Hence, when applying the scaling relations to estimate the masses of accreting BHs, the absence of obscured type-2 quasars from the AGN samples may introduce significant biases into the inferred evolution of BH mass. Only hard X-rays HXs can directly probe the central engine by penetrating the obscuring medium, therefore providing complete and unbiased samples of AGNs Ueda et al.
In the past decade, many semi-analytical models have attempted to explain the properties and evolution of AGNs. An important feature of these models is the inclusion of AGN feedback, a physical mechanism necessary for reproducing the morphology and colour of massive galaxies, and also the steep cut-off of the galaxy LF at low redshift Bower et al. In this paper, we present a study of AGN evolution using semi-analytic modelling see Baugh for a review. Our aim is to provide a robust framework for understanding the downsizing of AGNs within a self-consistent galaxy formation model.
This paper is organized as follows. In Section 2 , we present the galaxy formation model upon which we build our AGN model. In Section 3 , we study the evolution of BH mass and explore the scaling relations predicted between the BH mass and the mass of the host galaxy and dark matter DM halo. In Section 4 , we present the essential ingredients of the AGN model and study the evolution of the physical properties which, together with the mass, provide a complete description of accreting BHs.
Finally, in Section 6 , we explore the downsizing of AGNs in our model. We use the galform semi-analytical galaxy formation code Cole et al. Apart from some minor updates in the modelling of the different physical processes, which are described throughout this section, the galaxy formation part and associated parameters of the model is kept as in Bower et al.
Our starting point is the Bower et al.
This model invokes AGN feedback to suppress the cooling of gas in DM haloes with quasi-static hot atmospheres and has been shown to reproduce many observables, such as galaxy colours, stellar masses and LFs remarkably well. The model adopts a BH growth recipe based on that introduced by Malbon et al.
To distinguish between these two accretion channels, we refer to the accretion triggered by disc instabilities or galaxy mergers as the starburst mode and the accretion from quasi-hydrostatic hot haloes as the hot-halo mode. These correspond to the quasar and radio modes, respectively, in the terminology used by Croton et al. Finally, mergers between BHs, which occur when the galaxies which host the BHs merge, redistribute BH mass and contribute to the buildup of the most massive BHs in the universe see Fanidakis et al.
In this analysis, we are primarily interested in the properties that describe the BHs. Specifically, we output the BH mass, M BH , the amount of gas accreted in a starburst or during the hot-halo mode, M acc , the time that has passed since the last burst of SF experienced by the host galaxy, and the BH spin discussed in more detail below. The time since the last burst is necessary to determine the beginning of the active phase of BH growth in the starburst mode. The mechanical jet energy depends strongly on the accretion mode, as it most likely depends on the vertical poloidal magnetic field component close to the SMBH horizon.
In the Fanidakis et al. The collapse by two orders of magnitude in the scaleheight of the flow during the transition from an ADAF to a thin disc results in a similar drop in radio power. In this model, during the active phase gas is accreted on to the BH through a series of randomly oriented accretion discs, whose mass is limited by their self-gravity King et al. A correction to the mass of gas that is accreted is applied in order to account for the fraction of gas that turns into radiation during the accretion process. Finally, we note that we do not take into account BH ejection via gravitational-wave recoils during galaxy mergers Merritt et al.
This omission is not expected to have a significant impact on the evolution of BH mass in our model Libeskind et al. With this extension to the Bower et al. For completeness, we now list the model parameters which have an influence over the growth of BH mass: The fiducial model in this paper is denoted by F11b. We adopt the values for the above parameters that were used in the Bower et al.
These changes resulted in a small change in the distribution of accretion rates in the hot-halo mode. In this analysis, we use the original parameter values in order to be consistent with the version of the Bower et al. By reverting to them from the values in F11a, there is a small tail of objects accreting in the hot-halo mode, for which the accretion rate is higher than that typically associated with accretion via a thick disc. Summary of the revised parameter values in the variants of the Bower et al.
Note that F11a uses a different model for quiescent SF in discs. A further difference between our fiducial model, F11b, and the F11a model is the use of an improved SF law. Following Lagos et al. The BR06 model has the attraction that it is more physical than the previous parametric SF law used in Bower et al. The BR06 law distinguishes between molecular and atomic hydrogen, with only the molecular hydrogen taking part in SF.
The fraction of hydrogen in molecular form depends on the pressure within the galactic disc, which in turn is derived from the mass of gas and stars and the radius of the disc; these quantities are predicted by galform. The BR06 SF law contains no free parameters once it has been calibrated against observations. As shown by Lagos et al. This is due to a much weaker dependence of the effective SF time-scale on redshift with the new SF law, compared with that displayed by the Bower et al.
