The Formation of Brown Dwarfs: ObservationsKevin L. LuhmanThe Pennsylvania State UniversityViki JoergensThe University of LeidenCharles LadaSmithsonian Astrophysical ObservatoryJames Muzerolle and Ilaria PascucciThe University of ArizonaRussel WhiteThe University of AlabamaWe review the current state of observational work on the formation of brown dwarfs,focusing on their initial mass function, velocity and spatial distributions at birth, multiplicity,accretion, and circumstellar disks. The available measurements of these various properties areconsistent with a common formation mechanism for brown dwarfs and stars. In particular, theexistence of widely separated binary brown dwarfs and a probable isolated proto-brown dwarfindicate that some substellar objects are able to form in the same manner as stars through un-perturbed cloud fragmentation. Additional mechanisms such as ejection and photoevaporationmay play a role in the birth of some brown dwarfs, but there is no observational evidence todate to suggest that they are the key elements that make it possible for substellar bodies to form.1. INTRODUCTIONAlthough many of the details are not perfectly under-stood, stars and giant planets are generally believed to formthrough the collapse of molecular cloud cores and the ac-cretion of gas by rocky cores in circumstellar disks, respec-tively. In comparison, the formation of objects intermedi-ate between stars and planets – free-floating and companionbrown dwarfs – has no widely accepted explanation. A pri-ori, one might expect that brown dwarfs form in the samemanner as stars, just on a much smaller scale. However, al-though self-gravitating objects can form with initial massesof only ∼ 1 MJupin simulations of the fragmentation ofmolecular cloud cores, these fragments continue to accretematter from their surrounding cores, usually to the point ofeventually reaching stellar masses (Boss, 2001; Bate et al.,2003). Thus, standard cloud fragmentation in these modelsseems to have difficulty in making brown dwarfs. One pos-sible explanation is that the simulations lack an importantpiece of physics (e.g., turbulence), and brown dwarfs areable to form through cloud fragmentation despite their pre-dictions (e.g., Padoan and Nordlund, 2004). Another pos-sibility is that a brown dwarf is born when cloud fragmen-tation is modified by an additional process that prematurelyhalts accretion during the protostellar stage, such as dynam-ical ejection (Reipurth and Clarke, 2001; Boss, 2001; Bateet al., 2002) or photoevaporation by ionizing radiation frommassive stars (Kroupa and Bouvier, 2003; Whitworth andZinnecker, 2004). This uncertainty surrounding the forma-tion of brown dwarfs has motivated a great deal of theoreti-cal and observational work over the last decade.In this paper, we review the current observational con-straints on the formation process of brown dwarfs (BDs),which complements the theoretical review of this topic pro-vided in the chapter by Whitworth et al. By the nature ofthe topic of this review, we focus on observations of BDsat young ages (τ < 10 Myr), although we also considerproperties of evolved BDs that provide insight into BD for-mation (e.g., multiplicity). A convenient characteristic ofyoung BDs is their relatively bright luminosities and warmtemperatures compared to their older counterparts in the so-lar neighborhood, making them easier to observe. However,because the luminosities and temperatures of young BDsare continuous extensions of those of stars, positively iden-tifying a young object as either a low-mass star or a BD isoften not possible. The mass estimates for a given objectvary greatly with the adopted evolutionary models and themanner in which observations are compared to the modelpredictions. Using the models of Baraffe et al. (1998) andChabrier et al. (2000) and the temperature scale of Luhmanet al. (2003b), the hydrogen burning mass limit at ages of0.5-3 Myr corresponds to a spectral type of ∼M6.25, which1Fig. 1.— IMFs for Taurus (Luhman, 2004c), IC 348 (Luhman et al., 2003b), Chamaeleon I (Luhman, in preparation), and the TrapeziumCluster (Muench et al., 2002). The completeness limits for these measurements are near 0.02 M¯(dashed lines). In the units of thisdiagram, the Salpeter slope is 1.35.is consistent with the dynamical mass and spectral type ofthe first known eclipsing binary BD (Stassun et al., 2006)and other observational tests (Luhman and Potter, 2006).Therefore, we will treat young objects later than M6 as BDsfor the purposes of this review.2. INITIAL MASS FUNCTIONOne of the most fundamental properties of BDs is theirinitial mass function (IMF). Because BDs are brightestwhen they are young, star-forming regions and young clus-ters are the best sites for finding them in large numbersand at low masses, which is necessary for measuring sta-tistically significant IMFs. Spectroscopic surveys for BDshave been performed toward many young populations (τ <10 Myr) during the last decade, including IC 348 (Luh-man et al., 1998, 2003b, 2005a; Luhman, 1999), Taurus(Brice˜no et al., 1998, 2002; Mart´ın et al., 2001b; Luh-man, 2000, 2004c, 2006; Luhman et al., 2003a; Guieuet al., 2006), Chamaeleon I (Com´eron et al., 1999, 2000,2004; Neuh¨auser and Comer´on, 1999; Luhman, 2004a,b;Luhman et al., 2004), Ophiuchus (Luhman et al., 1997;Wilking et al., 1999; Cushing et al., 2000), Upper Scor-pius (Ardila et al., 2000; Mart´ın et al., 2004), Orion (Hil-lenbrand, 1997; Lucas et al., 2001; Slesnick et al., 2004),NGC 2024 (Levine et al., 2006), NGC 1333 (Wilking et al.,2004), TW Hya (Gizis, 2002; Scholz et al., 2005), λ Ori(Barrado y Navascu´es et al., 2004b), and σ Ori (Barradoy Navascu´es et al., 2001, 2002; B´ejar et al., 1999, 2001;Mart´ın et al., 2001a; Zapatero Osorio et al., 1999, 2000,2002a,b,c; Mart´ın and Zapatero Osorio, 2003).We now examine the IMF measurements for IC 348,Chamaeleon I, Taurus, and the Trapezium, which exhibitthe best combination of number statistics, completeness,and dynamic range in mass among the young populationsstudied to date. These IMFs are shown in Fig. 1. Becausethe same techniques and models were employed in convert-ing from data to masses for each population, one can beconfident in the validity of any differences in these IMFs.For the Trapezium, we use the IMF derived through infrared(IR) luminosity function modeling by Muench et al. (2002)(see also Luhman et al., 2000; Hillenbrand and Carpen-ter,
View Full Document
Unlocking...