Wednesday, December 31, 2014

Hot Jupiter Twins: WASP-94

Figure 1. Imaginary view of a binary system containing two Sun-like stars, each with a Hot Jupiter, one transiting.
A team of European astronomers recently announced an odd and interesting exoplanetary first: the WASP-94 system, which consists of a pair of F-type stars sharing a very wide binary orbit, each accompanied by a gas giant planet on a very tight orbit. Giant planets on such orbits (whether in binary systems or not) are commonly nicknamed Hot Jupiters. Until very recently, no binaries at all were known in which both stars supported a planetary system, and until this discovery, no double planetary systems were known in which both systems contained a Hot Jupiter.

One of the Hot Jupiters in this binary is also observed in transit across its host star (WASP-94A). This fortuitous arrangement permits measurements of the Rossiter-McLaughlin effect, a feature of stellar spectral lines that appears when transiting planets are misaligned with the spin axis of their parent stars. These measurements demonstrate that WASP-94Ab is so skewed in relation to its star’s rotation that its orbit is retrograde – i.e., it travels in the opposite direction to the stellar spin. Although the other Hot Jupiter is not observed in transit, even this null result offers meaningful evidence: the orbits of the two giants in WASP-94 must be mutually misaligned. Otherwise both planets would be observed in transit.

Several key questions in exoplanetary science converge on this remarkable system. Among them are the process of planet formation in binary systems, the assembly and evolution of Hot Jupiters, and the relative distribution of planetary system architectures in our Galaxy.

WASP-94AB as a binary

A large fraction of star systems are multiple, and the likelihood that a star will have one or more stellar companions appears to increase along with its mass (Raghavan et al. 2010). Thus M dwarfs, the least massive stars, are also the least likely to occur in multiples, while O stars, the most massive, are rarely observed in isolation. More than half of all Sun-like stars (those with spectral types ranging from mid K to late F) are estimated to occur in multiple systems.

The most common configuration is binary, like WASP-94, with numbers falling off as multiplicity increases. For example, in the well-studied sample of stars and brown dwarfs within 10 parsecs (32.6 light years) of our Sun, 52% are single, 30% occur in binaries, 13% occur in triples, and 5% occur in quadruples and quintuples (RECONS Survey 2012). This is similar to the multiplicity distribution among main sequence Sun-like stars (spectral types F6-K3; n = 454) within 25 parsecs: 56% are single, 33% are binary, and 11% contain three or more stars (Raghavan 2010).

Considering double stars only, systems range from “contact binaries,” in which the outer envelopes of the two stars directly interact and the binary orbit is measured in hours, to extremely remote pairs that require tens of thousands of years to complete a single orbit. Generally, pairs at least 200 astronomical units (AU) apart are considered “wide” binaries (where 1 AU equals the distance between the Earth and the Sun). It remains uncertain how far apart two stars can stray while remaining gravitationally bound. Separations in excess of 10,000 AU have been observed.

Stellar separation is a critical determinant of the likelihood and the nature of planet formation in binary systems. We now know that binaries with orbital periods shorter than about 50 days can support systems in which all planets orbit the binary’s center of mass. The Kepler Mission has revealed eight such circumbinary systems since 2011 (Welsh et al. 2014, Chavez et al. 2015). In each one, the primary is a Sun-like star between about 0.7 and 1.4 Msol, and the secondary is either a Sun-like star of similar mass or, more often, an M dwarf. The semimajor axis of the shared stellar orbit is smaller than 0.3 AU, so that observers on a hypothetical rocky planet in each system would see two suns in the sky. As a result, the most popular descriptor for such objects is “Tatooine planets.”

