexceptional, at about 0.84 MJUP, but its radius is a remarkable 1.751 RJUP. This value indicates the largest volume and lowest density of any exoplanet so far detected (Mandushev et al. 2007, Torres et al. 2008). Its discovery team speculated that the planet may be enveloped in an extended comet-like tail.

At the other end of the scale is HD 149026 b, which circles a highly metallic G-type subgiant located at a distance of about 79 parsecs. Its semimajor axis of 0.04 AU corresponds to an even shorter period of 2.9 days. At 0.359 MJUP, this planet is slightly more massive than Saturn, making it the most lightweight Hot Jupiter known (the lightest of all transiting exoplanets is GJ 436 b, a Hot Neptune with a very different internal structure). Yet the radius of HD 149026 b is anomalously small, measured at 0.654 RJUP (Torres et al. 2008). This value implies that the planet has a heavy-element core about 70 times as massive as Earth – much larger than the core masses usually predicted for gas giant planets (~10 MEA), and thus posing problems for the core-accretion theory of planet formation (see Broeg & Wuchterl 2007).

HD 209458 b, a more typical example of the species, is relatively well known through spectral analysis (Grillmair et al. 2007, Richardson et al. 2007, Swain et al. 2007). Its mass is 0.685 MJUP; its radius is 1.359 RJUP (Torres et al. 2008). It orbits a hot G0 star located 47 parsecs away, at a semimajor axis of about 0.05 AU in less than four days. The planet's spectrum indicates that it reflects very little light, and its extended atmosphere appears to contain silicate compounds, as predicted by Sudarsky and colleagues. Evidently HD 209458 b and similar planets are “cloaked in black silicate clouds,” so that they display dim, dark spheres unlike any planets in the Solar System (Sanderson 2007).

David Langton and Greg Laughlin have modeled the atmospheric dynamics of individual transiting Hot Jupiters by generating animated simulations (Langton & Laughlin 2007; see also Laughlin’s blog entry on Vorticity). They focus on the minority of transiting planets that follow eccentric orbits and thus experience substantial temperature variation. The models of Langton and Laughlin reveal extremely turbulent atmospheres, with enormous vortices rolling around the planets’ equatorial zones and in some cases devolving into regions of swirling chaos.

hot neptunes

So far no hard and fast rules are available to distinguish extrasolar ice giants from extrasolar gas giants. There seems to be general agreement that Uranus, Neptune, and GJ 436 b are all ice giants, spanning the range of masses from 14.5 to 22.6 MEA (0.046 to 0.071 MJUP). The next point of consensus is Saturn’s status as a gas giant, at 95.2 MEA (0.299 MJUP). Some theorists predicted a “planet desert” in the void between these two extremes (Ida & Lin 2004, 2005), but the desert has turned out to be fertile. As of February 2008, the Catalog of Nearby Exoplanets lists more than a dozen objects distributed in the range between 0.071 and 0.299 MJUP. Are these ice giants, gas giants, a transitional species, or some combination of the three?

One way out of the impasse would be to define the lower limit for gas giants as the minimum mass needed to open a gap in a primordial gas disk and initiate Type II migration to smaller orbits. While this parameter varies from system to system, a convenient rule of thumb is that planets of 0.2 MJUP or more are capable of opening a gap and thus qualify as gas giants (Armitage 2007, Raymond et al. 2007b). Planets of lesser mass (but still greater than 0.03 MJUP/10 MEA) are ice giants.

This approach brings us a trove of extrasolar ice giants – currently 18 objects between 10 and 65 MEA. All of them orbit within 0.78 AU of their primaries, and all but three orbit within 0.5 AU. Such a distribution of short-period orbits is no surprise. Because ice giants tug less strongly on their host stars, they are more difficult to detect by radial velocity observations than their more massive gaseous cousins. Thus they tend to be observed quite close to their host stars, where the effects of their orbital motion are most evident.

Four of these Hot (or Warm) Neptunes orbit within 0.1 AU of an M dwarf star (GJ 674 b, GJ 581 b, GJ 436 b, GJ 176). Nine are members of two- or multiple-planet systems around Sun-like stars (HD 69830, 55 Cancri, Mu Arae, GJ 777, HD 11964, HD 177830). The remaining five are, like most exoplanets, the sole companions of Sun-like stars.

