2. The Search for Brown Dwarfs

The search for brown dwarfs can be divided into three qualitatively different arenas. The most obvious is the search for old visible BDs, whose temperature and luminosity obviously lie below the minimum possible value for stars. This can be done by looking for stellar companions, or in the field. The second search arena is for dynamical BDs, where orbital information suggests a mass below the minimum stellar mass. Here one need not be able to actually see the BD; its effect on a stellar companion can reveal it. If the orbit is spatially resolved their actual masses can be found, otherwise only lower limits can be placed due to the unknown orbital inclination.

The third arena in which to search is for young BDs, which are visible and at their brightest. While these are the easiest to see, it is harder to be sure that they are really BDs. At early ages, the BDs occupy the same region of temperature and luminosity as very low mass stars (VLMS). One can trade off mass and age to infer either a BD or a VLMS at a given observed value of luminosity or temperature. For isolated objects in the field this is a particularly acute problem, since their age is not generally known. Even in a cluster, the mere fact that an object occupies a position in an HR or color-magnitude diagram where theory tells us to expect BDs at the age of the cluster, has not proven convincing by itself. This is partly because the theory which converts observational quantities to mass is still being refined, and partly because other factors may invalidate the conclusion. Among these are the possibility that the object may not actually be a member of the cluster, or that the age of the cluster may have a large spread or may not have been correctly determined.

2.1 A Brief History of the Searches

A review of early observational efforts can be found in Oppenheimer et al (2000). One of the first efforts to directly image BDs as companions to nearby stars was made by McCarthy et al (1985). Using an infrared speckle technique, they reported a companion to VB8, with inferred properties that would guarantee its substellar status. This was the highlight of the first conference on brown dwarfs (Kafatos 1986). Unfortunately, their result was never confirmed. Later surveys (eg. Skrutskie et al 1989; Henry & McCarthy 1990) did not find good BD candidates (but several VLMS companions). In a survey of white dwarfs, Becklin & Zuckerman (1988) turned up a very red and faint companion, GD 165B, whose spectrum was quite enigmatic. Kirkpatrick et al (1999b) argue that this is probably a BD.

The next good candidate came from a radial velocity survey. Latham et al (1989) were conducting a survey of about 1000 stars with 0.5 km/s precision. Among their roughly 20 radial velocity standards, HD 114762 exhibited periodic variations just at the limit of detectability. This orbit has been confirmed by the precision radial velocity groups, and implies a lower mass limit for the companion of about 11 jupiters. The difficulty is that the orbital inclination is not known. It would not be too surprising to find a very low inclination stellar companion in a sample of 1000 stars, but much more surprising in a sample of 20. This argument remains unsettled, though subsequent surveys have shown a real dearth of companions to solar-type stars in the BD mass range (see Sections 5.2,6.2). Until the actual orbital inclination for this object is measured (by a space interferometer?), it must remain unconfirmed but tantalizing.

During the early nineties, there were a number of surveys aimed at finding BDs in young clusters. Forrest et al (1989) announced a number of candidates in Taurus-Aurigae, which were later shown to be background giants (Stauffer et al 1991). Surveys of star forming regions (eg. Williams et al 1995) also found objects which might well be substellar, but with no obvious way to confirm them. Hambly et al (1993; HHJ) conducted a deep proper motion survey of the Pleiades, and found a number of objects that models suggested should be substellar. Stauffer et al (1994) were also conducting a survey for BDs in this cluster, working from color-magnitude diagrams. Both surveys went substantially deeper than before, and uncovered interesting objects. This set the stage for the next, ultimately successful, effort to find cluster BDs. Nonetheless, it is well to remember that at the ESO Munich conference on “The Bottom of the Main Sequence - and Beyond” (Tinney 1994), there was a palpable sense of frustration at the failure of many efforts to confirm a single BD.

Working from the new Pleiades lists, Basri and collaborators were finally able to announce at the June 1995 meeting of the AAS (Science News 147, p. 389) the first successful application of the lithium test for substellarity (Section 3.1). This was the first public declaration of a BD that is currently still solid. The object, PPl 15, would have an inferred mass well below the substellar limit, except that concurrently the age of the Pleiades was revised substantially upward (Section 3.2.1). This moved the mass of PPl 15 just under the substellar limit. Along with community unfamiliarity with the lithium test, it delayed acceptance of PPl 15 as a true BD (there is no question about it now, Section 5.3). Given this fact, any fainter Pleiades members should automatically be BDs. In September, Rebolo et al (1995) announced the discovery of such an object: Teide 1. Any remaining doubt could be removed by confirming lithium in it; which was accomplished by Rebolo et al (1996). These two objects are now accepted as undeniable BDs (along with many subsequently discovered faint Pleiades members). Their masses are in the 55-70 jupiter range.

2.2 The First Incontrovertible Brown Dwarf: Gl 229B

Only a month after the publication of Teide 1, any debate over the existence of brown dwarfs was ended by the announcement in Florence (at the Tenth Cambridge Cool Stars Workshop) of the discovery of a very faint companion to a nearby M star. Its temperature and luminosity are well below the minimum main sequence values. Along with the revelation at the same session of the first extrasolar planet, it was suddenly very clear that Nature has no problems manufacturing substellar objects.

Gl 229B was found in a coronographic survey of nearby low mass stars (Nakajima et al 1995). The survey was originally chosen to be biased towards younger M stars (though not strictly so). It ended up as a complete survey of stars to 8pc (almost 200 targets; Oppenheimer 1999). Of these, only Gl 229 shows a substellar companion. The companion was first detected in 1994, but the group showed commendable forbearance in waiting for proper motion confirmation that it was physically associated with the primary (allowing the known parallax of the primary to be applied to find its luminosity). They also obtained a spectrum that confirmed the remarkably low temperature implied by its luminosity (Oppenheimer et al 1995). In particular, the spectrum contains methane bands at 2 microns; features that had previously been detected only in planetary atmospheres (and which are not expected in any main sequence star).

The mass of Gl 229B is still somewhat uncertain. Its large separation from the primary means we will have to wait a few decades to find a dynamical mass from the orbit. The primary, though a member of the young disk population kinematically, is not a particularly active star. The uncertainty in age translates directly to a possible mass range. There is only a weak constraint on the gravity from atmospheric diagnostics. The allowed mass is from about 20-50 jupiters, with 40 jupiters as a reasonable value to take for now (given the inactivity of the primary, which implies an older age). There are a number of BDs found since that have masses lower than Gl 229B; it is distinguished by being the coolest BD (and therefore the oldest). This was a watershed discovery in the search for BDs; the next example of a similar object was not found until 1999.