2001ApJ...549..133G

We present an exhaustive statistical analysis of the associated (Δv_abs<5000 km.s^–1), high-ionization (C IV, N V, O VI) narrow absorption line (NAL) systems in a sample of 59 QSOs defined from the Hubble Space Telescope (HST) QSO Absorption Line Key Project. The goals of the research were twofold: (1) to determine the frequency of associated NALs at low redshift and in low luminosity QSOs, and (2) to address the question of what QSO properties either encourage or inhibit the presence of associated NAL gas. To that end, we have compiled the QSO rest-frame luminosities at 2500 Å, 5 GHz, and 2 keV, spectral indices at 2500 Å and 5 GHz, the Hβ emission-line FWHM, and the radio core fraction at observed 5 GHz. In addition, we have measured the C IV emission-line FWHM. We find 17 associated NALs (16 selected by C IV and one selected by O VI) toward 15 QSOs, of which ∼10 are statistically expected to be intrinsic. From a multivariate clustering analysis, we find that the QSOs group together (in parameter space) based primarily on radio luminosity, followed (in order of importance) by radio spectral index, C IV emission-line FWHM, and soft X-ray luminosity. We find that radio-loud QSOs that have compact radio morphologies, flat radio spectra [α(5 GHz)>-0.5], and mediocre C IV FWHM (≲6000 km.s^–1) do not have detectable associated NALs, down to W_r(C IV)=0.35 Å. We also find that broad absorption line (BAL) QSOs have an enhanced probability of hosting detectable NAL gas. In addition, we find that the velocity distribution of associated NALs is peaked around the emission redshifts rather than the systemic redshifts of the QSOs. Finally, we find only one strong NAL [W_r(C IV)≳1.5 Å] in our low-redshift sample. A comparison with previous higher redshift surveys reveals evolution in the number of strong NAL systems with redshift. We interpret these results in the context of an accretion disk model. We propose that NAL gas hugs the streamlines of the faster, denser, low-latitude wind, which has been associated with BALs. In the framework of this scenario, we can explain the observational clues as resulting from differences in orientation and wind properties, the latter presumably associated with the QSO radio properties.