Figure 2. Imaginary sunset over an Earth-like planet on a circumbinary orbit. Image credit: NASA/JPL-Caltech


Slightly wider binaries – those separated by about 0.5 AU to 15 AU – are incapable of supporting circumbinary planets, while their mutual proximity also prevents independent planet formation around either star. These conditions create a gap in the semimajor axis distribution of binary hosts, distinguishing circumbinaries from “circumprimary” systems. The former require semimajor axes smaller than 0.3 AU; the latter become possible at separations wider than 18 AU. We know of five planet-hosting binaries in which the shared semimajor axis of the two stars is between 18 and 24 AU (Gliese 86, Gamma Cephei, HD 196885, HD 41004, and Alpha Centauri). In each case, the proposed planet orbits only one of the two stars. To put those numbers in perspective: Uranus is separated from our Sun by a distance of 19 AU, completing a single orbital period in about 84 Earth years. The two stars of Alpha Centauri share a much more elliptical orbit with a semimajor axis of about 23 AU, completing a single period in 80 years.

However, the great majority of known planet-hosting binaries have separations wider than 100 AU, and more than half are wider than 1000 AU (Kaib et al. 2013). WASP-94 falls in this group, as the two stars are separated by at least 2700 AU.

Very few binaries in the current census support planetary systems around both stars, and those few have uniformly wide separations. Neveu-VanMalle and colleagues cite three previous detections – HD 20781/HD 20782, XO-2N/XO-2S, and Kepler-132AB – making WASP-94 the fourth overall. The Kepler-132AB system is poorly understood, apparently supporting only low-mass planets. The other three binaries host at least one gas giant each, providing a small but useful sample for comparison.

Table 1. Three Conjoined Systems with Gas Giants

Column 1 shows the star name; column 2, the star mass in Solar units (Msol); column 3, the stellar metallicity (Sun = 0); column 4, the projected separation of the two stars in astronomical units (AU); column 5, the spectral type; column 6, the system age in billions of years (Gyr); column 7, the planet designation; column 8, the planet mass in Jupiter units (Mjup); column 9, the orbital semimajor axis in AU; column 10, the orbital eccentricity; and column 11, the orbital period in days. Within errors (not reported), the two stars of WASP-94 have the same mass, metallicity, and effective temperature.

At a distance of only 35 parsecs, the yellow stars HD 20781 and HD 20782 co-host the nearest double planetary system. The former supports two planets with minimum masses in the range of Uranus and Neptune and semimajor axes smaller than 0.5 AU. The latter hosts a single gas giant on a long-period orbit with the highest eccentricity reported to date – a whopping 0.97, just shy of an escape trajectory. The two stars travel at an extremely wide separation. Both have approximately Solar metallicity, and one of them (HD 20782) is also a twin of our Sun in mass.

The XO-2 binary (XO-2 North and XO-2 South) is more distant, at about 150 parsecs, and it shared orbit is not as wide, though still immense. The metallicities of these late G stars are among the highest reported among planetary hosts. Burke and colleagues (2007) note that the stars’ unusual enrichment in metals explains why their spectral types are later than their masses would suggest. XO-2N supports a single hot Jupiter that is well-aligned with the stellar spin axis; XO-2S hosts one small and one large gas giant inside 0.5 AU.

WASP-94 is still farther away than XO-2, at about 180 parsecs. Both the stars and the planets in this system bear a clear family resemblance to those in the other two double systems. Although WASP-94A and B are not as widely separated as the other binaries, they are still well over 2000 AU apart. All three systems are mature, such that even the youngest (WASP-94) has probably completed at least 15 orbits of the Galactic Bulge. Notably, out of six stars in three double systems, exactly half host short-period gas giants.

WASP-94Ab and Bb as Hot Jupiters

Table 1 shows that both planets in WASP-94, as well as the single planet around XO-2N, are typical Hot Jupiters. Members of this species are defined as gas giants on orbits of 10 days or less, while gas giants are planets more massive than 0.18 Jupiter units (Mjup) consisting largely of hydrogen and helium (H/He). Within the full sample of well-characterized Hot Jupiters (now in excess of 260), the median mass is 0.92 Mjup, and more than 80% are between 0.3 Mjup and 3 Mjup. The median orbital period is 3.4 days, with 85% on periods shorter than 5 days. The WASP-94 twins are less massive than most (though still within the chunky 80%), while their periods are typical. The same applies to XO-2Nb, which has parameters very similar to WASP-94Bb.