The existence of these close-in Neptune-mass planets seems especially paradoxical, given the example of our Solar System and recent theories of evolutionary migration (Ida & Lin 2004, 2005). Planets of Neptune's mass are thought to assemble farther from their host stars than gas giant planets, in regions of the protoplanetary disk where planetesimal accretion occurs so slowly that Jupiter-mass planets have no time to form before the dispersal of primordial gas closes the window for gas giant accretion. Therefore, the presence of Neptune-mass planets in orbits interior to gas giants (as in the systems of Mu Arae, 55 Cancri, GJ 777, and HD 11964) poses problems for theories of planetary migration (see Evolution of Planetary Systems). In fact, Baraffe et al. (2005) have suggested that Hot Neptunes are actually Hot Jupiters that have suffered extensive mass loss as a result of intense stellar irradiation. In other words, Hot Neptunes were originally much more massive, but they have evaporated through long exposure to the heat of their primary stars.

Continuing observations and theoretical simulations, however, have failed to support this intriguing proposition. The evolutionary path for Hot Neptunes now tends to be discussed in terms of Type I migration, a mechanism by which planets less massive than Saturn undergo orbital decay in a gas-rich protoplanetary disk, eventually arriving at star-grazing orbits.

super earths

Yet another of the exotic bodies known or anticipated only in exoplanetary systems is the class of Super Earths, with masses between 2 and 10 times that of our own (non-Super) Earth. Radial velocity search programs have so far proposed eight candidates in this mass range, and microlensing searches have nominated two more. (Given the sparse data available through microlensing, the present discussion is limited to radial velocity detections.)

To date, the smallest Super Earth candidate is HD 40307 b, whose minimum mass is estimated at 4.2 MEA. With current techniques, this is the most lightweight object that can be identified in orbit around a Sun-like star. However, the planet’s true mass is unknown, because it cannot be calculated without data on the inclination of the planet’s orbit to the line of sight. This angle is undetectable by radial velocity searches. The same uncertainty applies to five other candidates – Gliese 581 c and d (5.03 and 7.7 MEA); HD 40307 c and d (6.8 and 9.2 MEA); and HD 181433 b (7.56 MEA).

Fortunately, we have a better understanding of two other potential Super Earths: Gliese 876 d, whose minimum mass of 5.89 MEA and approximate inclination of 50 degrees implies a true mass of 7.5 MEA (Rivera et al. 2005); and 55 Cancri e, whose minimum mass of 7.66 MEA and approximate inclination of 53 degrees implies a true mass of about 9.6 MEA (Fischer et al. 2008, Table 4).

Whether any given Super Earth represents a larger version of the Solar System’s terrestrial planets – Mercury, Mars, Venus, and Earth – or a smaller version of our ice giants – Uranus and Neptune – remains a mystery. Several recent studies have explored each alternative, presenting model planets composed of varying proportions of iron, silicate rocks, and ices (Valencia et al. 2006, Fortney et al. 2007, Baraffe et al. 2008).

The likeliest rocky Super Earths are Gliese 876 d and 55 Cancri e, for two reasons. First, each planet travels on an extremely tight orbit, meaning that it has been exposed to extreme irradiation for billions of years. A volatile-rich planet in such an environment presumably would have evaporated by now (Lecavelier des Etangs 2006). Second, each planet orbits immediately starward of a gas giant planet that evidently reached its present position by inward migration. Such a history implies that the inner planet was formed by the accretion of rocky planetesimals shepherded into ever smaller orbital radii by the migrating gas giant (Fogg & Nelson 2005, Mandell et al. 2007).

The candidate Super Earths around Gliese 581 and HD 40307 are more likely to be ice planets (AKA “Ocean Planets”), a species originally hypothesized by Leger and colleagues and then elaborated in subsequent studies (Leger et al. 2004, Sotin et al. 2007, Selsis et al. 2007). This conclusion is supported by two factors: the low metallicity of their host stars, indicating a shortage of metallic and rocky components in their primordial environments; and the absence of a shepherding mechanism to focus refractory materials in small orbits.

