5. BROWN DWARFS IN BINARY SYSTEMS

Many of the original searches for BDs were imaging or radial velocity surveys for companions to nearby stars. That these searches were unsuccessful or had very low yields caused some of the pessimism about finding BDs before 1995. This pessimism was codified in the phrase "brown dwarf desert" (eg. Marcy & Butler 1998). One must remember that while it is convenient to search around stars, this covers only a subset of possible places to find BDs. The search that discovered GD 165B included several hundred white dwarf primaries, and that which uncovered Gl 229B tested several hundred M dwarf primaries. There have been numerous searches from the ground and with HST that came up empty around solar neighborhood G-M stars (eg. Forrest et al 1988; Henry et al 1990; Simons et al 1996), and Hyades low mass stars (Macintosh et al 1996; Reid & Gizis 1997; Patience et al 1998). These have been pretty successful at finding VLMS companions, but not clearcut BDs. We can conclude that there is a relatively low (<1%) fraction of stars with well-separated visible BD companions. Recent searches have had slightly better luck. LHS 102B (Goldman et al 2000) is an L-type companion to an early M star (found in a proper motion study of EROS observations), although it fails the lithium test. That does not exclude it from being a BD, but it must have a mass greater than 60 jupiters. This is also true for GD 165B; unfortunately we cannot quite be sure that these are not VLMS unless their precise ages can be determined (or if it turns out they are just under the minimum main sequence temperature). In a survey careful to examine only young systems, Rebolo et al (1998b) found a BD companion (G196-3) after searching only 60 M primaries (Fig. 6). This is one of the lowest mass confirmed BDs (about 20 jupiters, based on the high activity of its primary which implies a Pleiades age). The companion to GG Tau B (White et al 1999) is an even younger example of such a system. Itoh et al (1999) find very promising objects in SFR. There is good candidate in the TW Hya association that awaits the lithium test (Lowrance et al 1999). Note that in all cases, the BD companions have a relatively low mass primary and are young (except for GD 165B). Although there is an observational selection effect against finding faint companions to brighter primaries, the results in the next subsection indicate this is not the main reason for the lack of companions around higher mass primaries.

5.2 Radial Velocity Brown Dwarf Candidates

Sensitive radial velocity surveys of G-M stars do not suffer from the fading of BDs with age or brightness contrast. They examine a separation range closer in than that in the imaging surveys. Precision radial velocity (PRV) searches will automatically find BDs more easily than planets. It is possible that the first BD was in fact found this way. HD114762 was detected as a companion to a solar-type star by Latham et al (1989; Section 2.1). It has been generally referred to as an extrasolar planet, but the minimum mass for it (11 jupiters) is quite near the planet/BD boundary, so that the inclination correction is likely to push it into the BD range. Of course, it is possible that it may be pushed all the way into the stellar range; the likelihood of that depends on how sparsely populated the "brown dwarf desert" really is (see Section 6.2) The most extensive survey for dynamical BDs has been that of Mayor and colleagues, first with the CORAVEL and then the ELODIE instruments (eg. Mayor et al 1997,1998). They found that several percent of solar-type systems have reflex velocities suggesting companions with lower mass limits in the substellar range. The difficulty with these candidates is exactly that they have lower mass limits. For a particular case, one is never quite sure whether the correction will push it into the stellar mass range. On statistical grounds one can argue that all the inclination corrections cannot be large. The extent to which this argument can be made, however, depends on the intrinsic mass function of binary companions. To see this, imagine that there are no BD companions to solar-type stars. Then one will only find BD candidates in PRV studies that are stellar systems with sufficiently low inclinations. Indeed, about half of the Mayor BD candidates were eliminated recently by finding their orbital inclinations using Hipparcos data (Halbwachs et al 2000). None of the remaining candidates is incontrovertibly substellar. The PRV searches have found very few companions in the BD mass range (Marcy & Butler 1998, Mayor et al 1998), but a number in the planetary mass range (which are harder to detect). Taking all this into account, one might fairly conclude that the incidence of BD companions to stars with masses of 0.5 solar masses or more is quite low (not more than about 1%). In contrast, the incidence of stellar companions to such primaries is in the range 20-40%. This result is discussed in more detail in Section 6.2. There are no examples of unambiguous dynamical BDs at present.

5.3 Double Brown Dwarfs

The search for binary brown dwarfs (BD pairs) is barely underway. It is striking that several have already been found. Color-magnitude diagrams of Pleiades VLMS show a large spread that has been interpreted as being due mainly to unresolved binaries (Steele & Jameson 1995; Zapatero-Osorio 1997a}. The presence of an unresolved substellar secondary has been inferred from infrared spectroscopy of the Pleiades VLMS HHJ54 (Steele et al 1995). A search for visible binaries among the Pleiades BDs using HST (Martín et al 1998b) identified a few such pairs (but it is turning out that they all may be non-members). If the distribution of binary frequencies among Pleiades BDs were similar to those of young stars and G dwarfs, they should have found 4.5 binaries. Dynamical stripping of wide companions of low mass primaries should not have proceeded too far in the Pleiades, though it could explain the dearth of wide substellar companions in the Hyades (Reid et al 1998). There is essentially only one BD which has been searched for radial velocity variations, and that is PPl 15. Its binarity was suggested by its position in the color-magnitude diagram (Zapatero-Osorio et al 1997a). The fact that it does turn out to be a double-lined spectroscopic binary (with an eccentric orbit and a period of 6 days; Basri & Martín 1999) is remarkable. It seems to bode well for the discovery of a reasonable number of spectroscopic BD binaries. We do not know whether the distribution of separations for substellar binaries is different from that for stellar binaries. Another surprisingly successful effort has been made to find field BD pairs. In only 2 pointings in an HST survey for binaries among the nearby field BDs, Martín et al (1999b) found that one of the 3 original DENIS objects is a sub-arcsec double (with a projected separation of about 5 AU). It is worth remarking that this system (DENIS-P J1228-1547) offers the first real chance for a dynamical confirmation of substellar masses. HST may be able reveal its orbit in only a few more years. Other groups have discovered several similar systems among the 2MASS and DENIS objects (as yet unpublished, possibly including a second of the first 3 DENIS objects). Thus, searches for BD pairs with small (<50 AU) separations have been remarkably successful (though it is hardly a statistical sample). This suggests that if one looks for substellar companions in systems where the mass ratio and separation are not too high, there may be many of them.