ARlogo Annu. Rev. Astron. Astrophys. 1999. 37: 487-531
Copyright © 1999 by Annual Reviews. All rights reserved

Next Contents Previous

3.4. Narrow Absorption Line Results

In contrast to the BALs, intrinsic NALs might be the best abundance probes we have for QSO environments. Resolved measurements of NAL multiplets allow us to measure both the coverage fractions and true column densities (Section 3.2.2). The NALs also allow separate measurements of important lines that are often blended in BAL systems, such as NV lambda1239, 1243-Lyalpha, OVI lambda1032, 1038-Lybeta, and many others. We therefore have potentially many more constraints on both the ionization and abundances.

Early NAL studies did not have the quality of data needed to derive column densities and abundances, but several groups noted a tendency for larger NV/CIV line strength ratios in za approx ze systems compared with za << ze (Weymann et al. 1981, Hartquist & Snijders 1982, Bergeron & Kunth 1983, Morris et al. 1986, Bergeron & Boissé 1986). This trend is probably not caused simply by higher ionization in za approx ze absorbers, because recent studies show that za << ze systems typically have strong OVI lines and therefore considerable high-ionization gas; NV appears to be weak relative to both CIV and OVI at za << ze (Lu & Savage 1993, Bergeron et al. 1994, Burles & Tytler 1996, Kirkman & Tytler 1997, Savage et al. 1998). The lower NV/OVI and NV/CIV line ratios at za << ze could be caused by an underabundance of nitrogen (compared with solar ratios) in metal-poor intervening gas (Bergeron et al. 1994, Hamann et al. 1997d, Kirkman & Tytler 1997). This would be the classic abundance pattern involving secondary nitrogen (Vila-Costas & Edmunds 1993). Relatively higher N abundances and thus stronger NV absorption lines should occur naturally in metal-rich environments near QSOs (see Section 2.6 above and Sections 6 and 7 below).

The first explicit estimates of za approx ze metallicities were by Wampler et al. (1993), Möller et al. (1994), Petitjean et al. (1994), Savaglio et al. (1994) for QSOs at redshifts of ~ 2-4. These studies found that za approx ze systems often have Z gtapprox Zodot, which is at least an order of magnitude larger than the za << ze systems measured in the same data. Several of the metal-rich za approx ze systems have doublet ratios implying partial coverage and thus, very likely, an intrinsic origin (Wampler et al. 1993, Petitjean et al. 1994). The location of the other za approx ze absorbers is not known, but Petitjean et al. (1994) noted a marked change from [C/H] ltapprox -1 to [C/H] gtapprox 0 at a blueshift of ~ 15,000 km s-1 relative to the emission lines. If high abundances occur only in intrinsic systems, then these results suggest that most za approx ze NALs are intrinsic (see also Möller et al. 1994).

More recent studies support these findings. Petitjean & Srianand (1999) measured Z gtapprox Zodot and [N/C] > 0 in an intrinsic (partial-coverage) za approx ze absorber. For za approx ze systems of unknown origin, Savage et al. (1998) estimated roughly solar metallicities and Tripp et al. (1997) obtained [N/C] gtapprox 0.1 and, very conservatively, [C/H] gtapprox -0.8. (The lower limit on [C/H] for the latter system is -0.2 when more likely ionizing spectral shapes are used in the calculations.) Savaglio et al. (1997) revised the metallicities downward slightly from those in their 1994 paper to -1 < [C/H] < 0, based on better data. Those systems are of special interest because of their high redshift (za approx 4.1). Wampler et al. (1996) estimated Z ~ 2 Zodot (based on a tentative detection of OI lambda1303) for the only other za approx ze systems studied so far at z > 4.

Hamann (1997), Hamann et al. (1995, 1997b, 1997e, 1999b) used the analysis outlined in Section 3.2 to determine metallicities or establish lower limits for several za approx ze systems, including some mentioned above and some that are clearly intrinsic by the indicators in Section 3.1. The results generally confirm the previous estimates and show further that, even when there are no constraints on the ionization (for example, when only Lyalpha and CIV lines are measured), the column densities can still require Z gtapprox Zodot. A quick survey of those results suggests that bona fide intrinsic systems and most others with Z gtapprox Zodot have [N/C] gtapprox 0.0.

3.4.1. Uncertainties and Conclusions

Most of the NAL studies mentioned above would benefit from better data (higher signal-to-noise ratios and higher spectral resolutions) and more ionization constraints (wider wavelength coverage), but the frequent result for Z gtapprox Zodot is convincing. Unlike the BALs, there are no obvious systematic effects that might lead to higher abundance estimates for za approx ze systems compared with za << ze. The possibility of ionization-dependent coverage fractions presents an uncertainty for those systems with partial coverage, but we do not expect that to cause systematic overestimates of the metallicities (Section 3.2.2). We conclude that many za approx ze NALs and, more importantly, all of the confirmed intrinsic systems have Z gtapprox 0.5 Zodot and usually Z gtapprox Zodot. The upper limits on Z are uncertain. The largest estimate for a well-measured system is Z ~ 10 Zodot (Petitjean et al. 1994), but those data are also consistent with metallicities as low as solar because of ionization uncertainties (Hamann 1997). There are mixed and confusing reports in the literature regarding metal-to-metal abundance ratios, most notably N/C. In contrast to Franceschini & Gratton (1997), we find no tendency for subsolar N/C in za approx ze systems. In fact, there is the general trend for stronger NV absorption at za approx ze compared with za << ze systems, and the most reliable abundance data suggest solar or higher N/C ratios whenever Z gtapprox Zodot.

The only serious problem is in interpreting the abundance results for absorbers of unknown origin. High metallicities might correlate strongly with absorption near QSOs, but the metallicities cannot define the absorber's location. For example, Tripp et al. (1996) estimated Z gtapprox Zodot and [N/C] gtapprox 0 for a za approx ze system in which the lack of excited-state absorption in CII* lambda1336 (compared with the measured CII lambda1335) implies that the density is low, ltapprox 7 cm-3, and thus the distance from the QSO is large, gtapprox 300 kpc. [The relationship between density and distance follows from the flux requirements for photoionization (Section 2.5.1.)] Super-solar metallicities at these large distances are surprising. At gtapprox 300 kpc from the QSO, we might have expected very low intergalactic or halo-like abundances. The solution might be that the absorbing gas was enriched much nearer the QSO and then ejected (Tripp et al. 1996).

Unfortunately, the excited-state lines used for density and distance estimates are not generally available for za approx ze systems (because they have low ionization energies, e.g. CII* and SiII*). Of the six za approx ze absorbers known to be far (gtapprox 10) from QSOs based on these indicators, three of them clearly have za > ze and are probably not intrinsic for that reason (Williams et al. 1975, Williams & Weymann 1976, Sargent et al. 1982, Morris et al. 1986, Barlow et al. 1997). Only one has a metallicity estimate - the system with Z gtapprox Zodot at gtapprox 300 kpc distance studied by Tripp et al. (1996).

Next Contents Previous