Models of the quasar population. II - The effects of dust obscuration

Abstract

A model for dust obscuration is proposed to account for the apparent lack of high-redshift quasars which are assumed to exist, but to be obscured by dust in intervening galaxies. The onset of dust obscuration is quite sudden with optical depth τ is propotional to (1 + z)^2.5^. Properties of the hypothesized intervening dusty galaxies are constrained to produce a sky-covering factor similar to that found for damped Lyα or Mg II absorption lines in high-redshift quasar spectra. The model successfully reproduces the apparent magnitude and redshift distributions of observed quasar samples at bright, intermediate, and faint magnitudes. Because spotty absorption tends to remove objects from the sample rather than redden them, observable quasars with redshifts less than 3.5 are reddened by only small amounts in conformity with observations. Hence, dust does not generally prevent quasars from being identified in surveys that select by color. Since X-rays penetrate the dust, the predicted 2-10 keV X-ray flux of quasars with an absolute blue magnitude brighter than -21.5 is larger than it would otherwise be and equals 82% of the observed background, with roughly one-half of the observed background arising from dust obscured quasars that appear to have an absolute magnitude fainter than -21.5. We would expect the Advanced X-ray Astrophysics Facility (AXAF) to detect a large number of these optically faint objects. The problem of ionization of the intergalactic medium noted by Shapiro to prevent large Lyα absorption troughs (Gunn-Peterson effect) is alleviated since the comoving UV photon density at redshifts greater than ~4 is larger by a factor of 15 than in equivalent models without dust. Quasars are also found to produce a substantial fraction of the 100 MeV γ-ray background. The predictions in the radio are consistent with the observations of QSOs in the Parkes radio survey. For a flux limit of ~0.5 Jy 2%-6% of sources should have optical counterparts fainter than B = 23 and most sources should be located at low redshift. If the dust model is correct, there should be many "blank" fields in fainter radio surveys that detect very low luminosity objects.