This leads to the buildup of larger gas reservoirs in discs at high redshift, resulting in more SF in bursts. This is shown in the top panel of Fig. This has an impact on the buildup of BH mass, by changing the amount of mass brought in through the starburst mode. The enhancement of the burst mode is further demonstrated in the bottom panel of Fig. Both channels show a significant increase in SFR density in the F11b model.
Since the BHs in our model grow in bursts of SF, we expect this enhancement to have a significant impact on the evolution of BH mass. We distinguish between the different accretion channels: The F11b model also predicts different hot-halo mode accretion. We note that, as shown in Fig. Hence, secular processes in galaxies are responsible for building most of the BH mass, while galaxy mergers become an important channel only when they occur between galaxies of similar mass i.
This merely reflects the importance of the disc-instability channel in triggering starbursts and building the stars in galactic bulges in our model. The fraction of baryons locked in BHs as a function of redshift black lines for the F11b solid lines and F11a dashed lines models. The plot also shows the contribution of each accretion channel, namely disc instabilities orange , galaxy mergers green and quasi-hydrostatic halo accretion blue , to the total fraction of baryons in BHs. The change in the total fraction of baryons locked up in BHs seen in the F11b model affects mostly the most massive BHs.
This is illustrated in Fig. In addition, we show two more variants of the Bower et al. This is because the feedback mechanism is most efficient in the most massive haloes and therefore it affects the most massive BHs. Modest values of the f Edd parameter mean that less accretion energy is available for heating the gas in the halo. This should in principle allow BHs to grow to significant masses. However, the model suggests that by choosing a relatively high e. Hence, to enhance the growth of massive BHs, we need to decrease the jet efficiency by a significant factor and allow more accretion energy to heat the gas in the host halo see e.
Furthermore, by changing the value of the f BH parameter, we can extend the dynamical range of BH masses built in the model since a high value of f BH implies more gas is being deposited into the BHs.
This produces an overall shift of mass function along the BH mass axis. The properties of the BHMF will be explored more thoroughly in the next section. Finally, we note that the total SFR density history in Fig. This implies that properties such as stellar masses and galaxy luminosities are fairly insensitive to the choice of SF law.
Indeed, Lagos et al. Since we have changed the normalization of the SF law, we recheck that our model still reproduces the observed LF of local galaxies, as shown in Fig. Hence, we are confident that we are building an AGN model within a realistic galaxy formation model. In the following sections, we will describe the AGN model extensively and present its predictions for the evolution of BH mass, spin and AGN abundances. When running galform , the parameters relating to the galaxy formation part of the code including those listed in Table 1 are kept fixed. The only free parameters in our model are strictly related to the accreting BHs and do not affect the evolution of their host galaxies.
These parameters, which are tuned to achieve a good fit to the available observational data on AGN LFs, will be described throughout the text and finally summarized in Table 3 see Section 4. The LF of galaxies in the local Universe. The left-hand panel compares the predictions of the F11b solid line and F11a dashed line models for the b J -band LF with the observational determination from the 2dF galaxy redshift survey by Norberg et al.
Similarly, the right-hand panel shows the predictions of the models line style as before for the K -band LF compared to the observational determinations by Cole et al. The theoretical predictions from both models include dust extinction. Proportionality factor in the accretion time-scale in equation 2.
BH demographics have been the topic of many studies in the past decade mainly because of the tight correlations between the properties of BHs and their host stellar spheroids. These correlations take various forms, relating, for example, the mass of the BH to the mass of the galactic bulge the M BH — M Bulge relation: These remarkable and unexpected correlations suggest a natural link between the BH evolution and the formation history of galaxies. The manifestation of this link could be associated with AGN activity triggered during the buildup of BHs.
Based on these MFs, and others inferred by independent studies, the total BH mass density in the local Universe has been estimated to be in the range see Table 2 for a list of local BH mass density estimates. For higher masses, the MF decreases steeply with increasing mass. Our predictions are compared to the local MFs estimated by Marconi et al.
The observationally estimated MFs are taken from Marconi et al. We classify as ellipticals all the galaxies whose bulge luminosity contributes more than 60 per cent to their total b J -band luminosity Cole et al. The rest of the galaxies are classified as spiral or S0 galaxies. As expected, the BHs in the elliptical galaxies dominate the high-mass end of the MF. Hence, the most massive BHs inhabit the centres of the most massive systems in our simulations.