By now it is well-established that Hot Jupiters tend to be less massive than planets on longer orbits. For all gas giants with semimajor axes larger than 0.5 AU, the median mass is more than double that of Hot Jupiters (2.0 Mjup vs. 0.92 Mjup). In these three double systems, for example, the most massive planets also have the widest orbits.

Both WASP-94 planets, as well as XO-2Nb, seem to be alone in their planetary systems, without any detectable neighbors. Again, this is a typical feature of Hot Jupiters: within the full sample, 97% have no reported companions. Nevertheless, we are justified in questioning whether the non-detection of additional planets in Hot Jupiter systems is truly evidence of absence. Few systems have been observed long enough or with sufficient precision to confirm or disprove their loneliness. Heather Knutson and colleagues (2014) recently assessed radial velocity trends in a sample of 51 transiting Hot Jupiter systems to estimate the occurrence rate of outer gas giant companions, concluding that about half of these systems had a second planet orbiting between 1 and 20 AU. Their findings contrast sharply with the depopulated systems implied by current data on confirmed planets.

enduring mysteries of Hot Jupiters

Since the discovery of the first Hot Jupiter two decades ago, astronomers have confronted three mysteries. The most immediately obvious, and hence the first, was this: How could a massive planet made mostly of H/He end up in a broiling orbit less than 15 million kilometers from its parent star? We now have several complementary answers, such that the creation of Hot Jupiters seems almost inevitable throughout the Galaxy.

The second mystery came with the earliest observations of transiting Hot Jupiters. Growing numbers of these objects had measured radii significantly larger than theoretical predictions, up to twice the radius of Jupiter (2 Rjup). A number of solutions have been proposed for that puzzle, too, but no single one seems to be universally accepted. For the purposes of the present discussion, therefore, we can dispense with further consideration of the “inflation problem,” especially since it doesn’t appear to be linked to stellar binarity or Hot Jupiter delivery.

The third mystery came after several measurements of the Rossiter-McLaughlin effect, again in transiting Hot Jupiters. It was just as shocking as the first: Many Hot Jupiters are severely misaligned with the spin axis of their parent stars, in some cases (such as WASP-94Ab) so tilted that their orbits are retrograde. These data suggest that mysteries number one and number three are closely connected. The formation mechanism of Hot Jupiters must be one of the principal factors, if not the deciding factor, in their orbital alignment. Given three competing origin stories, nevertheless, astronomers have yet to settle on a consistent scenario that can explain all the data.

formation theories

Formation in situ is universally rejected as unphysical. Instead, Hot Jupiters must have assembled at semimajor axes of at least a few AU, where solids were abundant in the protoplanetary nebula and orbital speeds favored rapid accretion of planetary cores. Then, at an uncertain phase in their evolution, some mechanism carried the young giants to hot orbits at their parent stars’ doorstep. Three different scenarios have been proposed:

Type II migration. This mechanism has been the prevailing explanation for Hot Jupiters since the 1990s. It involves gravitational interactions between a forming planet and the protoplanetary nebula in which it is embedded. A planet that reaches gas giant mass before the nebula disperses will open a gap in the cloud, and then spiral inward with the flow of gases until it reaches the nebula’s central cavity. In theory, the process of Type II migration is smooth, swift, and efficient.

Although astronomers originally thought that the inward spiral of a gas giant would prevent the formation of terrestrial planets in the inner nebula, and certainly destroy any small planets that had already formed there, several researchers conducted numerical simulations with quite different outcomes (Raymond et al. 2006, Mandell et al. 2007, Fogg & Nelson 2007). In these models, Type II migration most likely happens early in the evolution of the protoplanetary nebula. Although the passage of a Hot Jupiter through the inner system at this time would probably destroy or scatter a significant mass of solids, enough would remain for low-mass planets to coalesce in the giant’s wake. In fact, these studies indicated that watery or icy Super Earths orbiting between 0.1 and 1.0 AU were a potential outcome of Hot Jupiter migration.

