Next Contents Previous


We now turn to properties of the individual absorbers. Accurate spectra have been obtained at the MMT and CTIO 4-m in order to (a) test the damping hypothesis for the Ly-alpha profile of each `disk candidate', (b) confirm the redshift with the wavelengths of narrow metal lines, and (c) derive physical parameters for the absorbing gas. A full description of the observations and analysis is given in Turnshek et al. (1989) and Wolfe et al. (1988). The essential results are summarized in what follows.

All of the spectra blueward of Ly-alpha emission have been obtained with the MMT at a resolution of Delta lambda = 1 - 2 Å. Figure 1 is a sample of spectra in which redshifts determined from the metal lines have been used to shift the wavelength scale to the restframe of each damped Ly-alpha absorber. In QSOs with two damped systems, a restframe spectrum is shown for both redshifts. In every case Ly-alpha takes on the characteristic shape of a profile with damping wings. Quantitative tests show that most of the damped fits to the observed profiles are quite good. The wide range in W(Ly-alpha) illustrated in Figure 1 indicates that the damped systems exhibit wide variations in N(HI). Figure 1 also shows that in cases where the metal lines are predicted to occur redward of Ly-alpha emission, CII lambda1334 is detected. When low-ion metal lines such as CII are predicted to occur blueward of Ly-alpha emission, confusion with the Ly-alpha forest lines is inevitable. To determine redshifts in those cases, we obtained the FeII lambdalambda 2367-2600 multiplets and the MgII doublet with red spectra acquired with the CTIO 4-m.

Figure 1

Figure 1. Restframe Ly-alpha forest spectra of some sample QSOs with damped Ly-alpha absorption plotted on the same scale. Damped Ly-alpha is labeled with an arrow.

Comparison between the metal-line and Ly-alpha profiles on a velocity scale is particularly valuable, as it provides a sensitive test of the damping hypothesis, even in cases where the signal-to-noise ratio of the data is not optimum. In every case the metal-line profiles resemble the example in Figure 2, in which a narrow, unresolved (Deltav approx 150 km s-1) metal line is located at the center of a symmetric Ly-alpha feature which appears to have Deltav geq 3000 km s-1. The broadening mechanisms for the two transitions are obviously different. The CII line is broadened by Doppler motions, while quantum mechanical effects dominate the Ly-alpha profile. This is in contrast to CIV-selected systems with wide Ly-alpha lines. In those cases the metal lines break up into multiple narrow components (a) which span velocity intervals not much smaller than that covered by Ly-alpha , and (b) which are not symetrically displaced with respect to the center of Ly-alpha (Bechtold, Green and York 1987; Bechtold 1987). In this instance both Ly-alpha and the metal-lines are Doppler broadened, with Ly-alpha assuming a larger width due to an increase in saturation.

Figure 2

Figure 2. Comparison between z = 2.8 and damped Ly-alpha and CII profiles for Q1337+113.

The N(HI) inferred from the fits of Voigt profiles are better determined than in most absorption systems, because the fits are independent of velocity dispersion. The errors are dominated by inaccuracies in setting the continuum, and are unlikely to exceed 50% . Above 2 x 1020 cm-2, where the distribution of N(HI) is complete, we find that the mean column density, < N(HI) > = 1021 cm-2. While this is typical for the HI disks of spirals at R leq RHo, it greatly exceeds the upper limit of 1018 cm-2 set at R geq 2.5 x RHo, for a complete sample of spiral galaxies (Briggs et al. 1980). If the spirals have the same comoving number density as the damped systems, the latter are greater than the former in HI content as well as size.

Because the damped Ly-alpha systems are selected primarily on the basis of their HI properties, the resulting distribution of metal-line properties is unbiased with respect to metal-line selection criteria. Comparison between the detected W(CIV) and W(CII), for example, should provide a fair sample of the actual metal-line properties of the damped systems. We find that for the vast majority of damped systems, W(CII) geq W(CIV). This distinguishes the damped systems from most CIV-selected clouds in which W(CIV) geq W(CII) (Wolfe 1983; Danly, Blades and Norman 1987). Rather the damped systems resemble the MgII-selected clouds in this respect (Lanzetta, Turnshek and Wolfe 1987). This is an interesting result in view of the recent identification of MgII absorption systems with the halos of galaxies at z leq 0.8 (Bergeron 1988). The large W(CII) that are found for some damped systems may then indicate the presence of a turbulent halo associated with the quiescent disk.

In principle the damped systems are ideal for abundance determinations. Unlike most QSO absorption systems the ionization state of the abundant elements is known; i.e., H0 / H approx 1, C+ / C approx 1, etc. However, a spectral resolution of Deltav approx 10 km s-1 is required to separate the quiescent disk component from the turbulent halo component. Failure to do this leads one to underestimate the ionic column densities deduced from lines that are invariably saturated. A detailed description of the dilemma is given in Briggs et al. (1983). t the same time we (Lanzetta, Turnshek and Wolfe) are attempting to search for heavy elements locked up in grains. A program of absolute spectrophotometry was initiated in order to compare reddening in QSOs located behind damped systems with a control sample that shows no sign of damped Ly-alpha absorption. The absence of observed reddening leads to a preliminary constraint on the dust-to-gas ratio given by (D/G)damped leq 0.5 x (D/G)Galaxy (see also Pei and Fall 1987). This limit is consistent with the strict upper limits placed on the presence of H2 in one of the damped systems in our sample (Black, Chaffee and Foltz 1987).

Next Contents Previous