Such icy Super Earths probably form outside their system’s ice lines and are then transported to smaller orbits by planet-planet scattering or Type I migration. Given their native environments, they are likely to be 50% rock/metal and 50% ice (Leger et al. 2004, Selsis et al. 2007). Their subsequent evolution would depend on their final semimajor axes and on the luminosity of their host stars. In the habitable zone, the outer layers of an icy planet would melt, resulting in an “Ocean Planet” with a rocky core surrounded by a thick layer of ice supporting a global ocean 100 kilometers deep. Starward of this region, water cannot be sustained in liquid form, so the result would be a “Steam Planet” (AKA “Sauna World”) in which a thick, hot atmosphere of water vapor is in direct contact with a high-pressure mantle of ice.

megajovians

The median mass for all extrasolar gas giants is about 1.75 Mjup (assuming 0.2 Mjup as the lower mass limit). At least 15% of these giant planets are extreme heavyweights, with minimum masses of 5 Mjup or more. Beginning with the first batch of detections in the 1990s, astronomers have expressed concern that some of these heavyweights – often called “Super Jupiters” or “Megajovians” – might actually be brown dwarfs or failed stars (Marcy & Butler 1996, Baraffe et al. 2008). Recently, however, the trend has been to accept that gas giants can form at a wide range of masses, even as high as 25 Mjup. This is almost double the limit of 13 Mjup that is often proposed as the boundary between gas giants and brown dwarfs (e.g., Butler et al. 2006, Udry & Santos 2007). See Super Jupiter or Brown Dwarf?.

Regardless of which upper limit we accept, the current sample of Megajovians stands apart from the general exoplanet population in terms of native environment and orbital configuration. The present discussion refers to a sample of 41 high-mass planets detected within 200 parsecs of the Solar System; all are at least 5 times Jupiter's mass. The complete list is available here.

Megajovians have been confirmed only around relatively heavy stars of 0.78 Msol or more, in keeping with the hypothesis that exoplanet mass scales with host star mass. Except for the highly ambiguous case of HD 41004 Bb, none have been reported in orbit around M dwarfs, and only two are known to orbit K dwarfs (14 Herculis and HD 162020).

All other Megajovian hosts are either main-sequence stars of spectral classes G or F, or evolved stars that formerly belonged to classes G, F, or A. This finding conforms with the theory of planet formation by accretion, insofar as more massive stars typically sustain more massive protoplanetary disks, and more massive disks are likely to yield more massive planets.

Nevertheless, the host stars of Megajovians do not exhibit any particular enhancement in metal content. Their metallicity values run the gamut from -0.68 to +0.46, with a median of +0.08. This is virtually identical to the median metallicity of +0.12 for all exoplanet host stars.

Apart from mass, the most striking feature of this population is its preponderance of wide, eccentric orbits. The median semimajor axis for Megajovians is 1.91 AU, compared to a median of 0.63 AU for all exoplanets. Only 10% have semimajor axes smaller than 0.1 AU, compared to 33% for all exoplanets, while 71% orbit beyond 1 AU (the semimajor axis of Earth), compared to 41% for all exoplanets. Evidently the mechanisms that deliver gas giants to small orbits – e.g., Type II migration and planet-planet scattering – are less effective for objects of 5 Mjup or more.

Wide orbits are generally more eccentric than tight orbits, but even at the same range of semimajor axes, the orbits of Megajovians are notably more elliptical than those of less massive exoplanets. For semimajor axes of 0.1 AU or more, the median eccentricity among Megajovians is 0.36, while the median for less massive planets is 0.24. Two-thirds of Megajovians in this orbital space have substantial eccentricities of 0.3 or more, while 11% have extreme values of 0.7 or more. Only 40% of less massive planets orbiting at 0.1 AU and beyond have eccentricities of 0.3 or more, and only 3% have eccentricities of 0.7 or more.

A key unanswered question is whether these distinctive characteristics – large planet mass, large host star mass, large semimajor axis and eccentricity, absence of enhanced metallicity – imply a special evolutionary path. Fortunately, three Megajovians have so far been detected by photometric transit surveys (HD 147506 b, XO-3 b, and WASP-14 b), so the outlook for eventual resolution of this problem is good.

A transit detection enables the determination of a planet’s radius and true mass, facilitating estimates of its internal composition and providing clues to its formation. Among the trio of transiting Megajovians, WASP-14 b (7.73 Mjup) seems to fit prevailing theoretical models for a massive, extremely dense gas giant containing a large heavy element core (Joshi et al. 2008). The status of XO-3 b (11.8 Mjup) is less certain, since its radius appears to be slightly inflated, and a recent study could not decide whether it is metal-rich or metal-poor (Winn et al. 2008). HD 147506 b (also known as HAT-P-2b) is ambiguous for the opposite reason, because its mass of 8.78 Mjup and unusually small radius of 0.95 Rjup imply a density more appropriate for a brown dwarf or an M dwarf than for a gas giant (Loeillet et al. 2008).

