We report on the highly variable SiIV and CIV broad absorption lines in SDSS J113831.4+351725.2 across four observational epochs. Using the SiIV doublet components, we find that the blue component is usually saturated and non-black, with the ratio of
optical depths between the two components rarely being 2:1. This indicates that these absorbers do not fully cover the line-of-sight and thus a simple apparent optical depth model is insufficient when measuring the true opacity of the absorbers. Tests with inhomogeneous (power-law) and pure-partial coverage (step-function) models of the absorbing SiIV optical depth predict the most un-blended doublets component profiles equally well. However, when testing with Gaussian-fitted doublet components to all SiIV absorbers and averaging the total absorption predicted in each doublet, the upper limit of the power law index is mostly unconstrained. This leads us to favour pure partial coverage as a more accurate measure of the true optical depth than the inhomogeneous power law model. The pure-partial coverage model indicates no significant change in covering fraction across the epochs, with changes in the incident ionizing flux on the absorbing gas instead being favoured as the variability mechanism. This is supported by (a) the coordinated behaviour of the absorption troughs, (b) the behaviour of the continuum at the blue end of the spectrum and (c) the consistency of photoionization simulations of ionic column density dependencies on ionization parameter with the observed variations. Evidence from the simulations together with the CIV absorption profile indicates that the absorber lies outside the broad line region, though the precise distance and kinetic luminosity are not well constrained.
We present a two-epoch Sloan Digital Sky Survey and Gemini/GMOS+William Herschel Telescope/ISIS variability study of 50 broad absorption line quasars of redshift range 1.9 < z < 4.2, containing 38 Si IV and 59 C IV BALs and spanning rest-frame time i
ntervals of approximately 10 months to 3.7 years. We find that 35/50 quasars exhibit one or more variable BALs, with 58% of Si IV and 46% of C IV BALs showing variability across the entire sample. On average, Si IV BALs show larger fractional change in BAL pseudo equivalent width than C IV BALs, as referenced to an unabsorbed continuum+emission-line spectrum constructed using non-negative matrix factorisation. No correlation is found between BAL variability and quasar luminosity, suggesting that ionizing continuum changes do not play a significant role in BAL variability (assuming the gas is in photoionization equilibrium with the ionizing continuum). A subset of 14 quasars have one variable BAL from each of Si IV and C IV with significant overlap in velocity space and for which variations are in the same sense (strengthening or weakening) and which appear to be correlated (98% confidence). We find examples of both appearing and disappearing BALs in weaker/shallower lines with disappearance rates of 2.3% for C IV and 5.3% for Si IV, suggesting average lifetimes of 142 and 43 years respectively. We identify 5 objects in which the BAL is coincident with the broad emission-line, but appears to cover only the continuum source. Assuming a clumpy inhomogeneous absorber model and a typical size for the continuum source, we infer a maximum cloud radius of 10^13 to 10^14 cm, assuming Eddington limited accretion.
We carefully reconsider the problem of classifying broad absorption line quasars (BALQSOs) and derive a new, unbiased estimate of the intrinsic BALQSO fraction from the SDSS DR3 QSO catalogue. We first show that the distribution of objects selected b
y the so-called ``absorption index (AI) is clearly bimodal in log(AI), with only one mode corresponding to definite BALQSOs. The surprisingly high BALQSO fractions that have recently been inferred from AI-based samples are therefore likely to be overestimated. We then present two new approaches to the classification problem that are designed to be more robust than the AI, but also more complete than the traditional ``balnicity index (BI). Both approaches yield observed BALQSO fractions around 13.5 per cent, while a conservative third approach suggests an upper limit of 18.3 per cent. Finally, we discuss the selection biases that affect our observed BALQSO fraction. After correcting for these biases, we arrive at our final estimate of the intrinsic BALQSO fraction. This is f_BALQSO = 0.17 +/- 0.01 (stat) +/- 0.03 (sys), with an upper limit of f_BALQSO = 0.23. We conclude by pointing out that the bimodality of the log(AI) distribution may be evidence that the BAL-forming region has clearly delineated physical boundaries.
