Galactic and Circumgalactic O VI and its Impact on the Cosmological Metal and Baryon Budgets at 2 < z <~ 3.5

We present the first results from our NASA Keck Observatory Database of Ionized Absorbers toward Quasars (KODIAQ) survey which aims to characterize the properties of the highly ionized gas of galaxies and their circumgalactic medium (CGM) at 2 < z < 4. We select absorbers optically thick at the Lyman limit (τ_LL > 1, log N_{H\,\scriptsize{I}} \textgreater 17.3) as probes of these galaxies and their CGM where both transitions of the O VI doublet have little contamination from the Lyα, β forests. We found 20 absorbers that satisfy these rules: 7 Lyman limit systems (LLSs), 8 super-LLSs (SLLSs) and 5 damped Lyα (DLAs). The O VI detection rate is 100% for the DLAs, 71% for the LLSs, and 63% for the SLLSs. When O VI is detected, log \langle N_{O\,\scriptsize{VI}} \rangle = 14.9 +/- 0.3, an average O VI column density substantially larger and with a smaller dispersion than found in blind O VI surveys at similar redshifts. Strong O VI absorption is therefore nearly ubiquitous in the CGM of z ~ 2-3 galaxies. The total velocity widths of the O VI profiles are also large (200 \le \Delta v_{O\,\scriptsize{VI}} \le 400 km s^-1). These properties are quite similar to those seen for O VI in low-z star-forming galaxies, and therefore we hypothesize that these strong CGM O VI absorbers (with τ_LL > 1) at 2 < z <~ 3.5 also probe outflows of star-forming galaxies. The LLSs and SLLSs with no O VI absorption have properties consistent with those seen in cosmological simulations tracing cold streams feeding galaxies. When the highly ionized (Si IV and O VI) gas is taken into account, we determine that the τ_LL > 1 absorbers could contain as much as 3%-14% of the cosmic baryon budget at z ~ 2-3, only second to the Lyα forest. We conservatively show that 5%-20% of the metals ever produced at z ~ 2-3 are in form of highly ionized metals ejected in the CGM of galaxies.

We dedicate this paper and the KODIAQ project to the memory and families of Wal Sargent and Arthur M. Wolfe. Without the vision and terrific efforts of these two scientists, this survey would not exist. Their careers have greatly inspired and influenced our own, and we hope that their work continues to flourish with this archival data set.