In comparison to LRGs which are massive and quiescent, quasars powered by supermassive black holes also reside in massive halos of Mh ≳ 3 × 1012 M⊙ (Porciani et al., 2004, White et al., 2012, Shen et al., 2013b) but in an active phase that lasts ≈ 10−100 Myr (Martini, 2004). While the physical processes that drive the fueling and feedback of the central supermassive black holes are not well understood (e.g. Hopkins & Hernquist, 2009, Heckman & Best, 2014), quasar feedback is a critical ingredient in shutting down star formation in high-mass halos and producing red-and-dead elliptical galaxies in all theoretical frameworks (e.g. Benson et al., 2003, Croton et al., 2006). Indeed, most nearby early-type galaxies are found to harbor a supermassive black hole at the center (Kormendy & Richstone, 1995, Ho, 2008). Therefore, studying quasar hosts is an integral part of the effort to understand the growth of massive galaxies and observations of quasar host halos provide important clues for how the halo gas properties are shaped while the galaxies undergo an active quasar phase.
It has long been recognized that the intense radiation field from luminous quasars is sufficient to fully ionize their surrounding medium and to keep the IGM ionized over most of the cosmic history of the universe. Near the quasar redshifts, the incidence of Lyα absorption lines in the quasar spectra is reduced due to an enhanced radiation field from the quasars in the proximity zone (e.g. Bajtlik et al., 1988). The line-of-sight quasar proximity effect, which is also observed in different ionic transitions such as Mg II and C IV, offers a unique means of constraining both the metagalactic ionizing radiation and the extent of the ionizing bubble local to the quasar (Dall’Aglio et al., 2008, Wild et al., 2008). However, a reduced incidence of absorption features is not observed along the transverse direction from quasars (e.g. Fernandez-Soto et al., 1995, Schirber et al., 2004). In addition, there also appears to be an excess of cool clouds in quasar halos (Bowen et al., 2006, Hennawi & Prochaska, 2007, Farina et al., 2013, Farina et al., 2014, Prochaska et al., 2014).
The left panel of Figure 7 shows the mean covering fraction of cool halo gas κ as a function of halo mass for galaxies (closed triangles and squares) and quasars (closed circles) at different redshifts. For comparison, expectations from numerical simulations are also included (open star symbols). Observations of halos around galaxies and quasars at z < 2 are based on surveys of Mg II absorbers with Wr(Mg II) > 0.3 Å at projected distances less than the virial radius Rvir of the host dark matter halos. Observations of z ≈ 2 star-forming galaxies (Rudie et al., 2012) and quasars (Prochaska et al., 2014) are inferred from the observed incidence of optically-thick H I absorbers and C II absorption features, respectively. While halos around normal galaxies display a roughly constant cool gas covering fraction of κ ≈ 30% for the star-forming population at all redshifts studied (e.g. Chen et al., 2010a, Lovegrove & Simcoe, 2011, Rudie et al., 2012) and a suppressed cool gas content with κ ≲ 10% for the quiescent population at z ≈ 0.5 (e.g. Gauthier & Chen, 2011, Huang et al., 2016), there appears to be a surge of cool gas with κ ≈ 60% in active quasar halos (Prochaska et al., 2014, Farina et al., 2013, Farina et al., 2014, Johnson et al., 2015a) at both low and high redshifts.
 |
Figure 7. Elevated cool gas covering fraction in quasar host halos. Left: Mean gas covering fraction κ as a function of halo mass for galaxies and quasar hosts at different redshifts. The measurements for galaxies and quasars at z < 2 are based on surveys of Mg II absorbers with Wr(Mg II) > 0.3 Å and at d < Rvir, where Rvir represents the virial radius of the host dark matter halos. Measurements for z ≈ 2 star-forming galaxies (Rudie et al., 2012) and quasars (Prochaska et al., 2014) are inferred from the observed incidence of optically-thick H I absorbers and C II absorption observations, respectively. Closed triangles represent star-forming galaxies (Chen et al., 2010a, Lovegrove & Simcoe, 2011, Rudie et al., 2012, Werk et al., 2014), squares represent quiescent LRGs (Huang et al., 2016), and circles represent quasar hosts (Prochaska et al., 2014, Johnson et al., 2015b). Star symbols are simulation predictions for z ≈ 2 objects (Faucher-Giguère et al., 2016). Different colors indicate different redshift ranges of each dataset with red, orange, green, and blue corresponding to z ≈ 0.2, 0.5, 1.0, and 2.0, respectively. The mean gas covering fraction in quasar host halos is observed to be more than doubled from what is seen in star-forming galaxies at both low and high redshifts. Numerical simulations have attributed the elevated cool gas content in quasar hosts to either AGN feedback (Rahmati et al., 2015) or star formation feedback from neighboring satellite galaxies (Faucher-Giguère et al., 2016). Right: Mean gas covering fraction at d < 200 kpc for quasar hosts at different redshifts (top) and with different bolometric luminosity (bottom). Colors denote different redshift ranges as defined in the left panel. While the mean gas covering fraction in quasar hosts is found to evolve only weakly with redshift, a steep correlation is observed between cool gas covering fraction and quasar luminosity. The elevated cool halo gas covering fraction observed in quasar hosts appears to be driven primarily by the most luminous quasar population. (Figure credit: Sean Johnson) |
While state-of-the-art numerical simulations can reproduce the observed incidence of cool halo gas around star-forming galaxies and quasars at z ≈ 2, the underlying physical mechanisms are fundamentally different. The elevated cool gas content in quasar hosts are attributed to either AGN feedback (Rahmati et al., 2015) or star formation feedback from neighboring satellite galaxies (Faucher-Giguère et al., 2016).
