Chandra X-ray Observations of Quasars with Velocity-Offset Broad Lines: Assessing the Binary Supermassive Black Hole Hypothesis

Abstract

During the final stages of a galaxy merger, dynamical friction acting on the supermassive black holes (SMBHs) in the post-merger remnant can lead to the formation of a gravitationally bound binary SMBH. In the event that at least one of these SMBHs is actively accreting, the system can appear phenomenologically as an active galactic nucleus (AGN) with a broad line region (BLR) kinematically offset from the host galaxy rest frame. Such velocity offsets have been interpreted as signatures of binary SMBHs, recoiling SMBHs, or BLR gas dynamics within a single-SMBH system. We present deep Chandra X-ray observations of five nearby (0.1 < z < 0.2) Sloan Digital Sky Survey quasars whose broad emission lines are Doppler-shifted relative to their host galaxies' systemic velocities, along with archival Chandra observations of 11 additional sources from the same sample. Using our Chandra data, we constrain SMBH masses with multiple independent techniques. We find systematic, method-dependent differences among black hole mass estimates, with masses inferred from the fundamental plane of black hole activity generally lower and single-epoch virial masses typically higher than those obtained using other methods. We also compare the X-ray photon indices and optical-to-X-ray spectral indices of our quasars to the broader quasar population. While we find no strong differences in optical-to-X-ray spectral indices, we do find systematically harder X-ray photon indices than typically observed in comparable quasars. These results constrain competing physical models but do not provide conclusive evidence for or against a binary SMBH origin of the velocity-offset BLRs.

Figures (7)

• Figure 1: Top: Bolometric luminosity as a function of redshift for the full eracleous+12 sample of velocity-offset quasars (black points) and the Chandra-observed sub-sample analyzed in this work (orange points). Bolometric luminosities are taken from shen+11. Bottom: Bolometric luminosity as a function of the narrow-line FWHM([O iii] $\lambda5007$) for the same two samples, with matching symbols and colors. Narrow-line widths are taken from Table 3 of eracleous+12. Together, these panels show that the Chandra sub-sample spans similar ranges in luminosity and narrow-line kinematics as the parent sample.
• Figure 2: Spectral fits in the 2$-$10 keV observer frame for the Chandra GO observations detailed in Table \ref{['table:cxo_obs']}. The model is shown as a solid black curve, with data points plotted using 1$\sigma$ error bars. Spectra are grouped for visualization (the grouping is indicated by "gcts" in each panel), while all fits are performed on unbinned data. Source names are given in the upper right corner of each spectrum.
• Figure 3: Top row: Observer-frame $2-10$ keV spectra of three representative sources exhibiting Fe K$\alpha$ emission. These examples illustrate, respectively, an unresolved line (left panel), a potentially moderately resolved line (middle panel), and a fully resolved line (right panel). Solid black curves denote model fits without an emission-line component, while red curves include an Fe K$\alpha$ line model component. Source names are shown in the upper right; "gcts" gives the minimum number of photons per energy bin (spectra are grouped only for visualization, as all fitting is performed on unbinned data). Source names and gcts values also apply to the corresponding plots in the bottom row. Bottom row: Solid black curves show model fits without an emission-line component, with the corresponding residuals plotted below. Vertical red lines mark the observed-frame energies corresponding to 6.4 keV in the rest frame.
• Figure 4: Distribution of X-ray photon indices for our sub-sample of quasars with velocity-offset broad emission lines (blue), compared to two control samples shown in orange and black. The orange histogram is derived from the SDSS/XMM-Newton Quasar Survey young+09, and the black histogram is derived from the ChaMP survey green+09. All histograms are normalized such that the total area under each distribution is unity, allowing a direct comparison of the shapes of the photon-index distributions independent of sample size.
• Figure 5: Optical--to--X-ray spectral slope, $\alpha_{\mathrm{ox}}$, as a function of rest-frame monochromatic UV luminosity, $L_{2500}$, for our velocity-offset quasar sample. The orange line shows the best-fit $\alpha_{\mathrm{ox}}$--$L_{2500}$ relation from just+07, while the shaded region denotes the $\pm1\sigma$ dispersion determined solely by the intrinsic scatter in the just+07 sample fit. Black points represent our sample, which shows no significant systematic deviation from the canonical relation, consistent with typical Type 1 quasar optical--to--X-ray colors.