2004MNRAS.347..985P

Artificial absorption spectra are used to test a variety of instrumental and physical effects on the pixel correlation technique for the detection of weak Ovi absorption. At HI optical depths ≲0.3-1, the apparent Ovi detections are spurious coincidences due to HI absorption at other redshifts. In this range, the apparent Ovi optical depth is independent of HI optical depth. At larger HI optical depths, the apparent Ovi optical depth and HI optical depth are correlated. Detailed modelling is required in order to interpret the significance of this relation. High-resolution spectra of four quasi-stellar objects together with a large suite of synthetic spectra are used to show that the detection of Ovi in individual spectra is only statistically significant for overdensities ≳5. These overdensities are larger than would be naively inferred from the onset of the correlation and a tight optical depth-density relation. The lower limit for the volume filling factor of regions which are enriched by Ovi is 4 per cent at 95 per cent confidence. This is no larger than the observed volume filling factor of the winds from Lyman break galaxies. Previous claims that the observed Ovi absorption extends to underdense regions and requires a universal metal enrichment with large volume filling factor, as may be expected from Population III star formation at very high redshift, appear not to be warranted.