By contrast, spiral and S0 galaxies contribute only to the low-mass end of the MF. The growth of BHs in spiral and elliptical galaxies is driven in principle by different channels. In spiral galaxies, BHs grow mainly via accretion during the starburst mode, whereas the massive BHs in ellipticals grow via accretion during the hot-halo mode or BH—BH mergers. Evidently, the evolution of the BHMF is shaped by these growth channels.
The growth of these lower mass BHs is dominated by accretion during the starburst mode. In contrast, the abundance of less massive BHs is found to decline quickly as redshift increases. This evolutionary trend is an indication of the downsizing of BH mass, a topic that will be discussed extensively in Section 5. For the moment, we note that the fact that our model matches the observational constraints for the BH mass and also the LFs presented in Section 5 suggests that the models of the BHMF that indicate downsizing do not provide a unique interpretation of the observational data.
These BHs grow mainly during the hot-halo accretion mode or via mergers with other BHs see Fanidakis et al. Yet, the buildup of these BHs is not very efficient. This becomes clearer when we consider the evolution of the BH mass density. To gain more insight into the rapidly evolving population of accreting BHs, we consider the evolution of actively growing BHs, namely those BHs that accrete at relatively high rates greater than 1 per cent of their Eddington accretion rate, see Section 4.
The most striking characteristic of the MF is the dramatic change in shape and normalization with redshift. As illustrated in Fig. Its amplitude increases significantly with increasing redshift, suggesting that the space density of low-mass accreting BHs was higher at earlier epochs. This is because the high-mass end forms much later than the low-mass end in our model. Their results indicate that the fraction of active galaxies decreases rapidly for high BH masses.
The comparison is shown in Fig. Our predictions are in good agreement with both observational data sets. The model reproduces the turnover at suggested by the observations and the characteristic decline at higher BH masses, which is associated with the decline of accreting gas in massive galaxies. The decrease in the BHMF at lower masses in our model is due to the luminosity limit adopted to reproduce the selection applied to the observational samples.
The fraction of active galaxies in the top panel of Fig. This bump is related to the tail of BHs accreting during the hot-halo mode at rates higher than 1 per cent of the Eddington value. The BH duty cycle in our model strongly depends on the accretion time-scale, which is proportional to the dynamical time-scale of the host bulge see Section 4. The proportionality factor, f q , determines the normalization of the duty cycle and therefore the fraction of active BHs in a given mass bin. The value obtained from the best fit naturally reproduces also the duty cycle of BHs as shown in Fig.
At high redshifts, the accretion activity is dominated by 10 6 - BHs. This is mainly because more higher mass galactic systems are now in place and thus disc instabilities and galaxy mergers contribute to the growth of higher mass BHs compared to earlier epochs. Accretion on to these BHs leads to the fast buildup of the BH population, which will later be promoted via accretion during the hot-halo mode and BH—BH mergers to the most massive BHs in our model.
Eventually, the different physical processes that drive the evolution of BH mass result in a tight correlation between the BH and host galaxy mass, as shown in Fig. This is a consequence of the tight correlation between the accretion of gas and the SF that shapes the mass of the host bulges.
Even when the growth of BHs is dominated by other processes, which are only indirectly linked to the SF, the tight correlation remains. For example, the mass of the most massive BHs correlates with the mass of their host bulge because feedback energy regulates the cooling flows that supply gas for SF and therefore the growth of the bulge mass. The shaded areas indicate the 5—95 red and 25—75 blue percentile ranges of the distributions.
In a hierarchical universe, the most massive galactic bulges are usually found in the most massive DM haloes. Since the galaxies with the most massive bulges namely the passive ellipticals host the most massive BHs and those with less massive bulges namely the spiral and S0 galaxies host the low-mass BHs, it is natural to assume that there must be a similar hierarchy in the halo environments where these BHs can be found. Indeed, when plotting the relation between the BH mass and the mass of the host halo, we find a tight correlation.
Remarkably, the physical processes that shape the M BH — M Bulge relation in our model give rise to a well-defined link between these two quantities at all redshifts. The different slopes indicate the different efficiency with which BHs grow in haloes of different mass. Low-mass haloes are very efficient in growing BHs since in these environments the gas cooling is not suppressed.
The fast BH mass buildup slows down when hot gas in the host haloes enters the quasi-hydrostatic regime. In this case, AGN feedback suppresses the cooling flows that provide fresh cold gas to the galaxies and thus accretion through the starburst mode is reduced. In the quasi-hydrostatic regime, gas accretion during the hot-halo mode and mergers dominate the BH mass buildup.