The observed line-of-sight proximity effect, together with an absence of transverse proximity effect, indicates that the quasar radiation field is not isotropic and that cool clouds outside of the radiation zone can still survive in the host halos. Understanding whether and how the incidence of cool gas varies with quasar properties provides further insights into the origin of the observed excess of cool halo gas around quasars. Using a sample of 195 projected quasar pairs separated by less than 300 kpc in projected distance, it has been shown that the mean gas covering fraction (measured based on observations of Mg II absorption transitions) correlates strongly with the bolometric luminosity Lbol of the foreground quasar (Johnson et al., 2015a). Specifically, quasars that are ten times more luminous display on average 3−4 times higher cool gas covering fraction than low-luminosity quasars at d < 200 kpc in the halos (lower-right panel of Figure 7). The elevated cool halo gas covering fraction observed in quasar hosts appears to be driven primarily by the most luminous quasars in the sample.
The observed strong correlation between κ and Lbol is unlikely to be driven by an underlying mass dependence (c.f. Chen et al., 2010b), because the clustering amplitude of quasars remains roughly constant in different luminosity intervals (Shen et al., 2013b, Eftekharzadeh et al., 2015). The observed Lbol dependence in cool halo gas covering fraction therefore has profound implications for the connection between halo gas on the 100 kpc scale and quasar activities in the central parsec scale.
Possible explanations for the observed excess of cool gas along transverse sightlines include: (i) overdensity of galaxies in the quasar environment which is expected from the large clustering amplitude observed for luminous quasars, (ii) outflows from the quasar/nuclear starburst, and (iii) debris from galaxy interactions or mergers that trigger the luminous quasar phase. Of these scenarios, quasar outflows offer the most promising explanation but caveats remain. Significant contributions from correlated galaxies sharing the same large-scale overdensity peak can be ruled out based on the diminishing κ observed at d ≳ 200 kpc from quasars (Johnson et al., 2015a). Merger remnants are unlikely to explain the observed cool gas at d ∼ 100 kpc, given that the quasar lifetime is shorter than the dynamical time.
Spatially extended outflows have been detected in [O III] emission out to 10−30 kpc around luminous quasars at z ∼ 0.5 with outflow velocities as high as vout ≈ 1000 km s−1 (e.g. Greene et al., 2012, Liu et al., 2013, Liu et al., 2014). The observed spherical morphologies in these [O III] emitting nebulae suggest that outflows in these quasar hosts are not well-collimated. It is therefore possible that the absorbers detected at d ∼ 100 kpc along transverse sightlines in quasar halos originate in these extended outflows with densities too low to be detected in emission.
An outflow origin also provides a natural explanation for the observed strong correlation between κ and Lbol with more luminous quasars driving higher mass outflow rates (e.g. Carniani et al., 2015). This is at least qualitatively consistent with the extreme kinematics displayed in some ( ≈ 10%) quasar absorbers that spread over a velocity interval of |Δv| ≳ 1000 km s−1 (e.g. Johnson et al., 2015a). However, free-expanding outflows, traveling at vout = 1000 km s−1 at 15 kpc from the quasar, are expected to reach 100 kpc in ≈ 100 Myr and longer if the outflows decelerate due to the gravitational potential of the host halo. With a typical quasar lifetime of 10−100 Myr (e.g. Martini, 2004), the required outflow speed would need to exceed 1000 km s−1 in order for outflows near the peak of an active quasar phase to explain the observed excess of cool gas at d ∼ 100 kpc. With the energetics associated with fast-moving outflows, the gas is expected to be heated and highly ionized. If the cool clumps form within the hot outflows due to efficient cooling (e.g. Costa et al., 2015), then a strong correlation is expected between the incidence of cool gas and the presence of highly-ionized gas as probed by either [O III] emission (e.g. Greene et al., 2012) or O VI/C IV absorption (e.g. Grimes et al., 2009). Observations of highly-ionized gas associated with the low-ionization gas detected in quasar host halos will provide new insights into the origin of the κ vs. Lbol correlation.
Alternatively, the cool gas could originate in outflows from star-forming satellites as suggested by a recent simulation study (Faucher-Giguère et al., 2016). In this scenario, the distance required for the outflows to travel within the quasar lifetime would be ∼ 10 kpc, rather than ∼ 100 kpc, and the required outflow energetics would be less extreme. However, the implication would be a more quiescent satellite environment around low-luminosity quasars in order to explain a reduced κ. Comparing the star formation histories of satellites around luminous and low-luminosity quasars will provide a necessary test for this scenario.