Monday, January 2, 2017

2016: Backyard Bonanza

Figure 1. A snapshot of 79 exoplanetary systems located within 20 parsecs (65 light years), arranged by increasing distance from our Sun. Names in red mark planets added to the census in 2016. Spectral types are indicated by the color key at lower right. The inner circle has a radius of 5 parsecs, and each successive ring represents an increment of 5 parsecs. Stellar icons are arranged in 2 dimensions by right ascension, which is marked at the edge of the outer circle; declination is ignored. Note that this diagram shows the relative distance of planetary systems from our Sun, but not from each other.

Without doubt, the biggest exoplanetary news of 2016 was the announcement of a small planet orbiting Proxima Centauri, a dim red star that happens to be our Sun’s closest neighbor. Not only did this story dominate headlines in the popular media: a recent query of the SAO/NASA Astrophysics Data System returned a total of 28 scientific articles about Proxima Centauri (including the discovery paper) either published or in press over the past 5 months. My blog post on the detection (The Perils of Proxima) was the biggest click magnet for Back Alley Astronomy in 2016.

No other discovery, analysis, or commentary in the field of exoplanetary astronomy rivaled the news from Proxima. Given the imbalance, it would be a stretch to concoct a “Top Ten” list of extrasolar news for the year just ended. So I’ll content myself with a Top Five: 1) the announcement of Proxima Centauri b by Guillem Anglada-Escudé & colleagues (blogged here), 2) the radial velocity detection of a sixth planet in the well-known Kepler-20 system by Lars Buchhave & colleagues (blogged here), 3) the final data dump of the Kepler Mission, which added 1,284 mostly lonely planets to the extrasolar census in one fell swoop (blogged here), 4) the return of Rho Coronae Borealis b, one of the first exoplanets ever announced, which was widely rejected as a bona fide planet a few years ago and then restored to consensual reality by Benjamin Fulton & colleagues, and 5) the discovery by Michael Endl & colleagues of a new Jupiter analog, HD 95872 b, which like Proxima Centauri b is located right in the Sun’s back yard (announced in a preprint late in 2015 but not formally published – or noticed by me – until 2016). Since I haven’t yet written about the last two items, I’ll devote some words to these less heralded new arrivals, and then assess the message their discoveries might be telling us about our immediate Galactic neighborhood.

the newest, nearest Jupiter analog

HD 95872 b is a gas giant planet in a more or less circular orbit whose estimated period of 4375 days falls within 1% of the orbital period of Jupiter. It was reported along with another cool gas giant – the latter following a notably eccentric orbit around a somewhat more distant star, Psi Draconis – by Endl et al. (2016).

The host star, HD 95872, is located just 25 light years away (7.56 parsecs) in the constellation of Crater the Wine Bowl. By any definition, it is a Sun-like star, included in the respectable Henry Draper Catalog and assigned a spectral type of K0 V in the SIMBAD database. However, Endl’s group reports a few stellar oddities. First, this object was omitted from the Hipparcos Catalog of nearby stars, from which most targets monitored by radial velocity searches have been drawn. The omission means that HD 95872 did not receive a precise distance estimate during the Hipparcos mission, a factor noted without comment by Endl & colleagues. Second, the star’s assigned spectral type seems rather late (i.e., reddish) for the mass of 0.95 ±0.4 Solar (0.95 ±0.4 Msol) determined by Endl’s group. For context, Kepler-20 has a virtually identical mass, but its spectral type is G2. Third, at an estimated age of 10 billion years, a star near Solar mass is likely to be evolving into a subgiant – an expectation contradicted by the classification of HD 95872 as a main sequence star by SIMBAD as well as Endl’s group.

