12.7.4. HI in Quasars and in Their Spectra
The host galaxies of quasars (QSOs) are difficult to observe at most wavelengths because the line and continuum radiation from the underlying galaxy is usually overwhelmed by the central QSO itself. And, although the optical galactic envelopes of QSOs have been observed to a redshift of about 0.5, these envelopes are sufficiently small over such distances that it is difficult to choose whether they are better fit by a spheroidal r1/4 law or by an exponential disk. The optical "fuzz" surrounding the nearest QSOs does seem to be fainter than expected for typical luminous ellipticals and sometimes suggests a spiral, though deformed, morphology. Just as the characteristic two-horned 21-cm profile can be used to place Seyfert and starburst galaxies in the heart of an otherwise faint or even unseen spiral disk, the detection of 21-cm emission from QSOs can be used as further evidence that the host object is likely a spiral (Condon et al. 1985). The detection of HI indicates the presence of gas in the host. Symmetric HI profiles can be used to determine an accurate systemic redshift that can be compared with the optical emission line values to establish the kinematics and radiation transfer within the regions producing. the narrow and broad optical lines. Asymmetric 21-cm line profiles may indicate the presence of tidal disruption or other global disturbances that may contribute to the QSO activity or result from it. Several QSOs are now known to reside in galaxies containing on the order of 7 × 109 M of HI, a typical HI mass for a spiral galaxy but much larger than that expected for an elliptical.
The optical spectra of QSOs frequently contain very narrow absorption lines nearly always lying at lower redshift zabs than the emission redshift zem, itself derived from the broad high-excitation emission lines also present. The location of the absorbing clouds has always been controversial: do they lie close to the quasar itself and move away from the quasar with relativistic velocities or are they located at some intermediate distance, corresponding to a strict cosmological recession-velocity interpretation of zabs ?
Under the assumption that clouds producing the narrow absorption lines are cold, the spectra of many quasars which are radio continuum sources have been searched for redshifted 21-cm absorption. Detectable absorption is expected if a cloud of NH > 1.0 × 1020 cm-2 and Ts < 1000 K intercepts the line of sight. Such clouds are prevalent in the disk of the Milky Way. Searches are made difficult by bandwidth and spectral resolution restrictions of current instrumentation, which limits the instantaneous frequency search range to a few tens of MHz, and by the presence of man-made interference - an increasingly severe handicap - at frequencies below 1 GHz. Nevertheless, absorption of redshifted HI has been detected in about ten QSOs.
The first detection of the 21-cm line in a quasar spectrum was made by Brown and Roberts (1973), who detected it in absorption at 839 MHz in 3C286, corresponding to a redshift of zabs = 0.69. The half-width of the line in the rest frame was observed to be 8.2 km s-1. VLBI observations reveal the existence of two narrow-velocity features of dispersion 1.6 and 3.0 km s-1, each located in front of a compact component in the continuum source. The inferred column densities are high, in excess of 8.5 × 1019(Ts / 100 K) cm-2, reminiscent of those seen in the disk of the Milky Way. After the discovery of the absorption at 21 cm, weak absorption features identified with transitions in MgI, MgII, and FeII were detected at optical wavelengths. In Milky Way clouds, the element Mg appears mainly in its singly ionized form; therefore, it appears sensible to search for MgII and HI arising from the same absorber. In such a search, Briggs and Wolfe (1983) found only two coincidences among eighteen redshifted clouds searched; most optical redshift systems do not contain highly opaque HI gas. Briggs and Woffe concluded that there is no correlation between 21-cm optical depth and such optical properties as MgII equivalent width, MgII doublet ratio, or MgI equivalent width. This lack of correlation between the radio and optical properties can be explained if the opacity is only occasionally high enough to produce HI absorption, and even in those cases, the HI clouds do not contribute significantly to the optical equivalent widths. Two gas phases are inferred: one showing only optical absorption, similar to Milky Way halo clouds, and one showing also 21-cm absorption characteristic of clouds in a galaxian disk similar to our own.
Perhaps the most intriguing 21-cm absorption line system is the one exhibited by the BL Lac object AO 0235+164, discovered by Roberts et al. (1976). This system is the first in which both radio and optical high-redshift absorption lines were measured at coincident redshifts. Figure 12.12 shows the zabs = 0.52 line, which occurs at 932 MHz. This spectrum, observed with a resolution of 1.6 km s-1, reveals five separate HI clouds in front of the source. Two gas phases are required to explain both the distinct narrow features and the overall absorption. The narrow features represent high-column-density clouds, each characterized by a velocity dispersion of about 3 km s-1 spread over a-range in mean velocity of 105 km s-1. In order to fit adequately the optical curve of growth, however, an additional component, optically thin at 21 cm and having a velocity dispersion of about 40 km s-1, must also be present.
If intervening galaxian disks are to be invoked to explain absorption lines in the spectra of quasars, the extrapolation to high redshift of a "normal" luminosity function and of commonly observed sizes of gaseous disks at low redshift cannot explain the observed number of absorption features. A different population of objects from those seen at low redshift or spiral disks that are much larger than their present-day counterparts appear to be necessary. Models of galaxy evolution that have been constructed to explain both the present metallicity and stellar luminosity function predict that in its early history, the typical column density of HI in the disk of the Milky Way was three to ten times higher than it is today; the velocity dispersion in the gas might have been as much as ten times larger than the current value of 10 km s-1. It is not yet clear whether galaxy disks existed at a redshift of 2 or more, or alternatively whether the growth of disks-was a slow process taking more than 1010 years after collapse to reach the Holmberg radius. As this field of research is still in a highly speculative stage, the 21-cm line appears to be a most promising probe of the universe at high redshifts.
This chapter was initially compiled in mid-1985 and revised in March 1987. The authors wish to thank M.S. Roberts, R. Sancisi and the editors for careful reading and useful suggestions on earlier versions of the manuscript.