The discovery and characterization of methane-bearing brown dwarfs and the definition of the T spectral class

Abstract

I present the discovery of 18 T dwarfs, brown dwarfs exhibiting CH4 absorption analogous to Gliese 229B, identified in the Two Micron All Sky Survey. Follow-up spectroscopic observations reveal the presence of strong H2O and CH4 bands in these objects, as well as broadened Na I and K I absorption in the red optical; fine lines of K I, Cs I, and Rb I; and FeH absorption at 9896 ® Three objects are analyzed in detail; the widely-separated companion brown dwarf Gliese 570D, the coolest known brown dwarf with Teff = 810+/-45 K; the active T dwarf 2MASS 1237+6526, whose unique and steady H alpha emission may be the result of Roche-lobe overflow accretion from a closely-separated companion; and 2MASS 0559-1404, the brightest T dwarf currently known, which appears to be overluminous but unresolved in HST images. The variation in spectral features amongst these objects and those identified by the Sloan Digital Sky Survey have been used to derive a near-infrared spectral classification scheme, tied to the observed strengths of H2O and CH4 bands, color ratios, and K-band spectral morphology. The grid of subclasses segregates the currently known population into seven distinct groups, ranging from T1 V to T8 V. I show that the possible presence of CH4 in the L7 V DENIS 0205-1159AB argues for few subtypes between the latest known L dwarfs and the earliest T dwarfs. One peculiar object, 2MASS 0937+2931, has a highly suppressed K-band peak, likely due to increased H2 opacity in a high-gravity or low-metallicity atmosphere. Examination of absolute brightness and effective temperature across the L/T transition indicates rapid evolution of spectral features, possibly linked to heterogenous cloud coverage as condensibles rain out of the photosphere. Finally, I have used the T dwarf search samples to constrain the substellar field mass function. Through rigorous analysis of selection biases and Monte Carlo simulations, I show that my results are consistent with a power-law mass function scaling as 0.5 [less than] alpha [less than] 1.0, consistent with young stellar cluster surveys but significantly less than current estimates in the field.