HD 95872 bears a notable resemblance to 55 Cancri, another nearby star that also happens to host a gas giant in an orbit similar to Jupiter’s. Both the mass (0.90 ±0.015 Msol) and the age (10.2 billion years) of 55 Cancri are similar to those of HD 95872, and both stars are also significantly enriched in metals. According to von Braun & colleagues (2011), the metallicity of 55 Cancri is +0.31, while Endl & colleagues report +0.41 for HD 95872. Given these similarities, it’s illuminating to remember that the spectral classification of 55 Cancri is subject to disagreement, with published types ranging from G8 V to K0 IV. (Note that alternative values of 3-9 billion years are available for the age of 55 Cancri; see Teske et al. 2013.)

As substantial research has shown, metal-rich stars like 55 Cancri and HD 95872 have a high likelihood of supporting gas giant planets. 55 Cancri hosts two: planet b with an orbital radius of 0.11 astronomical units (AU) and a minimum mass 0.83 times Jupiter (0.83 Mjup), and planet d with an orbital radius of 5.76 AU and a minimum mass of 3.84 Mjup. 55 Cancri also hosts two tweens (objects intermediate in mass between Neptune and Saturn) in the gap between planets b and d, plus a dense, highly irradiated Super Earth interior to planet b with a mass of about 8 Earth units (8 Mea) and an orbital period shorter than 24 hours. These massive progeny illustrate the superior planet-forming potential of metal-rich protoplanetary disks.

Our new friend, HD 95872, has only a single reported planet, for which Endl & colleagues provide a minimum mass of 4.6 Mjup. This is similar to the estimate for 55 Cancri d, but well above the median for the full census of gas giants. For orbital period and semimajor axis, Endl’s group reports values almost identical to those of Jupiter: respectively 4375 days and 5.2 AU. In addition, they find a relatively circular orbit, although their results for eccentricity are understandably imprecise: 0.06 ±0.04. Within uncertainties, this value is consistent with the well-measured eccentricity of Jupiter: 0.048.

Endl & colleagues confidently identify both HD 95872 b and their other new planet, Psi Draconis b, as Jupiter analogs. They define such planets as “within a factor of a few Jupiter-masses and in orbits longer than 8 years.” As explained in a previous post (Almost Jupiter), I find this definition incomplete, since I’m more interested in analogs of our Solar System than in cool gas giants per se. My definition of a Jupiter analog is 1) a gas giant planet (minimum mass 0.16 Mjup/50 Mea) that is 2) located outside the system’s liquid water zone 3) with an orbit whose eccentricity is under 0.3 and 4) with a semimajor axis that permits the survival of terrestrial planets on habitable orbits but 5) without any gas giant companions on interior orbits. A Solar System analog contains a cool giant that meets all these criteria and is centered around a Sun-like star in the mass range of 0.65-1.30 Msol, corresponding to spectral types between mid K and late F.

HD 95872 nicely satisfies all these criteria and thereby assumes the status of the nearest Solar System analog, demoting HD 154345 to the second-nearest. But Psi Draconis, the other system reported by Endl et al., does not meet my standard. Despite its appealingly memorable name, this system is disqualified by the significant orbital eccentricity (0.4) of its gas giant, Psi Draconis b. It’s a disappointing result, since the host star is very similar to our Sun in age, mass, and spectral type, and it’s only 22 parsecs away.

In light of these recent discoveries, I conducted a new review of the full sample of exoplanets detected by radial velocity measurements (the only technique that has managed to characterize any Solar System analogs). I based the review on a December download of the catalog of the Extrasolar Planets Encyclopaedia (EPE), with additional revisions based on discovery papers. The results appear in Table 1. Systems are listed by increasing semimajor axis of the Jupiter analog.

Table 1. Seventeen potential analogs of our Solar System
Tags: Type = spectral type; Msol = star mass in Solar units; [Fe/H] = metallicity; Dist. = distance in parsecs (rounded); Mjup = planet mass in Jupiter units; a = semimajor axis in Earth units; e = orbital eccentricity; Period = orbital period in years. Selection criteria: star mass 0.65-1.30 Msol; a > 3 AU; e < 0.3, no interior giants perturbing the system habitable zone. Note: HD 30177 has a second planet, a massive “Saturn analog” on an orbit exterior to planet b, with an uncertain orbital period.