The hot-halo mode is, however, characterized by low accretion rates see Section 4. Therefore, the BH mass buildup slows down, establishing haloes of as environments where BH growth is not very efficient. Given the efficiencies that characterize the two different regimes in Fig. This is in contrast with the common expectation that the brightest quasars should be found in the most massive haloes. The ability of an accretion disc to produce electromagnetic radiation is attributed to gravity, yet its radiative efficiency is controlled primarily by the properties of the gas.
In the next sections, we describe how we model the physics of accretion flows and present predictions for the most fundamental properties characterizing the accreting BH systems in our model. For such low accretion rates, the gas flow is unable to cool efficiently since radiative cooling does not balance the energy generated by viscosity. Thus, the viscous energy is trapped in the gas and ultimately advected into the hole.
ADAFs have a number of distinct properties, some of which will be essential for the analysis in later sections see Sections 5 and 6. The remainder of the viscously dissipated energy is stored in the gas as entropy, resulting in hot flows with almost virial temperatures. We note that, as shown by Ichimaru , the ions and the electrons in an ADAF are not thermally coupled and, thus, reach different temperatures.
However, the accretion is entirely due to dissipation via viscous forces rather than gravity. The bolometric luminosity of the flow in this case is equal to the luminosity emitted by the various cooling processes. Finally, in Table 3 , we summarize the most important parameters in the AGN model. In practice, only f q has a significant impact on the model predictions and can be adjusted to match the observed AGN LF.
Data are taken from Hopkins et al. In Fanidakis et al. In brief, the spins of BHs change during gas accretion whenever a starburst is triggered by disc instabilities or galaxy mergers in the starburst mode or during the cooling of gas from the hydrostatic halo in the hot-halo mode and mergers with other BHs. In this redshift range, BHs grow predominantly during disc instabilities see Fig. As shown by Fanidakis et al. Hence, as indicated also by the different percentiles in the plot shaded regions , the bulk of BHs acquire low spins.
In particular, the field-corrected X-ray-selected AGN density in clusters evolves at least 3 times more rapidly than that of the field population. Their space density was significantly higher at earlier epochs, an evolutionary trend which suggests that AGN activity in the past was much more intense. In the highest redshift intervals, the obscuration becomes more prominent affecting even the brightest sources. For a more comprehensible representation of the objects populating the LF, we refer the reader to Fig. The second mode is linked to spheroids with low SF and governed by a power-law distribution of low Eddington ratios with a universal slope at least in the range probed by observations. The logarithmic dependence of luminosity on the accretion rate has a strong impact on the shape of the LF, resulting in a very steep slope at the bright end. Our starting point is the Bower et al.
Also shown in each panel are the 10—90 red shaded areas and 25—65 blue shaded areas percentile ranges of the BH spin distribution. Eventually, at low redshifts, BH mergers give rise to a population of rapidly rotating BHs. The environmental dependence of BH spins is illustrated in Fig. In contrast, slowly rotating BHs are found preferentially in low-mass haloes with some scatter.
This is a consequence of the correlation between BH mass and spin in our model: These BHs are hosted by massive elliptical galaxies in our model. In a forthcoming paper, we will explore the clustering properties of rapidly rotating BHs and show quantitatively the environmental dependence of BH spin. The size of the spheres is proportional to the spin of the BH that the galaxy hosts. In contrast, the median spin of actively growing BHs in our model does not reveal the presence of rapidly spinning BHs. This is demonstrated in the lower panel of Fig.
Hence, these BHs dominate the sample and determine the overall trend of the median. For higher BH masses, the efficiency can reach significantly higher values. This is a manifestation of the fact that high-mass BHs have high spins and thus high accretion efficiencies when they accrete in the thin-disc regime. Hence, at all redshifts, BHs have very well determined accretion efficiencies. It is therefore only the accretion rate that regulates the luminosity output from an accreting BH. In this case, a plot similar to the one in the bottom panel of Fig.
Hence, BH spins and efficiencies can display different distributions depending on the AGN population being probed. Having explored the evolution of BH mass and spin, and calculated the accretion efficiencies for the BHs accreting in the thin-disc regime, we now investigate the disc luminosities of the accreting BHs predicted by the model.
We calculate the bolometric disc luminosity in the ADAF and thin-disc regime using the formulation described in Section 4. The plane is divided into the radiatively-inefficient regime of ADAFs shaded region and radiatively-efficient regime of thin discs. The denser shading denotes the region of the plane where the discontinuous transition from the ADAF to the thin-disc regime takes place.
The first important property that is clearly apparent in Fig. The bimodality is exhibited at nearly all redshifts with both modes being characterized by a lognormal distribution. The second peak is located in the thin-disc regime and in this mode we find the most luminous objects in our model. The objects populating the second mode are exclusively AGNs accreting in the starburst mode. Since the thin-disc regime is radiatively efficient, these objects are expected to dominate the LF of AGNs in all bands except the radio and perhaps the HXs.