The puzzle seemed to have only two missing pieces: a stopping mechanism that would halt migrating giants before they collided with the central star, and some kind of on/off switch that would enable some giants (e.g., Jupiter, Saturn, and the growing population of long-period extrasolar gas giants) to avoid Type II migration altogether and remain in the cool vicinity of their birthplace. However, neither of these apparent gaps is fatal, since various plausible explanations have been proposed for each contingency.

Yet the popularity of this scenario for Hot Jupiter delivery has fallen considerably since 2010 (Triaud et al. 2010). Interactions with a primordial gas disk cannot produce the polar or retrograde orbits observed in many Hot Jupiters. These cockeyed giants persuaded many astronomers that Type II migration was at best a partial, and at worst a fallacious explanation for Hot Jupiters. Even more recently, abundant data from the Kepler Mission have demonstrated that Hot Jupiters truly are alone on their blazing orbits. No low-mass planets have been detected in any Hot Jupiter system; those pleasant Super Earths from the simulations don’t seem to exist. Indeed, Hot Jupiters represent an architecture that appears antithetical to the formation and survival of low-mass planets in system habitable zones.

Planet scattering. Many extrasolar giants on orbits wider than about 0.1 AU have large eccentricities, in sharp contrast to the circular orbits of the gas giants in our Solar System. The most popular explanation for such orbits involves gravitational interactions that produce dynamical instabilities among young planets. Known as planet scattering, this mechanism can excite high orbital eccentricities and shrink or expand semimajor axes. It has evidently operated among a large fraction of extrasolar gas giants. Accordingly, planet scattering was another early suspect in the mystery of Hot Jupiter delivery (Ford & Rasio 1996).

Although some investigators found that this mechanism could not scatter planets as far inward as 0.05 AU, where most Hot Jupiters orbit (Adams & Laughlin 2003), later studies (e.g., Nagasawa et al. 2008) argued that hot, circular orbits could be achieved in two steps. First, dynamical interactions occur among two or more gas giants on cool orbits. One of the giants is scattered inward on a highly eccentric orbit that brings it within 0.1 AU at periastron (closest approach to the parent star). Then, tidal interactions between the star and the planet eventually circularize its orbit at a period shorter than 10 days.

Just as the stock of Type II migration fell, that of planet scattering soared. This mechanism appears capable of causing all the misalignments observed in transiting systems.

Kozai oscillations. But another explanation remains very much in play. It’s variously known as Kozai cycles and Kozai oscillations. (I’ve always found this one complicated.) In its standard form, it operates in planets orbiting one member of a stellar binary, but some research argues that it can also occur within a single planetary system, regardless of binarity, in the form of perturbations of a smaller planet by a bigger one.

A recent study found that binaries wider than 1000 AU are unstable over the main sequence lifetimes of their components, since Galactic tides are likely to perturb their orbits with every passage around the Bulge (Kaib et al. 2013). The older the system, the more times it has circled the Galaxy, and the more likely it is that the stars’ orbits have been tweaked into high eccentricities. Decentered binary orbits can then perturb the stars’ planetary systems, and even lead to the collision and merger of the two stars.

summing up

It’s difficult to assess the relative contributions of these three mechanisms to the architectures of the known Hot Jupiter systems. Some studies have tried to set a maximum percentage for the contribution of planet scattering and Kozai oscillations. For example, Naoz et al. (2012) argued that 30% of all Hot Jupiters and “up to 100%” of all misaligned Hot Jupiters were produced by the Kozai mechanism in stellar binaries. But the full sample has yet to be studied in enough detail to establish which host stars reside in binary or higher multiple systems, the likely semimajor axis or at least projected separation of each multiple system, the percentage of systems with outer planets in addition to Hot Jupiters, and the nature of those outer systems, if any.

(Happy New Year!)