This is the third time I’ve searched the EPE sample for Solar System analogs during the past few years (previously blogged here and here), and each time the population has grown. HD 95872 is an especially welcome addition. Not only is it the nearest Jupiter analog – it also has the longest orbital period of the lot, making it Jupiter’s closest rival within this subset.
The 17 systems summarized in Table 1 bear a strong family resemblance, at least in terms of their gross parameters. But how much of the apparent similarity might stem from selection biases?

Distance: All but 2 are located at distances between 25 and 60 parsecs; no Solar System analogs have been reported outside that space. Their scarcity inside 25 parsecs might be an accurate reflection of the absolute rarity of such systems, but their absence beyond 60 parsecs almost certainly results from historical limits on the sensitivity of radial velocity searches.

Star mass: Even given my inclusion criterion of 0.65-1.3 Msol, the host stars fall in a narrow range of masses – 0.88-1.18 Msol – and center on a narrow range of spectral types – early to mid G. All are thus extremely similar to our Sun (type G2), while stars less massive than 0.85 Msol are conspicuously absent from the list. Since the entire population is based on continuing searches that began decades ago, this outcome might simply reflect the preference of early search programs for G-type stars. On the other hand, the non-appearance of stars in the range of 0.65-0.85 Msol might be significant. Maybe architectures like the Solar System are restricted to more massive stars.

Metallicity: Eleven out of 17 host stars have Solar metallicity or less; only 4 have [Fe/H] greater than +0.2, which would indicate a notable enhancement in heavy elements. My own (not very educated) guess is that this relative indifference to metallicity might be real. In any case, the scarcity of extremely metal-rich stars in this population supports the premise that our Sun is a typical host of Jupiter analogs. It also warrants a reminder that even the Sun is metal-rich compared to the average star in our Galactic neighborhood, where typical metallicities are about -0.10. Whether or not cool giants on circular orbits are intrinsically rare around stars impoverished in heavy elements, no extreme metallic enrichment seems necessary to form them.

Planet mass: The median gas giant mass in this sample is approximately 2 Mjup, and only 2 of these “Jupiter analogs” have masses smaller than 0.99 Mjup. This outcome might be just another reflection of the limited sensitivity of the radial velocity method, since planets of lower mass are more difficult to detect on long-period orbits. But it might instead be a clue that Jupiter is relatively small for a cool giant, and that most of its analogs are more massive.

Orbital eccentricity: Most gas giants with semimajor axes of 3 AU or more have eccentricities in excess of 0.5, and many giants with smaller eccentricities have giant companions on interior orbits that rule out Earth-like planets. Even in Table 1, which includes only unaccompanied giants with eccentricities < 0.3, the median value of 0.1 is relatively high by the standards of our Solar System. Of course, it’s possible that exoplanetary eccentricities are systematically overestimated (Shen & Turner 2008), and that the giants in Table 1 are actually less eccentric (and therefore friendlier) than they look on the screen. But it’s also possible that cool giants on orbits as circular as Jupiter’s are even rarer than Hot Jupiters. If eccentricity < 0.1 turns out to be the limiting factor for Solar System analogs, then they are much less numerous than I project here.


On the other hand, if Table 1 presents a reasonable survey of systems like our own, the outlook for Earth-like planets is rosy. The available data now show that 2 out of 260 star systems (0.8%) identified within 10 parsecs – including almost 3% of the 70 FGK stars in this space – host a gas giant on a cool orbit that would permit the survival of Earth-like planets in the habitable zone. Since one of those two star systems (our own) is known to support such a world, we’re entitled to conjecture that the other one (HD 95872) does, too.

