Gas Giant vs. Brown Dwarf. The image on the left is Jupiter, courtesy NASA/JPL-Caltech. The image on the right is an unknown artist's impression of a brown dwarf or "failed star." These two objects are shown at their approximate relative sizes. As I. Neill Reid points out, a brown dwarf of spectral type T, with a mass about 40 times that of Jupiter, will probably have a radius almost identical to Jupiter's. The same applies to so-called Megajovian planets with masses between 5 and 25 MJUP – all are likely to have radii within 10% of Jupiter's.

As exoplanet discoveries began to accumulate in the late 1990s, one of the many surprises that they presented was their enormous weight – often two, three, or more times the mass of Jupiter. Currently, the median mass for known extrasolar gas giants is about 1.75 Mjup, which exceeds the aggregate mass of all eight planets in the Solar System. The heaviest 10% of exoplanets have masses well in excess of 7 Mjup, meaning that each one contains the equivalent of four or five Solar Systems' worth of planets (EPE, CNE).

The commonly cited parameters for exoplanets actually understate their likely masses, because in all but a few systems, we have only minimum values and not actual values. Radial velocity data cannot yield precise exoplanetary masses unless the orbital inclination of the object is also known. Minimum masses are calculated on the assumption that we view the planet's orbit precisely edge-on.

For example, radial velocity data alone indicate a minimum mass of 1.67 Mjup for Mu Arae b, but one dynamically feasible orbital solution requires an actual mass of 5.07 Mjup for this planet (Short et al. 2008). The true mass of some exoplanets may exceed the minimum mass estimate by several hundred percent.

Since the 1990s, terms like Megajovian and Super Jupiter have been used to denote planets many times heavier than Jupiter. One of the earliest to be detected was the single companion of the nearby Sun-like star 70 Virginis. The object's minimum mass was originally calculated as 6.6 Mjup, and its maximum mass was estimated as 20-40 Mjup (Marcy & Butler 1996). As the discovery team noted, the higher values would require this object to be a brown dwarf or "failed star" rather than a gas giant planet. Nevertheless, they were able to conclude on statistical grounds that 70 Virginis b is most likely a gas giant between 6.6 Mjup and 9 Mjup – and their conclusion has held up well in the dozen years since.

The problem of distinguishing between large planets and small brown dwarfs has also persisted. By now, more than 40 objects on planet-like orbits have been announced in the mass range between 5 Mjup and 23 Mjup. Various criteria have been have been suggested to separate the Megajovians from the dwarfs.

1. The most popular is mass. Brown dwarfs are hot gaseous bodies composed mostly of hydrogen, ranging from 13 Mjup to 80 Mjup (Grether & Lineweaver 2006). The bottom of this range is determined by the minimum mass needed to initiate nuclear fusion in deuterium atoms. Thus a common strategy is to classify all objects less massive than 13 Mjup as planets, and all objects more massive as brown dwarfs. As noted above, however, for all but a few exoplanets the true mass is unknown, so this criterion is difficult to apply in practice. Still worse, as I. Baraffe and colleagues point out, even a bona fide gas giant of 12 Mjup or more will experience deuterium fusion, thus highlighting “the utter confusion provided by a definition of a planet based on the deuterium-burning limit” (Baraffe et al. 2008).
   
   
2. Another obvious distinction is history. Brown dwarfs form like other stars in the fragmentation and collapse of hydrogen clouds – perhaps in the accretion disks surrounding pre-main sequence binary stars (Jiang et al. 2004) – while gas giants assemble by accretion of solids and gases in a circumstellar disk, during or after the last stage of star formation. In theory these mutually exclusive histories could provide a foolproof criterion, but in practice it is impossible to know the precise evolutionary history of any given substellar-mass object.
   
   
3. The least ambiguous marker so far suggested is orbital environment. Except for the poorly understood category of runaway or ejected bodies, planets are found exclusively in orbit around stars, while brown dwarfs typically occur as single objects. Only about 10%-20% of brown dwarfs occur in binary systems, and in this group both the primary and the secondary stars are usually brown dwarfs with a separation no larger than 20 AU (Grether & Lineweaver 2006, Ahmic et al. 2007, Kraus et al. 2008). Only a tiny fraction occur as wide companions (50-150 AU), whether around other brown dwarfs or, rarest of all, around Sun-like stars (spectral types F, G, K).

   As Marcy and Butler noted in 2000, and as many succeeding studies have confirmed, gas giant planets are actually much more abundant than brown dwarfs within 3 AU of Sun-like stars (Marcy & Butler 2000). Marcy & Butler estimated that less than 0.5% of FGK stars have brown dwarf companions orbiting within 3 AU, whereas recent data indicate that at least 10.5% of Sun-like stars host gas giants in the same orbital space (Cumming et al. 2008).

The dearth of dwarfs on planet-like orbits is often called the “brown dwarf desert” (Halbwachs et al. 2000). Robust surveys have shown that the desert extends to large radii, from tens of AU to more than 1000 AU. Fewer than 1% of Sun-like stars, whether in the immediate Solar neighborhood or in nearby star-forming regions, are likely to harbor a brown dwarf companion at any detectable radius (Gizis et al. 2001, McCarthy & Zuckerman 2004, Grether & Lineweaver 2006, Kraus et al. 2008). By contrast, about 18% of Sun-like stars are expected to host gas giants within 20 AU (Cumming et al. 2008).

Thus an object with a minimum mass of about 25 Mjup or less, orbiting within 20 AU of a Sun-like star, is much more likely to be a planet than a brown dwarf.

Last update September 2008