3.1. Metals in the Ly Forest
The lack of associated metal lines was originally one of the defining characteristics of the Ly forest and was interpreted as evidence for a primordial origin of the clouds (Sargent et al. 1980). However, this picture was shown to be an oversimplification by the first observations - using the HIRES spectrograph on the Keck I telescope - with sufficient sensitivity to detect the weak C IV 1548, 1550 doublet associated with Ly clouds with column densities log N(H I) 14.5 (Cowie et al. 1995; Tytler et al. 1995). Typical column density ratios in these clouds are N(C IV) / N(H I) 10-2 - 10-3, indicative of a carbon abundance of about 1/300 of the solar value, or [C/H] - 2.5 in the usual notation, and with a scatter of perhaps a factor of ~ 3 (Davé et al. 1998).
The question of interest is `Where do these metals come from?'. Obviously from stars (we do not know of any other way to produce carbon!), but are these stars located in the vicinity of the Ly clouds observed - which after all are still at the high column density end of the distribution of values of N(H I) for intergalactic absorption - or are we seeing a more widespread level of metal enrichment, perhaps associated with the formation of the first stars which re-ionised the universe at z > 6 (Songaila & Cowie 2002)?
To answer this question we should like to search for metals in low density regions of the IGM, away from the overdensities where galaxies form. Observationally, this is a very difficult task - the associated absorption lines, if present at all, would be very weak indeed. Ellison et al. (1999, 2000) made some progress towards probing such regions using extremely long exposures with HIRES of two of the brightest known high-z QSOs, both gravitationally lensed: APM 08279+5255 and Q1422+231. The latter set of data in particular (Figures 13 and 14) is of exceptionally high quality, reaching a signal-to-noise ratio S/N 300 which translates to a limiting rest-frame equivalent width limit W0(3) 1 mÅ; this in turn corresponds to a sensitivity to C IV absorbers with column densities as low as N(C IV) 4 × 1011 cm-2.
And indeed C IV lines are found at these low levels (see Figure 16), showing that metals are present in the lowest column density Ly clouds probed, at least down to N(H I) = 1014 cm-2.
Figure 16. (Reproduced from Ellison et al. 2000). Examples of weak C IV lines identified in the spectrum of Q1422+231; most of these would have remained undetected in spectra of lower signal-to-noise ratios. Green (grey) lines show the profile fits used to deduce the column density of C IV. The weakest C IV systems are indicated with tick marks to guide the eye.
As can be seen from Figure 17, the number of weak C IV lines continues to rise as the signal-to-noise ratio of the spectra increases and any levelling off in the column density distribution presumably occurs at N(C IV) < 5 × 1011 cm-2. This limit is one order of magnitude more sensitive than those reached previously. In other words, we have yet to find any evidence in the Ly forest for regions of the IGM which are truly of primordial composition or have abundances as low as those of the most metal-poor stars in the Milky Way halo. These conclusions are further supported by the recent detection of O VI 1032, 1038 absorption in the Ly forest at z = 2 by Carswell, Schaye, & Kim (2002). In agreement with the results of Ellison et al. (2000), these authors found that most Ly forest clouds with N(H I) 1014 cm-2 have associated O VI absorption and that [O/H] is in the range -3 to -2. Weak O VI lines from regions of lower Ly optical depth have not yet been detected directly, but their presence is inferred from statistical considerations (Schaye et al. 2000).
Figure 17. (Reproduced from Ellison 2000). C IV column density distribution in Q1422+231 at <z> = 3.15; f(N) is the number of C IV systems per column density interval and per unit redshift path. The filled circles are the data; the straight line shows the best fitting power-law slope = 1.44, assuming the distribution to be of the form f(N)dN = BN- dN. The open circles show the values corrected for incompleteness at the low column density end; with these correction factors there is no indication of a turnover in the column density distribution down to the lowest values of N(C IV) reached up to now. Earlier indications of a turnover shown by the grey (Petitjean & Bergeron 1994) and dashed (Songaila 1997) curves are now seen to be due to the less sensitive detection limits of those studies, rather than to a real paucity of weak Ly lines.