These ambiguities have prompted speculation that the Megajovians represent a unique class unto themselves. As Loeillet and colleagues suggest, “Supermassive planets like HAT-P-2b may constitute a new class of stellar companions, in between Hot-Jupiters and low-mass stars and near the brown dwarf population. . . . . This could suggest a different formation process” (Loeillet et al. 2008). In a study of transiting planets, Torres and colleagues agreed on the oddball status of HD 147506 b/HAT-P-2b, observing that this object “may belong to a different category of planet” (Torres et al. 2008).

Yet in many regards the Megajovians behave much like less massive gas giants. They have been found in “hot” orbits smaller than 0.1 AU and “cool” orbits larger than 3 AU; they occur in systems containing one or more additional planets (HD 68988, HD 74156, HD 38529, HD 202206, HD 168443), on interior as well as exterior orbits; and they have been detected in systems that harbor massive debris belts (70 Virginis, HD 38529). Since the core accretion theory has been successful in explaining all of these system architectures, it may also prove adequate to explain the formation of Megajovians.

cool giants and jupiter analogs

Once a rarity, cooler gas giants with semimajor axes of 0.5 AU or more are now the most commonly detected of all extrasolar planets. Most of them travel on orbits far more eccentric than those of the Solar planets, and they also tend to be more massive than the epistellar giants.

Only four exoplanets beyond 0.5 AU have minimum masses less than 0.5 MJUP. These are HD 69830 c, a Warm Neptune; 55 Cancri f, a transitional object whose mass of 0.14 MJUP (46 MEA) falls between the ice giants and the gas giants, and whose semimajor axis is similar to that of Venus; HD 74156 c, which at 0.4 MJUP is about one-third again as massive as Saturn, and which travels in a moderately eccentric orbit at about 1 AU; and HD 164922 b, an apparent Saturn analog whose circular orbit has a semimajor axis of 2.11 AU. On the other hand, a few dozen planets orbiting outside 0.5 AU have masses between 5 and 13 MJUP, the cutoff for brown dwarfs. It is uncertain whether this deficit of lower-mass planets is real or simply an artifact of the limitations of the radial velocity method.

Of the 54 exoplanets so far detected within 20 parsecs, 18 (33%) qualify as “cool eccentric giants,” with semimajor axes greater than 0.5 AU and eccentricities greater than 0.1. Nine of the 54 (17%) follow relatively circular orbits (e < 0.1) with comparable semimajor axes. The remaining 27 planets (50%) orbit inside 0.5 AU.

Within the larger sample of exoplanets, only three currently meet Sean Raymond’s criteria for “Jupiter analogs.” These are gas giants with [1] semimajor axes of 3.5 AU or more, [2] eccentricities near 0, and [3] no interior giant companions with semimajor axes exceeding 0.5 AU (Raymond 2006).

This select group comprises:

• HD 154345 b, the only planet of a nearby G8 star, with a minimum mass identical to Jupiter’s, a semimajor axis of 4.18 AU, and an eccentricity of 0.04.
• HD 24040 b, another single planet orbiting a G0 star, with a minimum mass about four times Jupiter’s, a semimajor axis of 4.68 AU, and an eccentricity of 0.07.
• HD 68988 c, the outer planet of a two-planet system around another G0 star, with a minimum mass about five times Jupiter’s, a semimajor axis very similar to Jupiter's (5.32 AU vs. 5.2 AU), and an eccentricity approaching 0.

Unfortunately, none of these systems are well constrained, and published physical and orbital parameters have been subject to large revisions on an annual basis.

The theories of Sudarsky and colleagues provide an interesting preview of the appearance of this varied group of long-period giants (Sudarsky et al. 2003). They suggest two alternatives: colorful methane-ammonia clouds like those of Jupiter and Saturn for the coldest planets (probably beyond 2 AU for a G-type star), and water-ice clouds like those of Earth for planets in smaller orbits (but still cooler than 300 K). This latter type, which they term “Water Class II,” is especially interesting, because such planets may look deceptively like giant versions of our own homeworld. As Sudarsky and colleagues point out, even at large orbital separations the most massive planets will maintain high core temperatures and thus present the appearance of water-cloud giants or – if young and massive enough – clear blue spheres like the “Gaseous Class III” model (Sudarsky et al. 2003).

Last update September 2008