Largest Sample of Quasar Line Velocity Shifts: Statistical Analysis and Implications

We have found 25 quasars showing velocity-shift signatures in BALs, making up the largest sample of line velocity shifts so far. This is still a small sample, but we can get some preliminary statistical results based on it, as detailed below.

Distribution of Velocity Shift The velocity shifts of these BALs are mainly distributed in the range from -10000 to +10,000 km s−1 (fig2), but the maximum for the acceleration candidates can reach 2061 km s−1 (SDSS J083817.00+295526.5), and the maximum for the deceleration candidates can reach -2953 km s−1 (SDSS J145615.24+432217.3).

Velocity Shift as a Function of Timescale The velocity shift of BALs is positively correlated with the time interval (∆t) between each two epochs (fig3, 0.626619, 1.83532e-011). That is to say, as the ∆t increases, BAL tends to exhibit larger velocity-shift values. This is similar to the situation for EW variations of BAL that as the ∆t increases, BALs tend to exhibit larger EW variations.

Radial Velocity In our sample consisting of 25 quasars, compared to BALs without velocity shift, those with velocity shifts tend to show larger absolute values of radial velocities (fig4). KS test results indicate that the distributions of radial velocity show significant (P < -1E5) differences between these two sub-samples. The median of radial velocity for these two sub-samples are -18192.0 (for BALs without velocity shift) and -12289.5 km/s (for BALs with velocity shifts), respectively.

Fractional EW Variation In our sample of 25 quasars, compared to BALs without velocity shift, those with velocity shifts tend to show larger absolute values of △EW/ (fig5). KS test results indicate that the distributions of △EW/ show significant (P=0.02) differences between the two sub-samples. The △EW/ distribution of the shifted sub-sample (with the standard deviation σ=0.28 for the best-fitting Gaussian component) shows larger absolute values than the unshifted sub-sample (with the standard deviation σ=0.09 for the best-fitting Gaussian component).

The Correlated Fractional Variations of BALs and Quasar Continuum We plotted the △EW/ of C IV and Si IV BALs versus the fractional variation of the continuum in Figure 6. Significant moderate anticorrelations between the fractional flux variations of the continuum and △EW/ for both C IV and Si IV BALs are confirmed. Before this paper, anti-correlation between the fractional variations of the continuum and that of absorption line has also been found for several BAL (Lu et al. 2018; Lu & Lin 2018; Vivek 2019) and NAL (Lu et al. 2017; Chen et al. 2018a,b) quasar samples. Here, we confirm that such anti-correlation is also true for the quasar sample showing line velocity shifts, which serves as evidence for the following idea: the ionizing continuum fluctuation is the primary driving factor of BAL variability of the quasar sample with velocity shift.

Kinematic Shift As mentioned in the introduction, it is still difficult to determine the physical mechanism that causes the line velocity shift to date, and there are no existing models that can perfectly explain the observational cases of velocity shift of absorption line. In this paper, for the sample of 25 quasars with line velocity shift, we also cannot determine the physical mechanism of the velocity shift of the BALs only by the data of several SDSS observations. However, we infer that the line velocity shift shown in part of these 25 quasars may indicate actual line-of-sight accelerations of outflows owing to radiation pressures from the central source. The reasons for making this judgment are as follows. On the one hand, several quasars show variation characteristics (Section 3), which offer evidence for the IC driven BAL variation. On the other hand, the more important thing is that significant moderate anticorrelations between the fractional flux variations of the continuum and fractional EW variations for both C IV and Si IV BALs have been confirmed. The above variation characteristics convincingly indicate that at least the majority of BAL outflows in these 25 quasars are affected by the background radiation energy.

Conclusion This study conducted a comprehensive search for BALs that show velocity-shift signatures mainly from a BAL catalog (He et al. 2017) including 2005 SDSS quasars that have at least two-epoch observations. As a result, we have identified 35 absorption systems showing velocity-shift phenomenon, including 30 C IV BALs and five Si IV BALs, in 25 quasars, making up the largest sample of line velocity shifts so far. In addition to these velocity-shift BALs, we have also identified other CIVs and SiIV BALs presented in the spectra of the 25 quasars. Based on several measurements of these identified BALs, such as the equivalent width (EW), fractional EW variations (△EW/), velocity shift, and some properties of the spectra, we have mainly obtained the following statistical results:

1. As the ∆t increases, BAL tends to exhibit larger velocity-shift values.
2. Compared to BALs without velocity shift, those with velocity shifts tend to show larger absolute values of radial velocities.
3. Compared to BALs without velocity shift, those with velocity shifts tend to show larger absolute values of △EW/.
4. Significant moderate anticorrelations between the fractional flux variations of the continuum and △EW/ for both C IV and Si IV BALs are confirmed, which serves as evidence for the following idea: the ionizing continuum fluctuation is the primary driving factor of BAL variability of the quasar sample with velocity shift.
5. Based on the BAL variation characteristics and the significant moderate anticorrelations between the fractional flux variations of the continuum and fractional EW variations for both C IV and Si IV BALs, we guess that at least the majority of BAL outflows in these 25 quasars are affected by the background radiation energy.