2011ApJ...733..111T

We have serendipitously detected a strong O VI-bearing Lyman-limit system (LLS) at z_abs= 0.3558 toward the quasi-stellar object (QSO) J1009+0713 (z_em= 0.456) in our survey of low-redshift galaxy halos with the Hubble Space Telescope's (HST) Cosmic Origins Spectrograph. Its total rest-frame equivalent width of W_r= 835±49 mÅ and column density of log N(O VI) = 15.0 are the highest for an intervening absorber yet detected in any low-redshift QSO sightline, with absorption spanning at least four major kinematic component groups over 400 km/s in its rest frame. HST/Wide Field Camera 3 images of the galaxy field show that the absorber is associated with two galaxies lying at 14 and 46 kpc from the QSO line of sight. The absorber is kinematically complex and there are no less than nine individual Mg II components spanning 200 km/s in our Keck/HIRES optical data. The bulk of the absorbing gas traced by H I resides in two strong, blended component groups that possess a total log N(H I) ≃ 18-18.8, but most of the O VI is associated with two outlying components with log N(H I) = 14.8 and 16.5. The ion ratios and column densities of C, N, O, Mg, Si, S, and Fe, except the O VI, can be accommodated into a simple photoionization model in which diffuse, low-metallicity halo gas is exposed to a photoionizing field from stars in the nearby galaxies that propagates into the halo at 10% efficiency. In this model, the clouds have neutral fractions of ∼1%-10% and thus total hydrogen column densities of log N(H) ≃ 19.5. Direct measurement of the gas metallicity is precluded by saturation of the main components of H I, but we constrain the metallicity firmly within the range 0.1-1 Z_☉, and photoionization modeling indirectly indicates a subsolar metallicity of 0.05-0.5 Z_☉. This highly ionized, multiphase, possibly low-metallicity halo gas resembles gas with similar properties in the Milky Way halo and other low-redshift LLS, suggesting that at least some other galaxies have their star formation fueled by metal-poor gas accreting from the intergalactic medium and ionized by the stars in the host galaxy. As observed in the Milky Way high-velocity clouds, the strong detected O VI is not consistent with the photoionization scenario but is consistent with general picture in which O VI arises in interface material surrounding the photoionized clouds or in a hotter, diffuse component of the halo. The appearance of strong O VI and nine Mg II components in this system, and our review of similar systems in the literature, offer some support to this "interface" picture of high-velocity O VI: the total strength of the O VI shows a positive correlation with the number of detected components in the low-ionization gas.