MG II absorption in the spectra of 103 QSOs - Implications for the evolution of gas in high-redshift galaxies

Abstract

We have obtained spectra of 103 low- and intermediate-redshift QSOs (roughly half of which are radio sources) with a resolution of 4-6 A in order to investigate the statistics of Mg II λλ2796, 2803 absorption systems over the redshift range 0.2 &lt;= Z_abs_ &lt;= 2.2. The QSOs were selected using criteria that are independent of their absorption properties, and in order to maximize the total redshift coverage of the survey. The signal-to-noise ratio of each spectrum was carefully monitored in order to attain a uniform minimum 5 σ equivalent width threshold of W_0_(λ2796) = 0.3 A for the detection of Mg II doublets over the entire observed redshift range. Comparable observations of most of the 103 QSOs had not previously been obtained. We generated lists of absorption lines for each QSO using criteria which ensure that a very high proportion of the lines are real. Standard methods were used to identify a total of 107 redshifts based on the Mg II doublet. This sample was augmented by the addition of Mg II redshifts determined in the course of the C IV survey of Sargent, Boksenberg &amp; Steidel. Uniform samples were selected with rest equivalent width thresholds W^min^_0_(λ 2796) of 0.3, 0.6, and 1.0 A, containing 111, 67, and 36 systems, respectively. The total redshift path covered by the survey is about 5 times larger than the largest previously existing homogeneous survey of similar sensitivity. New emission-line redshifts, based mostly on Mg II λ2800 and C III] λ1909, were determined for the 103 QSOs. A full discussion of the emission-line properties of the sample is deferred to a later paper. The principal results obtained from a study of the Mg II samples were the following: 1. The mean number density of Mg II absorbers per unit redshift range &lt;N(z)&gt; decreases from 0.97 +/- 0.10 for W^min^_0_ = 0.3 A to 0.27 +/- 0.05 for W^min^_0_ = 1.0 A. The results are in agreement with comparable previous work. 2. The observed equivalent width distribution n(W) could be fitted with either exponential or power-law analytic distributions, with parameters in line with those found in earlier studies. However, as suspected by earlier workers, there is a deficiency of weak Mg II lines. Consequently, we estimate that at least ~80% of the Mg II absorbers are accounted for in a sample sensitive to W^min^_0_ = 0.3 A. 3. Statistical tests show that the Mg II absorbers are uniformly distributed in both comoving coordinates and velocity relative to the QSO. In particular, there is no discernible tendency for Mg II absorption redshifts to congregate around Z_em_ as has been suspected for a restricted class of C IV absorption systems. 4. Our spectral resolution is insufficient for studying clustering on scales {DELTA}&lt;~500 km s^-1^, characteristic of the relative motions of clouds within galaxies. On larger scales, 600&lt;={DELTA}v&lt;=5000 km s^-1^, we find an excess in the two-point correlation function ξ({DELTA}v) which is significant at the 3.7 σ level, comparable in amplitude and formal significance to the excess found on the same large velocity scales for C IV absorbers. 5. The number density of Mg II absorbers, when expressed as a power law N(z) is proportional to (1 + z)<SUP>gamma</SUP>, yields γ = 0.78 +/- 0.42 for the sample with W^min^_0_ = 0.3 A. This value is formally consistent with nonnevolving absorbers in any Friedmann cosmological model with 0 &lt;= q_0_ &lt;= 0.5. As expected on the basis of photoionization models, both the slope and the number density of Mg II absorbers are consistent with the values found for the H I Lyman limit absorption systems. However, the small minority of very strong Mg II systems (W_0_ &gt; 1.0 A) have γ = 2.24 +/- 0.76, indicating very strong cosmological evolution. There is a trend for γ to increase with increasing W^min^_0_; it results from a systematic change in the Mg II rest equivalent width distribution with redshift. W_o_(λ2796) is correlated with redshift at the 97% confidence level, and &lt;W_0_&gt; is ~ 40% larger at &lt;z&gt; = 1.8 than at &lt;z&gt; = 0.55. 6. We give the dependence of N(z) on {OMEGA}_0_ for the models with finite cosmological constant {LAMBDA} ("zero-curvature" models) which result from the inflationary scenario; γ(z, {OMEGA}_0_) is a very sensitive function of {OMEGA}_0_ (and hence {LAMBDA}) at low redshift. The present data do not provide a useful constraint on such models; more stringent constraints could result from observations of larger samples of Mg II absorbers near z = 0. 7. A sample of 91 unblended Mg II doublets yields a mean doublet ratio DR = W_0_(λ2796)/W_0_ (λ2803) = 1.30, with a dispersion σ(DR)=0.25. The value of DR is uncorrelated with redshift (unlike the case of the C IV absorbers), while there is a significant anticorrelation of DR with W_0_(λ2796). The distribution of DR indicates that the great majority of the Mg II absorbers are strongly saturated. 8. At most a few percent of the detected Mg II absorbers could have been responsible for amplifying the background QSO into our sample by gravitational lensing. 9. Of the 103 QSOs, 35 had been observed for C IV absorption in the homogenous survey of Sargent, Boksenberg, &amp; Steidel, thus permitting a systematic intercomparison of the properties of Mg II and C IV in the same absorption systems over the redshift range 1.2 &lt;= Z_abs_ &lt;= 2.2. Because of the different spectral resolutions of the two surveys, complexes of absorption found in C IV were treated as single systems. Of the 105 absorption systems in the resulting sample, 27(26%) were detected in both Mg II and C IV, one was detected in Mg II but not in C IV, and 77(73%) were only detected in C IV. Thus, with spectra of comparable sensitivity in the C IV and Mg II spectral regions, one detects more than 99% of all heavy-element absorption complexes by using C IV alone. Systems which show Mg II also show C II, and vice versa. For z &gt; 2.2 we find that for the small available C II sample, N(z) = 0.94 +/- 0.33 at(z) = 2.35, while the power-law index for C II absorption over the range 1.5 &lt;= z_abs_ &lt;= 3.7 is γ = -1.46+/-1.73. Thus C II-selected absorbers (and presumably Mg II absorbers) show a trend for N( z) to decrease with increasing z, which has already been observed for C IV absorbers. Incorporating results from previous work on the behavior of C IV and Lyman limit selected absorption systems, as well as the results of direct observations of Mg II absorbing galaxies at moderately low redshifts, we present a scenario for the evolution of halo gas in galaxies over the redshift range 4 &gt;= z &gt;= 0.