3.2. C IV at the Highest Redshifts

A level of metal enrichment of 10^-3 to 10^-2 of solar in regions of the IGM with N(H I)   10¹⁴ cm^-2 may still be understood in terms of supernova driven winds from galaxies. The work of Aguirre et al. (2001) shows that such outflows which, as we shall shortly see (Section 4.5) are observed directly in Lyman break galaxies at z = 3, may propagate out to radii of several hundred kpc before they stall. However, if O VI is also present in Ly forest clouds of lower column density, as claimed by Schaye et al. (2000), an origin in pregalactic stars at much earlier epochs is probably required (Madau, Ferrara, & Rees 2001).

In order to investigate this possibility, Songaila (2001) extended the search for intergalactic C IV to z = 5.5, taking advantage of the large number of QSOs with z_em > 5 discovered by the Sloan Digital Sky Survey. The surprising result, reproduced in Figure 18, is that there seems to be no discernable evolution in the integral of the column density distribution of C IV from z = 1.5 to z = 5.5. (The reality of a possible drop in (CIV) beyond z = 4.5 is questioned by Songaila because incompleteness effects have not been properly quantified in this difficult region of the optical spectrum, at _obs > 8500Å). This finding was unexpected and has not yet been properly assessed. The observed column density of C IV depends not only on the overall abundance of Carbon, but also on the shape and normalisation of the ionising background and on the densities associated with a given N(H I). Thus, we would have predicted large changes in (CIV) between z = 5.5 and 1.5 in response to the evolving density of ionising sources (QSOs) and the development of structure in the universe, even if the metallicity of the IGM had remained constant between these two epochs.

Figure 18. (Reproduced from Songaila 2001). Mass density in C IV (expressed as fraction of the closure density - see eqs. 2.2 and 2.3) as a function of redshift. The filled symbols are for C IV absorption systems in the range 12 log N(C IV) 15 (cm-2). The dot-dash lines show values of (CIV) computed assuming b h2 = 0.022, h = 0.65, a C IV ionisation fraction of 0.5, and metallicities Z = 0.0001 and 0.001 Z, respectively.

Whatever lies behind the apparent lack of redshift evolution of (CIV), it is clear that the IGM was enriched with the products of stellar nucleosynthesis from the earliest times we have been able to probe with QSO absorption line spectroscopy, only ~ 1 Gyr after the Big Bang. The measurements of (CIV) in Figure 18 suggest a metallicity Z_Ly 10^-3 Z; this is a lower limit because it assumes that the ionisation of the gas is such that the ratio C IV / C_tot is near its maximum value of about 0.5. This minimum metallicity can in turn can be used to infer a minimum number of hydrogen ionising photons (with energy h 13.6eV, corresponding to 912Å) in the IGM, because the progenitors of the supernovae which produce Oxygen, for example, are the same massive stars that emit most of the (stellar) ionising photons. Assuming a solar relative abundance scale (i.e. [C/O] = 0), Madau & Shull (1996) calculated that the energy of Lyman continuum photons emitted is 0.2% of the rest-mass energy of the heavy elements produced. ⁽¹⁾ From this it follows that

(3.6)

(Miralda-Escudé & Rees 1997), where Z is the metallicity (by mass) and m_p the mass of the proton. Since Z = 0.02 (Grevesse & Sauval 1998), if the Ly forest at z 5 had already been enriched to a metallicity Z_Ly 10^-3  Z, eq. (3.6) implies that by that epoch stars had emitted approximately three Lyman continuum (LyC) photons per baryon in the universe. Whether this photon production is sufficient to have reionised the IGM by these redshifts depends critically on the unknown escape fraction of LyC photons from the sites of star formation.

¹ This is a lower limit if [C/O] < 0, as is the case for low metallicity gas in nearby galaxies (e.g. Garnett et al. 1999). Back.