Abstract

Aims. We present a study of the highly ionized gas (plasma) associated with damped Lyman-a (DLA) systems at z = 2.1-3.1. Methods. We search for O VI absorption and corresponding Si IV, CIV, and N V in a Very Large Telescope/Ultraviolet-Visible Echelle Spectrograph (VLT/UVES) sample of 35 DLA systems with data covering O VI at S/N > 10. We then use optical depth profile comparisons and ionization modelling to investigate the properties, phase structure, and origin of the plasma. Results. We report twelve DLAs (nine intervening and three at <5000 km s -1 from the QSO redshift) with detections of O VI absorption. There are no clear O VI non-detections, so the incidence of O VI in DLAs is between 34% (12/35) and 100%. Among these 12 DLAs, CIV and Si IV are seen whenever data is available, and N V is detected in 3 cases. Analysis of the line widths together with photoionization modelling suggests that two phases of DLA plasma exist: a hot, collisionally ionized phase (seen in broad O VI components), and a warm, photoionized phase (seen just in narrow C IV and Si IV components). The presence of inflows and/or outflows is indicated by individual O VI and C IV components displaced from the neutral gas (either blueshifted or redshifted) by up to 400 km s -1. We find tentative evidence (98% confidence) for correlations between the DLA metallicity (measured in the neutral gas) and high-ion column density, and between the DLA metallicity and high-ion line width, as would be expected if supernova-driven galactic outflows rather than accretion produced the high ions. Using conservative ionization corrections, we find lower limits to the total hydrogen column densities in the hot (O VI-bearing) and warm (C IV-bearing) phases in the range log N H II Hot > 19.5 to >21.1, and log N HII Warm 19.4 to > 20.9. On average, the hot and warm phases thus contain ?40% and ?20% of the baryonic mass of the neutral phase in DLAs, respectively. Conclusions. If the temperature in the O VI phase is ?10 6 K and so f O VI = O VI/O ? 0.2, the plasma can make a significant contribution to the metal budget at high redshift. Additional searches for O VI in Lyman Limit Systems (QSO absorbers with 17.0 < N H I < 20.3) will be necessary to determine the total quantity of baryons and metals hidden in hot halos at z ? 2. © ESO 2007.