The relative space density of the objects in each mode changes dramatically with redshift. This picture changes in the low-redshift universe. As redshift decreases, more haloes enter the quasi-hydrostatic cooling regime and thus more BHs start to accrete in the ADAF regime see also the evolution of the hot-halo accretion channel in Fig. In this mode, we also find AGNs powered by gas accretion during gas-poor disc instabilities and galaxy mergers.
The first mode is associated with galactic spheroids that are undergoing significant SF and are characterized by a lognormal distribution of relatively high Eddington ratios. The second mode is linked to spheroids with low SF and governed by a power-law distribution of low Eddington ratios with a universal slope at least in the range probed by observations.
Our model suggests a similar behaviour since active galaxies in the thin-disc regime are starbursts, whereas in the ADAF regime, we have mostly passive galaxies with insignificant or no ongoing SF. In a forthcoming paper, we will present the distribution of Eddington ratios for different galaxy populations, and BH masses, and explore the growth of BHs in these two regimes. Whether these objects represent the most luminous AGNs at some given redshift is not obvious. To unravel the relation between BH mass and accreted luminosity, we plot in Fig.
We show predictions both for objects accreting in the thin-disc lower panels and ADAF upper panels regimes. The different percentiles of the distribution are colour-coded according to the bar on the right-hand side of each row. The shaded regions represent the regime where the accretion becomes super-Eddington.
For brighter luminosities, we find a strong correlation between M BH and L bol which indicates that the most massive BHs must have higher accretion rates compared to the lower mass ones. However, these BHs are very rare: For higher redshifts, their space density declines, and as a consequence, the maximum disc luminosity produced in the ADAF regime is also reduced. The M BH — L bol correlation in this regime increases monotonically until the slope becomes significantly shallower near the highest luminosities achieved at a given redshift.
The nature of the break in the slope is determined by the AGN feedback prescription in our model. When massive haloes reach quasi-hydrostatic equilibrium, they become subject to AGN feedback that suppresses the cooling flows; some of the mass which would have been involved in the cooling flow is instead accreted on to the BH.
Therefore, these BHs are expected to live in gas-poor environments and when they accrete gas they usually do so via an ADAF disc. In this regime, the suppression of cooling flows forces the M BH — L bol correlation to evolve only along the L bol axis since accretion via a thin disc on to BHs becomes very rare. This is due to the fact that in a hierarchical universe accretion shifts to lower BH masses at higher redshifts Section 3.
In addition, the break in the slope at high luminosities becomes less prominent since fewer haloes are in quasi-hydrostatic equilibrium at high redshifts. This implies that the most luminous quasars are expected to be found in DM haloes and not in the most massive ones remember the M BH — M Halo relation in Fig. There is evidence that the fraction of obscured AGNs decreases with increasing X-ray luminosity Steffen et al.
The question of whether the fraction of obscured AGNs depends also on redshift is more uncertain.
If the obscuring medium is indeed gas and dust in the galaxy, then we would expect the fraction of obscured AGNs to be redshift-dependent since in principle galaxy properties evolve with redshift. In addition, a strongly evolving population of obscured AGNs are required by AGN population synthesis models to reproduce the properties of the X-ray background Comastri et al. The b J -band quasar LF in the redshift range 0.
Predictions are shown for the model described in this paper F11b; solid black lines and the F11a solid orange lines model. Netzer's text spans the intellectual range from basic theory to observational connections, better than any other treatment of AGN that I know of. This is especially important since the things we know with most assurance are observational, and the things we want to know are theoretical conclusions at various degrees of separation from the data; here one sees the whole chains of reasoning in rich detail.
The text also folds in a range of recent developments, from the connections between accretion disks and outflows, through the rich statistics from recent sky surveys, to the unfolding links between AGN and their host galaxies as a normal part of galaxy evolution Keel, University of Alabama ' It is authoritative and complete The text also folds in a range of recent developments, from the connections between accretion disks and outflows, through the rich statistics from recent sky surveys, to the unfolding links between AGN and their host galaxies as a normal part of galaxy evolution.
This is a working reference, with useful tables and literature citations. Overige kenmerken Extra groot lettertype Nee. Reviews Schrijf een review. In winkelwagen Op verlanglijstje. Gratis verzending 30 dagen bedenktijd en gratis retourneren Ophalen bij een bol. Krolik Active Galactic Nuclei 86, Springer Active Galactic Nuclei , Springer Central Activity in Galaxies , Springer Accretion Disks - New Aspects ,