@article{TurnerSchayeCrainetal.2016,
  author    = {Turner, Monica L. and Schaye, Joop and Crain, Robert A. and Theuns, Tom and Wendt, Martin},
  title     = {Observations of metals in the z approximate to 3.5 intergalactic medium and comparison to the EAGLE simulations},
  series = {Monthly notices of the Royal Astronomical Society},
  volume    = {462},
  journal   = {Monthly notices of the Royal Astronomical Society},
  publisher = {Oxford Univ. Press},
  address   = {Oxford},
  issn      = {0035-8711},
  doi       = {10.1093/mnras/stw1816},
  pages     = {2440 -- 2464},
  year      = {2016},
  abstract  = {We study the z approximate to 3.5 intergalactic medium (IGM) by comparing new, high-quality absorption spectra of eight QSOs with < z(QSO)> = 3.75, to virtual observations of the Evolution and Assembly of Galaxies and their Environments (EAGLE) cosmological hydrodynamical simulations. We employ the pixel optical depth method and uncover strong correlations between various combinations of H I, C III, C IV, Si III, Si IV, and O VI. We find good agreement between many of the simulated and observed correlations, including tau(O) (VI) (tau(H) (I)). However, the observed median optical depths for the tau(C) (IV) (tau(H) (I)) and tau(Si) (IV) (tau(H) (I)) relations are higher than those measured from the mock spectra. The discrepancy increases from up to approximate to 0.1 dex at tau(H) (I) = 1 to approximate to 1 dex at tau(H) (I) = 10(2), where we are likely probing dense regions at small galactocentric distances. As possible solutions, we invoke (a) models of ionizing radiation softened above 4 Ryd to account for delayed completion of He II reionization; (b) simulations run at higher resolution; (c) the inclusion of additional line broadening due to unresolved turbulence; and (d) increased elemental abundances; however, none of these factors can fully explain the observed differences. Enhanced photoionization of H I by local sources, which was not modelled, could offer a solution. However, the much better agreement with the observed O VI(H I) relation, which we find probes a hot and likely collisionally ionized gas phase, indicates that the simulations are not in tension with the hot phase of the IGM, and suggests that the simulated outflows may entrain insufficient cool gas.},
  language  = {en}
}