[ABRIDGED] We have carried out a systematic search for close supermassive black hole binaries among z < 0.7 SDSS quasars Such binaries are predicted by models of supermassive black hole and host galaxy co-evolution, therefore their census and populat
ion properties constitute an important test of these models. We used an automatic technique based on spectroscopic principal component analysis to search for broad H-beta lines that are displaced from the rest-frame of the quasar by more than 1,000 km/s This method can also yield candidates for rapidly recoiling black holes. Our search yielded 88 candidates, several of which were previously identified and discussed in the literature. The widths of the broad H-beta lines are typical among quasars but the shifts are extreme. We found a correlation between the peak offset and skewness of the broad H-beta profiles, which suggests that the profiles we have selected share a common physical explanation. The general properties of the narrow emission lines are typical of quasars. We carried out followup spectroscopic observations of 68 objects to search for changes in the peak velocities of the H-beta lines (the time interval in the observers frame between the original and new observations is 1-10 yr). We measured significant changes in 14 objects, with resulting accelerations between -120 and +120 km/s/yr. We emphasize that interpretation of the offset broad emission lines as signatures of supermassive binaries is subject to many significant caveats. Many more followup observations over a long temporal baseline are needed to characterize the variability pattern of the broad lines and test that this pattern is indeed consistent with orbital motion. The possibility that some of the objects in this sample are rapidly recoiling black holes remains open as the available data do not provide strong constraints for this scenario.
The Karhunen-Loeve (KL) transform can compactly represent the information contained in large, complex datasets, cleanly eliminating noise from the data and identifying elements of the dataset with extreme or inconsistent characteristics. We develop t
echniques to apply the KL transform to the 4000-5700A region of 9,800 QSO spectra with z < 0.619 from the SDSS archive. Up to 200 eigenspectra are needed to fully reconstruct the spectra in this sample to the limit of their signal/noise. We propose a simple formula for selecting the optimum number of eigenspectra to use to reconstruct any given spectrum, based on the signal/noise of the spectrum, but validated by formal cross-validation tests. We show that such reconstructions can boost the effective signal/noise of the observations by a factor of 6 as well as fill in gaps in the data. The improved signal/noise of the resulting set will allow for better measurement and analysis of these spectra. The distribution of the QSO spectra within the eigenspace identifies regions of enhanced density of interesting subclasses, such as Narrow Line Seyfert 1s (NLS1s). The weightings, as well as the inability of the eigenspectra to fit some of the objects, also identifies outliers, which may be objects that are not valid members of the sample or objects with rare or unique properties. We identify 48 spectra from the sample that show no broad emission lines, 21 objects with unusual [O III] emission line properties, and 9 objects with peculiar H-beta emission line profiles. We also use this technique to identify a binary supermassive black hole candidate. We provide the eigenspectra and the reconstructed spectra of the QSO sample.
We present HST WFPC2/PC images and KPNO 4-m longslit spectroscopy of the QSO SDSS J153636.22+044127.0, which we advanced as a candidate binary supermassive black hole. The images reveal a close companion coincident with the radio source identified by
Wrobel & Laor (2009). It appears to be consistent with a M_g ~ -21.4 elliptical galaxy, if it is at the QSO redshift. The spectroscopy, however, shows no spatial offset of the red or blue Balmer line subcomponents. The companion is thus not the source of either the red or blue broad line systems; SDSS J153636.22+044127.0 cannot be explained as a chance superposition of objects, or as an ejected black hole. Over the Delta T=0.75 yr difference between the rest frame epochs of the present and SDSS spectroscopy, we find no velocity shift to within 40 km/s, nor any amplitude change in either broad line system. The lack of a shift can be admitted under the binary hypothesis if the implied radial velocity is a larger component of the full orbital velocity than was assumed in our earlier work. A strong test of the binary hypothesis requires yet longer temporal baselines. The lack of amplitude variations is unusual for the alternative explanation of this object as a double-peaked emitter; we further argue that SDSS J153636.22+044127.0 has unique spectral features that have no obvious analogue with other members of this class.
We identify SDSS J153636.22+044127.0, a QSO discovered in the Sloan Digital Sky Survey, as a promising candidate for a binary black hole system. This QSO has two broad-line emission systems separated by 3500 km/sec. The redder system at z=0.3889 also
has a typical set of narrow forbidden lines. The bluer system (z=0.3727) shows only broad Balmer lines and UV Fe II emission, making it highly unusual in its lack of narrow lines. A third system, which includes only unresolved absorption lines, is seen at a redshift, z=0.3878, intermediate between the two emission-line systems. While the observational signatures of binary nuclear black holes remain unclear, J1536+0441 is unique among all QSOs known in having two broad-line regions, indicative of two separate black holes presently accreting gas. The interpretation of this as a bound binary system of two black holes having masses of 10^8.9 and 10^7.3 solar masses, yields a separation of ~ 0.1 parsec and an orbital period of ~100 years. The separation implies that the two black holes are orbiting within a single narrow-line region, consistent with the characteristics of the spectrum. This object was identified as an extreme outlier of a Karhunen-Loeve Transform of 17,500 z < 0.7 QSO spectra from the SDSS. The probability of the spectrum resulting from a chance superposition of two QSOs with similar redshifts is estimated at 2X10^-7, leading to the expectation of 0.003 such objects in the sample studied; however, even in this case, the spectrum of the lower redshift QSO remains highly unusual.
