Quasar winds are the fastest outflows in the universe and they are observed as blue-shifted Broad Absorption Lines (BALs) in quasar spectra. Quasar winds are substantial part of the nuclear environment; fast outflows play a key role on galaxy feedback by evacuating gas and heat to the host galaxy (Di Matteo et al., 2005; Springel et al., 2005; King, 2010). Therefore, understanding the mechanisms behind these outflows would shed light on dynamics and evolution of super-massive black holes.

BAL troughs observed in quasar spectra present characteristic variations in their equivalent widths (EW), line profiles, and velocities (Barlow et al., 1992; Lundgren et al., 2007; Filiz Ak et al., 2012, 2013, 2014). The timescales of significant variations ranges between a few years to a few tens of hours (Capellupo et al., 2012; Filiz Ak et al., 2012, 2013, 2014; Grier et al., 2015).

In this study, we investigate BAL variations in multi-epoch spectroscopic observations of SDSS J141955.28+522741.4 (hereafter J1419). The Sloan Digital Sky Survey (SDSS) DR 12 Quasar Catalog lists MJD-PLATE-FiberID key parameters for 32 spectroscopic observations of J1419 and the catalog categorizes J1419 as a BAL quasar with z = 2.14 (Pâris et al., 2017). Frequent observations allow us to investigate significant rapid BAL variations and correlated variations of multiple BAL troughs.

The main driving mechanisms behind the BAL variations is largely debated in the literature. One scenario involves transverse motion of absorbing gas across the observer's line of sight producing changes in the coverage fraction (e.g., Rogerson et al., 2016). A second scenario considers ionization level changes of the outflowing gas (e.g., Filiz Ak et al., 2013, 2014). Other scenarios (e.g., intrinsic instabilities of an absorbing gas driving BAL variations) are usually found potential but problematic (Capellupo et al., 2012).

SDSS BOSS carried out spectroscopic observations of 297301 quasars using a 2.5 m dedicated telescope at Apache Point Observatory (Gunn et al., 2006) between 2009 and 2014 (Eisenstein et al., 2011; Dawson et al., 2013). The main aim of BOSS is to map the spatial distribution of luminous red galaxies and quasars to detect the characteristic scale imprinted by baryon acoustic oscillations in the early universe. Spectral wavelength coverage of BOSS is between 3,600 and 10,400 Å with a spectral resolution varying between 1,300 and 3,000 (Smee et al., 2013).

SDSS obtained 32 spectroscopic observations of J1419 between MJD 56397 and MJD 56837 with a time spread of 140 days in the quasar rest frame. We follow some simple steps to prepare the spectra: We correct the Galactic extinction using a Milky Way extinction model (Cardelli et al., 1989) for R_v = 3.1 and A_V values from Schlafly and Finkbeiner (2011). We fit the continuum with a power-law model that is intrinsically reddened using SMC-like reddening model from Pei (1992). We transform all the available spectra to the quasars rest frame using visually inspected redshift value of z = 2.14 (Pâris et al., 2017).

To detect BAL troughs, we follow classical BAL definition that requires absorption lines to have velocity widths > 2,000 km s⁻¹, and reach at least 10% under the continuum level (Weymann et al., 1991). Considering the variable nature of BAL troughs, we follow Filiz Ak et al. (2013) to determine BAL complexes using multiple-epoch observations.

We identify three individual C iv BAL troughs that are denoted as C_A, C_B, and C_C. Their minimum and maximum velocity limits (v_min and v_max, respectively) are as follows: −2,000 and −7,800 km s⁻¹ for C_A, −8,200 and −10,200 km s⁻¹ for C_B and −11,200 and −15,600 km s⁻¹ for C_C. We also find a Si iv BAL trough that have v_min and v_max velocities similar to that of C_A. Multi-epoch observations show that C_A is a BAL trough complex, rather than a single trough with at least two constituent absorption features (see Figure 1). Similarly, the detected Si iv BAL trough is likely to be a BAL complex.

Figure 1 shows emission lines and absorption regions for C iv and Si iv transitions in mean spectrum. The mean spectrum is calculated by averaging the 32 spectra for a given wavelength. Figure 1 shows the identified C iv BAL troughs C_A, C_B, and C_C, and Si iv BAL trough.

Traditionally, BAL variability has been assessed considering the time-dependent variations of EWs measured for the identified absorption features. Thus, we measure EW and uncertainties on EW using Equations 1 and 2 of Kaspi et al. (2002). In order to study time dependent variations on EW, we calculate ΔEW = EW₂−EW₁ where EW₂ is BAL trough EW measured in a latter epoch of the two consecutive spectra. The uncertainties on EW₁ and EW₂ are propagated to calculate uncertainty on ΔEW.

We investigate 32 epochs of spectrum for J1419 to assess characteristics of its BAL variations. We identify three individual C iv BAL troughs that one of them appear to have at least two constituent absorption features. In addition, we identify a Si iv BAL trough that lies in similar velocity ranges of the slowest C iv BAL trough. Studying time-dependent EW variations for these BAL troughs, we highlighted three main findings: (1) The strongest BAL trough of J1419 (i.e., C_A) show a rapid significant variation at timescales of ~31 h where EW variations are as strong as 5.1σ. (2) The faster component of C_A disappears and re-emerges in a short timescale. The BAL component starts weakening compared to the first epoch spectra and disappears at t = 126.7 days. Following observations show that the component regains its strength within 4 days. (3) The time-dependent EW variations of four BAL troughs identified in J1419 spectra show strong and significant correlations where Spearman rank correlation coefficients are larger than 80%.

The shortest timescale BAL variation is presented by Grier et al. (2015) showing that a significant (4.67σ) EW variation is detected for C iv BAL trough of SDSS J141007.74+541203.3 at timescales as short as 1.2 days. Given that our findings is consistent with that of Grier et al. (2015), BAL EW variations over timescales of a few 10 h is likely to be a common behavior. In order to assess this suggestion, a larger number of quasars with frequent spectroscopic observations should be investigated.

Time dependent EW variability of BAL troughs is largely investigated at the timescales of years (e.g., Barlow et al., 1992; Lundgren et al., 2007; Capellupo et al., 2012; Filiz Ak et al., 2012, 2013, 2014). So far, however, only a small number of disappearance events are recorded (Filiz Ak et al., 2012; McGraw et al., 2017). The number of quasars presenting BAL re-emergence is only a few (e.g., Lundgren et al., 2007; Filiz Ak et al., 2012; Rogerson et al., 2016). Our findings show that J1419 is the first example of BAL disappearance and re-emergence at timescales as short as ~4 days.

All the other BAL troughs present in J1419 spectra show weakening and strengthening while C_Ab disappears and re-emerges suggesting that the BAL variability is not due to bulk motion of the absorbers. Furthermore, event of re-emergence in less than 14 days support that bulk motion is not the likely scenario to explain BAL variations for J1419.

Coordinated EW variations of BAL troughs that have a large velocity separations in between suggest that cause of BAL variability should effect a large portion of the BAL region for a quasar. Therefore, we conclude that BAL variations is not due to intrinsic instabilities of an absorbing gas. Our finding favor a scenario in which a change in ionization state of the absorbing gas is likely to be dominant mechanism to drive the BAL variability.

For further analysis, we assess emission line variations and photometric variations in coordination with BAL variations. Therefore, physical constraints will be discussed on the light of current models.

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.