Adams F, Laughlin G. (2003) Migration and dynamical relaxation in crowded systems of giant planets. Icarus, 163: 290-306.
Albrecht S, Winn JN, Johnson JA, Howard AW, Marcy GW, Butler RP, Arriagada P, Crane JD, Shectman SA, Thompson IB, Hirano T, Bakos G, Hartman JD. (2012) Obliquities of Hot Jupiter host stars: Evidence for tidal interactions and primordial misalignments. Astrophysical Journal, 757:18. doi:10.1088/0004-637X/757/1/18
Burke GJ, McCullough PR, Valenti JA, Johns-Krull CM, Janes KA, Heasley JN, Summers FJ, Stys JE, Bissinger R, Fleenor ML, Foote CN, Garcia-Melendo E. et al. (2007) XO-2b: Transiting Hot Jupiter in a metal-rich common proper motion binary. Astrophysical Journal 671, 2115-2128.
Chavez CE, Georgakarakos N, Prodan S, Reyes-Ruiz M, Aceves H, Betancourt F, Perez-Tijerina E (2015) A dynamical stability study of Kepler circumbinary planetary systems with one planet. Monthly Notices of the Royal Astronomical Society 446, 1283-1292.
Desidera S, Barbieri M. (2007) Properties of planets in binary systems: The role of binary separation. Astronomy & Astrophysics 462, 345-353.
Desidera S, Bonomo AS, Claudi RU, Damasso M, Biazzo K, Sozzetti S, Marzari F, et al. (2014) The GAPS programme with HARPS-N@TNG IV: A planetary system around XO-2S. Astronomy & Astrophysics 567, L6. Abstract:
Fogg MJ, Nelson RP. (2007) On the formation of terrestrial planets in hot-Jupiter systems. Astronomy & Astrophysics 461, 1195–1208. doi: 10.1051/0004-6361:20066171
Kaib N, Raymond S, Duncan M. (2013) Planetary system disruption by Galactic perturbations to wide binary stars. Nature 493, 381-384. Abstract:
Knutson HA, Fulton BJ, Montet BT, Kao M, Ngo H, Howard AW, Crepp JR, Hinkley S, Bakos GA, Batygin K, Johnson JA, Morton RD, Muirhead PS. (2014) Friends of Hot Jupiters I: A radial velocity search for massive, long-period companions to close-in gas giant planets. Astrophysical Journal 785, 126.
Mandell A, Raymond S, Sigurdsson S. (2007) Formation of Earth-like planets during and after giant planet migration. Astrophysical Journal 660, 823-844. Abstract.
Naoz S, Farr WM, Rasio FA. (2012) On the formation of Hot Jupiters in stellar binaries. Astrophysical Journal Letters, 754:L36. doi:10.1088/2041-8205/754/2/L36
Nagasawa M, Ida S, Bessho T. (2008) Formation of hot planets by a combination of planet scattering, tidal circularization, and the Kozai mechanism. Astrophysical Journal, 678: 498-508. Abstract.
Neveu-VanMalle M, Queloz D, Anderson DR, Charbonnel C, Collier Cameron A, Delrez L, Gillon M, Hellier C, Jehin E, Lendl M, Maxted P, Pepe F, Pollacco D, S├ęgransan D, Smalley B, Smith A, Southworth J, Triaud A, Udry S, West RG. (2014) WASP-94 A and B planets: hot-Jupiter cousins in a twin-star system. Astronomy & Astrophysics 572, A49.
Triaud AHMJ, Collier Cameron A, Queloz D, et al. (2010) Spin-orbit angle measurements for six southern transiting planets: New insights into the dynamical origins of hot Jupiters. Astronomy & Astrophysics 524, A25.
Raghavan D, McAlister HA, Henry TJ, Latham DW, Marcy GW, Mason BD, Gies DR, White RJ, ten Brummelaar TA. (2010) A survey of stellar families: Multiplicity of Solar-type stars. Astrophysical Journal Supplement 190, 1-42. Abstract.
Raymond SN, Mandell AM, Sigurdsson S. (2006) Exotic Earths: forming habitable planets with giant planet migration. Science 313,1413-1416.
Rasio FA, Ford EB. (1996) Dynamical instabilities and the formation of extrasolar planetary systems. Science 274, 954-956.
Welsh WF, Orosz JA, Short DR, Cochran WD, Endl M, Brugamyer E, Haghighipour N, et al. (2014) KIC 9632895: The tenth Kepler transiting circumbinary planet. September 2014. eprint arXiv:1409.1605. Abstract: