The previous section demonstrates that, while most AGN obscuration is likely to occur in nuclear regions within the sphere of influence of the SMBH (Section 4.1), a significant fraction of the obscuration may originate on larger scales (Sections 4.2 and 4.3). Obscuration during discrete events such as starbursts or galaxy mergers point to a link between SMBH growth and the cosmological formation of galaxies and large-scale structures. In this section we explore three of the implications of obscured SMBH growth for observational cosmology: (1) The SMBH-galaxy evolutionary sequence, (2) Obscured SMBH growth in the early Universe, and (3) The “missing” AGN population and the radiative efficiency of SMBH accretion.
5.1. The evolutionary sequence and the SMBH-galaxy connection
As discussed in Section 1.2, connections between SMBHs and galaxies in their cosmic evolution have attracted a great deal of interest, motivated in part by observations of concurrent AGN and starbursts in well-studied local systems (e.g., Sanders & Mirabel, 1996, Farrah et al., 2003), statistical connections between AGN activity and star formation or stellar mass in extragalactic surveys (e.g., Chen et al., 2013, Azadi et al., 2015), and observed correlations between SMBH masses and host galaxy properties (velocity dispersion, stellar mass, etc.; McConnell & Ma 2013, Kormendy & Ho 2013, Graham 2016). The growth of SMBHs releases enormous amounts of energy in the form of radiation, outflows, and relativistic jets that can significantly influence the evolution of the host galaxies (e.g., Alexander & Hickox, 2012, Fabian, 2012). Many galaxy formation models require energy input from AGN to produce the observed population of quiescent galaxies (e.g., Bower et al., 2006, Dubois et al., 2016). Ultimately, there is likely to be a complex interplay between SMBH and galaxy growth, in which AGN activity follows or enhances the growth of stars in some cases, and shuts off or prevents new star formation in others (see Harrison, 2017, for a review).
An important role in many models of SMBH-galaxy co-evolution is played by obscured AGN. Some models posit rapid phases of galaxy and SMBH growth triggered by mergers, interactions, or violent instabilities that can also disrupt the gas content of the galaxy and shroud the AGN (e.g., Sanders et al., 1988, Hopkins et al., 2008). The majority of the SMBH and galaxy growth is predicted to occur in an early obscured phase (e.g., Blecha et al. 2017; Figure 14), followed by a “blowout” phase in which AGN feedback both limits SMBH growth and ejects gas from the galaxy potential, quenching the starburst and preventing further star formation (e.g., Ishibashi & Fabian, 2016). In this picture the period of obscured AGN activity represents the key phase for building up the mass of the SMBH, while the subsequent AGN luminosity limits the growth and sets the relationship between the SMBH and galaxy. The impact of the AGN on surrounding gas can be observed in outflows, heating, and turbulence in molecular, atomic, and ionized gas (e.g., Greene et al., 2011, Harrison et al., 2014, Feruglio et al., 2015).
Figure 14. An illustration of SMBH fueling and obscuration during a galaxy merger (Blecha et al., 2017). The left panel shows images of simulated galaxies in four different stages of a merger, with the time dependence of AGN Lbol, NH, and WISE W1–W2 color are shown in the right column. The coalescence of the galaxies and SMBHs produces a spike in SMBH accretion and Lbol, along with a redder W1–W2 color and a jump in NH that reaches approximately Compton-thick (NH ∼ 1024 cm−2) level. Figure from Blecha et al. (2017), courtesy of L. Blecha.
However, whether galaxy and SMBH evolution is primarily driven by discrete, dramatic phases of evolution remains unclear. Some studies have suggested that the relationships between SMBHs and galaxies progress slowly over cosmic time, with the fueling (and obscuration) of AGN being primarily a stochastic process (e.g., Cisternas et al., 2011, Mullaney et al., 2012), so that obscured AGN do not represent an especially important phase in SMBH growth or feedback. It may also be possible that the importance of obscured AGN activity depends strongly on the type of galaxy and its evolutionary history; since massive ellipticals and galaxy bulges have old, α-enhanced stellar populations that formed in rapid starbursts (e.g., Zhu, Blanton & Moustakas, 2010, McDermid et al., 2015), obscured AGN activity may be more important in the formation of these systems than in disk-dominated galaxies with more quiescent SF histories (e.g., Ishibashi & Fabian, 2017).
A key clue in uncovering the role of obscured AGN in the cosmological growth of SMBHs is the determination of whether obscuration is connected with processes in the nuclear torus (small enough to be decoupled from the broader galaxy formation) or on the scale of the galaxy. Further observations are required to determine the fractions of AGN that are obscured due to material on “torus” and “galaxy” scales (as discussed in Section 4.3), and to determine the sub-populations of galaxies for which obscured AGN may play a particularly important role.
5.2. The evolution of obscured SMBHs at high redshift
If AGN obscuration is connected to the gas content of the host galaxy, then we may expect obscuration to be enhanced at higher redshift, where the fraction of mass of galaxies in atomic or molecular gas is far higher than in the local Universe (e.g., Carilli & Walter, 2013). As discussed in Section 3, X-ray observations have shown hints that the obscured fraction may increase with redshift. For a galaxy with a high enough mass of gas, even large-scale (∼kpc) obscuring clouds can be heavily absorbing or even Compton-thick, so X-ray observations may miss an increasing fraction of the SMBH growth at higher redshifts. Powerful, mid-IR-bright obscured quasars at z ∼ 2 frequently show weak or absent X-ray detections even in deep observations (Stern et al., 2014, Del Moro et al., 2016), suggesting that the Compton-thick fraction for these AGN are ∼ 25–50%, comparable to local samples (see Section 3) but at higher AGN luminosities. If this high Compton-thick fraction is also present for lower luminosity AGN at high redshift, a substantial fraction of the SMBH growth at high redshift might not be captured by current X-ray observations (see Section 5.3 for the implications of this on the global radiative efficiency).
The question of “missing” obscured AGN is particularly interesting at high redshift (z > 3). Luminous quasars with MBH > 109 M⊙ are observed to emerge at z = 6–7 (e.g., Mortlock et al., 2011, Wu et al., 2015) and increase in space density to lower redshifts with a peak at z ∼ 2 (e.g., Kelly et al., 2010). The growth of lower-mass SMBHs that ultimately power these massive quasars can be probed with deep X-ray studies of lower-luminosity AGN (see Volonteri, 2010, for a review). Recent studies of the deepest Chandra fields reveal a steep drop-off in the X-ray selected AGN space density at z > 3, with the evolution strongest for soft X-ray luminosity < 1044 erg s−1 (Vito et al. 2018; see Figure 15). If the X-ray observations do probe the complete radiative output of AGN at these redshifts, then models of SMBH “seeds” may encounter problems with insufficient SMBH growth to produce the observed massive quasars at lower redshifts. One potential solution is if the AGN are heavily obscured, which would allow for the presence of many more growing SMBHs that lie below the Chandra detection (or stacking) thresholds (e.g., Novak, 2013, Vito et al., 2018).
Figure 15. Evolution in the AGN X-ray luminosity density at high z, as determined through Chandra Deep Field Observations (Vito et al., 2018). The rapid drop-off of the luminosity density (shown by the black points) suggests that SMBH growth falls with redshift faster than does star formation (shown by the gray filled region). This fast evolution rules out a number of SMBH evolution models (shown as blue lines); this tension may be resolved by the existence of a highly-obscured AGN population at these redshifts. Figure from Vito et al. (2018), courtesy of F. Vito. Observational and model results are from: Vito et al. (2016), Ranalli et al. (2016), Aird et al. (2015b), Georgakakis et al. (2015), Vito et al. (2014), Bouwens et al. (2015), Volonteri et al. (2016), Sijacki et al. (2015), Bonoli, Mayer & Callegari (2014), Shankar, Weinberg & Miralda-Escudé (2013), Volonteri (2010), Lodato & Natarajan (2006).
At the highest redshifts, obscured accretion is an essential component to some of the SMBH seed models themselves. In particular “direct collapse” models posit the growth of early SMBHs in gas-rich dark matter halos (e.g., Volonteri & Begelman, 2010, Mayer et al., 2010) and the accretion process in these models implies high covering factors with Compton-thick absorption. In these models, the earliest growth of SMBHs is necessarily heavily obscured, although some signatures of these direct-collapsing systems may be observable through reprocessed IR radiation (e.g., Natarajan et al., 2017).
5.3. Obscured AGN, the cosmic X-ray background, and the radiative efficiency of black hole accretion
The presence of a population of heavily obscured AGN has important consequences for the fundamental physics of SMBH accretion (in particular the radiative efficiency), and the corresponding origin of the cosmic background radiation. As first proposed by Soltan (1982), the total radiation emitted by SMBHs over cosmic time provides a powerful clue to the accretion process that produces the population of SMBHs observed at low redshift. The concept is elegantly simple: The total radiation density produced by SMBHs (UT) is equal to the product of the mass density of SMBHs in the local Universe (assumed to have been accumulated via accretion) ρSMBH and the radiative efficiency є. Including a factor of 1−є to account for the mass-energy lost as radiation, this relationship can be written as:
A number of early studies used the space density of optically-selected unobscured quasars to estimate є (e.g., Yu & Tremaine, 2002). These studies included no obscured AGN and so naturally produced on a lower limit on UT, or required an estimate of the fraction of radiation that was obscured, as well as an estimate of the optical bolometric correction (the scale factor kopt to convert from the optical radiation density Uopt to the total radiation density; UT = kopt Uopt). Subsequent studies attempted to directly account for the obscured sources through measurements of the cosmic X-ray background (CXB); sensitive X-ray observations have confirmed that the CXB is dominated by emission from individual AGN (e.g., Bauer et al. 2004, Hickox & Markevitch 2006, see Brandt & Alexander 2015 for a recent review) and dominate the CXB even to E > 10 keV (e.g., Aird et al., 2015a, Harrison et al., 2016). Successful CXB synthesis models universally require a population of obscured AGN to produce the observed peak in the spectrum at E ∼ 30 keV (Gilli, Comastri & Hasinger, 2007, Treister, Urry & Virani, 2009, Ballantyne et al., 2011, Akylas et al., 2012, Ueda et al., 2014, Aird et al., 2015b, see Figure 16). After correcting the total CXB radiation for absorption (either empirically, or through AGN synthesis models), and assuming an X-ray bolometric correction, it can be used to estimate є (e.g., Fabian & Iwasawa, 1999).
Figure 16. ynthesis model of the cosmic X-ray background (CXB; Aird et al. 2015b), showing the contributions to the CXB from unobscured AGN (blue dotted line) and obscured AGN with varying levels of NH. The high-energy peak of the CXB is dominated by obscured sources, with a significant contribution from Compton-thick AGN (thick red line). CXB synthesis models have been used to estimate the SMBH radiative efficiency є, but may not account for the presence of an extremely obscured population (with NH > 1025 cm−2) that does not contribute significantly to the CXB (Section 5.3). Figure from Aird et al. (2015b), courtesy of J. Aird.
In recent years there has been substantial progress in understanding the cosmic synthesis of SMBHs. Large obscured AGN populations have been discovered that could substantially increase UT (Section 3) and updates to the local SMBH mass density have come from new dynamical measurements of SMBH masses and re-assessments of relationships between SMBH masses and galaxy properties (e.g., Kormendy & Ho, 2013). These new scaling relations significantly increased the estimate of ρSMBH, by up to a factor of 5; for the same value of UT, this would lead to a corresponding decrease in є that would fall uncomfortably far below the theoretically expected value of є ≈ 0.1 (Kerr 1963, Shapiro & Teukolsky 1983; see Section 1).
However, in these analyses one fundamental uncertainty is the fraction of AGN that are so heavily obscured that they contribute little or nothing to the observed radiation fields used to compute UT and thus є (see e.g., Martínez-Sansigre & Taylor 2009, Novak 2013). The total obscured fraction and NH distribution is often inferred from local optical or X-ray studies (e.g., Burlon et al., 2011, Ricci et al., 2017b). However, in principle a substantial fraction of sources may be missed through extremely heavy absorption that is not accounted for in the selection function (as discussed in Section 2), and the NH distribution may evolve significantly with redshift. The “missing” AGN population could therefore contribute a substantially higher fraction of the total SMBH growth than is assumed based on local studies. Due to energy conservation, the radiation from such sources must ultimately emerge at far-IR and submm wavelengths as it is reprocessed into thermal emission from cold dust. A census of the AGN population obtained from fitting broad-band SEDs (including far-IR data from Herschel) suggests that the evolution of the luminosity function of AGN identified in the IR is comparable to that determined from other wavebands (Delvecchio et al., 2014). However, the signal from extremely heavily buried AGN may be challenging or impossible to distinguish from emission powered by star formation processes that completely dominate the IR background at > 10 µm (Shi et al., 2013). Comastri et al. (2015) showed that the total radiation output from accreting SMBHs could be increased by a factor of ∼2 in the form of extremely Compton-thick, X-ray faint AGN without violating constraints in the X-ray and IR backgrounds. This, in turn, would increase є by a similar factor, potentially resolving tension with theoretical expectations.