If we restrict our scope to stars other than our Sun with confirmed planets, rather than all known stars, we can say that 2% (N=2) of the 79 exoplanetary systems located within 20 parsecs qualify as Solar System analogs. Furthermore, if Jupiter analogs are at least as common as Hot Jupiters, as various analyses suggest, then 3 more systems resembling our own might be awaiting discovery within the same volume of space. That prediction rests on the fact that five Hot Jupiters are already known in this region. If we further extend this back-of-the-envelope approach to a radius of 60 parsecs, we can say that a dozen more Solar System analogs await detection in the immediate Solar neighborhood, since more than 30 Hot Jupiters systems have been reported in that space, but only 17 bona fide analogs of our system.

counting planets in the Sun’s back yard

The statistics on exoplanets within any volume-limited sample depend sensitively on how many such planets we regard as validated. Unfortunately, different exoplanetary catalogs offer different pictures even of the well-studied space within 20 parsecs. Accordingly, I have to approach this problem with a consistent set of exclusions: Alpha Centauri b, Kapteyn’s b & c, all companions identified as brown dwarfs, and all objects reported by direct imaging or astrometry, with the exception of Beta Pictoris b.

When I apply these criteria to the full census maintained by EPE, along with corrections for stellar parameters found in the literature, I get the 79 systems pictured in Figure 1. Given additional exclusions discussed in a series of older blog posts (starting here, most recently here), these 79 systems support a total of 141 planets.

However, if I apply the same criteria to the NASA Exoplanet Archive, I get a notably smaller sample of 65 systems supporting only 113 planets. Much of the mismatch can be explained by NASA’s exclusion of candidates reported in an unpublished manuscript by Michel Mayor & colleagues (2011), as well as other candidates proposed by other investigators on the strength of Bayesian reanalyses of existing radial velocity data (see list of exclusions here). Thus even my relatively skeptical approach might err on the side of optimism.

The difficulty of establishing an accurate count of the known exoplanetary systems, let alone the very nearest, is illustrated by the observational history of Rho Coronae Borealis. This Sun-like star, one of the first to be identified as an exoplanetary host, is located 17.24 parsecs (56 light years) away. In 1997, Robert Noyes & colleagues reported an object (b) with a minimum mass of 1.1 Mjup orbiting this star in a period of 39 days with an eccentricity of 0.03. Although their radial velocity measurements could not rule out much higher masses, they presented theoretical arguments in favor of a gas giant planet instead of a brown dwarf star.

Follow-up studies summarized by Fulton et al. (2016) presented additional astrometric data that, according to several astronomers, unmasked the so-called exoplanet as a very dim M dwarf or massive brown dwarf.  In this argument, the original mass value reported by Noyes & colleagues was a drastic underestimate, because we happen to observe the system from a face-on viewing angle. Such a geometry minimizes the Doppler shift detectable from Earth, whereas an edge-on angle maximizes it. According to this argument, the companion must be at least 100 times more massive than Jupiter.

Most radial velocity detections that are not supported by transit observation are vulnerable to a similar challenge, but the prevailing attitude in the exoplanet community seems to be “innocent until proven guilty.” In the case of Rho Coronae Borealis b, the skeptical argument quietly won the day, and the proposed planet was excluded from EPE sometime around 2011.

Now Fulton’s group has conducted a concentrated observing program to test the planetary hypothesis. Their findings confirm the existence of planet b while ruling out even the dimmest of M dwarfs. Thus Rho Coronae Borealis, jewel in the Northern Crown, has returned to the fold of exoplanet host stars.

Fulton & colleagues also report a second planet (c) with an orbital period of 102 days and a semimajor axis of 0.41 AU, similar to Mercury’s. Depending on our viewing angle, the planet’s minimum mass of 25 Earth units (Mea) suggests either a large low-mass planet like GJ 436 b (a “Super Neptune” or tween) or a small gas giant. In the first alternative, Rho Coronae Borealis would present a relatively rare and extremely interesting architectural feature: a pair of adjacent planets in which the inner object is a gas giant and the outer is a low-mass planet like Neptune. Apart from Rho Coronae Borealis, just five examples of this architecture have been observed: our Solar System (Saturn + Uranus), GJ 876 (planets b + e), Kepler-87 (planets b + c), Kepler-89 (planets d + e), and WASP-47 (planets b + d).

