The evolution of ΩH iand the epoch of formation of damped Lyman α absorbers

We present a study of the evolution of the column density distribution, f(N, z), and total neutral hydrogen mass in high column density quasar absorbers using candidates from a recent high-redshift survey for damped Lyman α (DLA) and Lyman-limit system (LLS) absorbers. The observed number of LLS N(H i) >1.6 × 1017 atom cm−2 is used to constrain f(N, z) below the classical DLA definition of 2 × 1020 atom cm−2. The evolution of the number density of LLS is consistent with our previous work but steeper than previously published work of other authors. At z = 5, the number density of Lyman-limit systems per unit redshift is ∼5, implying that these systems are a major source of ultraviolet (UV) opacity in the high-redshift Universe. The joint LLS–DLA analysis shows unambiguously that f(N, z) deviates significantly from a single power law and that a Γ-law distribution of the form ƒ(N, z) = (f*/N*)(N/N*)−βexp(−N/N*) provides a better description of the observations. These results are used to determine the amount of neutral gas contained in DLAs and in systems with lower column density. Whilst in the redshift range 2–3.5, ∼90 per cent of the neutral H i mass is in DLAs, we find that at z > 3.5 this fraction drops to only 55 per cent and that the remaining ‘missing’ mass fraction of the neutral gas lies in sub-DLAs with N(H i) 1019–2 × 1020 atom cm−2. The characteristic column density, N*, changes from 1.6 × 1021 atom cm−2 at z 3.5, supporting a picture where at z > 3.5, we are directly observing the formation of high column density neutral hydrogen DLA systems from lower column density units. Moreover, since current metallicity studies of DLA systems focus on the higher column density systems they may be giving a biased or incomplete view of global galactic chemical evolution at z > 3. After correcting the observed mass in H i for the ‘missing’ neutral gas the comoving mass density now shows no evidence for a decrease above z = 2.