7.2. Preliminary Results at z > 4

Quasars at the highest known redshifts potentially offer the strongest leverage on cosmological parameters, and the opportunity for studying high-z QSOs has grown rapidly in recent years with the discovery of an increasing number of sources at z > 4. Observational results for z > 4 QSOs were reviewed at this conference by Shields; in mid-1998 there are approximately 90 such objects reported in the published literature. Spectroscopic properties for major subsets of these sources have been reported previously by Schneider et al. (1991), Storrie-Lombardi et al. (1996), and Shields & Hamann (1997).

To first order, QSOs at high z appear very similar spectroscopically to sources at lower redshift. A composite spectrum derived from observations of 21 QSOs with z > 4 is shown in Figure 8. The apparent normality of the emission-line spectra of z > 4 quasars is actually a striking feature when one considers that the lookback time is greater than 90% of the age of the universe (for q₀ = 0.5, = 0). To the extent the emission spectra of quasars reflect the abundances of elements like C, N, and O, one might expect to see changes as a result of the chemical evolution expected in the host galaxies of quasars over cosmological time scales. In general, however, order-of-magnitude differences in abundances lead to much smaller variations in BLR line strengths in photoionization models, due to thermostatic feedback effects. An important exception may be present in the N V 1240 emission from QSOs, which may be sensitive to secondary nitrogen enrichment (Section 3.1), and does not display a BE. The N V feature remains relatively strong at z > 4, pointing to rapid early enrichment in the quasar nucleus environs (see Hamann & Ferland 1992, 1993 for details). As discussed by Shields & Hamann (1997), these sources also show evidence of elevated O I 1304 emission. This feature forms primarily via fluorescence pumped by H Ly line coincidence, and is consequently expected to scale with the O/H ratio, although factors other than metallicity may also affect its strength.

Figure 8. Composite spectrum obtained by averaging spectra of 21 QSOs at z > 4, observed with the Multiple Mirror Telescope (from Shields et al. 1998).

Quasars detected at z > 4 are almost invariably high-luminosity objects, and provide useful data points at high L for inclusion in the Baldwin diagrams. A summary of results to date is shown in Figure 9 for the C IV line, with data taken from Shields et al. (1998) for a sample of objects that are largely selected on the basis of optical colors (filled points), and from Schneider et al. (1991) for grism-selected objects (x). The dotted line represents the fit obtained to the BE by OPG for quasars at lower z. Schneider et al. noted a tendency for many of their sources to fall below the extrapolated BE in this diagram; with the addition of the Shields et al. points, the results appear consistent with the low-redshift fit, accompanied by the customary degree of scatter.

Figure 9. Baldwin diagram for quasars at z > 4. The filled points represent MMT observations from Shields et al. (1998), while the x's represent data from Schneider et al. (1991). The C IV BE reported by OPG, based on measurements at lower redshift, is shown by the dotted line.

Selection effects remain a point of concern in comparing the z > 4 findings with low-redshift results. The color-selected objects in particular are sensitive to inclusion of high-W objects that are preferentially found in magnitude-limited surveys, due to the added flux contributed by the emission line (typically Ly observed in the R bandpass ⁽⁸⁾; see Kennefick et al. 1995 for quantitative details). This bias may contribute to a weak tendency for the color-selected sources to exhibit larger equivalent widths than are found for the grism-selected objects in Figure 9 ⁽⁹⁾.

The existing results for quasars at the highest known redshifts nonetheless do not appear markedly different from their low-redshift counterparts, in terms of the BE as seen in the strong UV lines. We note, however, that this statement is ultimately sensitive to the choice of cosmology in computing L. (Figure 9 assumes H₀ = 50 km s^-1 and q₀ = 0.5.) This ambiguity is inherent in studying the BE and its possible evolution with samples in which L is strongly correlated with z; the resolution of these issues awaits construction of samples extending over a wide range of L at a given z. Researchers can also aid their colleagues by publishing actual measurements of continuum flux and redshift, rather than simply the derived quantity L, which depends on the specific choice of cosmology as well as measured z.

⁸ Note that the influence of the Ly line in this regard is exacerbated at large redshift by the (1 + z) scaling of observed W, and the diminished continuum blueward of the line caused by the Ly forest.

⁹ Interestingly, this pattern runs counter to the historical problems of selection bias, which favored large-W sources in grism-selected samples, in contrast with color selection techniques (Section 2.2). Objective selection algorithms for analysis of grism observations have largely removed this bias (e.g., Schmidt et al. 1995), while selection effects in high-redshift color-selected samples may be nonnegligible.