In the second alternative, Rho Coronae Borealis might a scaled-down version of 55 Cancri, which presents an adjacent pair in which the inner object is a massive gas giant (planet b) and the outer is a tween (planet c) of about 55 Mea.


One of my favorite articles of 2016 was “Challenges in Planet Formation,” by Alessandro Morbidelli & Sean Raymond, two seasoned veterans of exoplanetary science. The authors frankly discuss everything we don’t really know about the evolution of planetary systems, which their opening sentence describes as “a vast, complex, and still quite mysterious subject.” The same characterization could easily apply to our understanding of exoplanetology, given the vast, mysterious, and constantly expanding nature of the extrasolar census. As with planetary evolution, when it comes to the demographics of exoplanets, “even our most successful models are built on a shaky foundation” (Morbidelli & Raymond 2016).

So here’s looking forward to the mysterious, terrifying, and inevitably fascinating events and revelations to come in 2017 . . . .

Endl M, Brugamyer EJ, Cochran WD, MacQueen PJ, Robertson P, Meschiari S, Ramirez I, Shetrone M, Gullikson K, Johnson MC, Wittenmyer R, Horner J, Ciardi DR, Horch E, Simon AE, Howell SB, Everett M, Caldwell C, Castanheira BG. (2016) Two new long-period giant planets from the McDonald Observatory Planet Search and two stars with long-period radial velocity signals related to stellar activity cycles. Astrophysical Journal 818, 34. Abstract: 2016ApJ...818...34E
Fulton BJ, Howard AW, Weiss LM, Sinukoff E, Petigura EA, Isaacson H, Hirsch L, Marcy GW, Henry GW, Grunblatt SK, Huber D, von Braun K, Boyajian TS, Kane SR, Wittrock J, Horch E, Ciardi DR, Howell SB, Wright JT, Ford EB. (2016) Three temperate Neptunes orbiting nearby stars. Astrophysical Journal 830, 46. Abstract: 2016ApJ...830...46F
Mayor M, Marmier M, Lovis C, Udry S, Ségransan D, Pepe F, Benz W, Bertaux J-L, Bouchy F, Dumusque X, Lo Curto G, Mordasini C, Queloz D, Santos NC. (2011) The HARPS search for southern extra-solar planets XXXIV. Occurrence, mass distribution and orbital properties of super-Earths and Neptune-mass planets. Unpublished; abstract at
Morbidelli A, Raymond S. (2016) Challenges in planet formation. Invited review; in press. Abstract: 2016arXiv161007202M
Noyes R, Jha S, Korzennik SG, Krockenberger M, Nisenson P, Brown TM, et al. (1997) A planet orbiting the star Rho Coronae Borealis. Astrophysical Journal 483, L111-L114.
Shen Y, Turner EL. (2008) On the eccentricity distribution of exoplanets from radial velocity surveys. Astrophysical Journal 685, 553-559.
Teske JK, Cunha K, Schuler SC, Griffith CA, Smith VV. (2013) Carbon and oxygen abundances in cool metal-rich exoplanet hosts: A case study of the C/O ratio of 55 Cancri. Astrophysical Journal 778, 132.
von Braun K, Boyajian TS, ten Brummelaar TA, Kane SR, van Belle GT, Ciardi DR, Raymond SN, Lopez-Morales M, McAlister HA, Schaefer G, Ridgway ST, Sturmann L, Sturmann J, White R, Turner NH, Farrington C, Goldfinger PJ. (2011) 55 Cancri: Stellar astrophysical parameters, a planet in the habitable zone, and implications for the radius of a transiting Super-Earth. Astrophysical Journal 740, 49.