The ultimate aim of studies of the host galaxies and environments of radio AGN is to understand how their AGN are triggered as part of the evolution of the general (non-active) galaxy populations. An holistic approach is required. A limited study of, say, the host galaxy morphologies alone, or the star formation properties alone, would not be sufficient. Fortunately, this is another area that has profited from observations with the new generation of ground- and space-based telescope facilities, which have provided a far more complete picture of the host galaxies than was possible in the recent past.
3.1. Overall morphologies, light profiles, kinematics and masses
It was already clear from the earliest photographic imaging studies that radio AGN are associated with early-type host galaxies. Matthews, Morgan & Schmidt (1964) showed that, above a certain radio power limit, the hosts of AGN are elliptical, D, or cD galaxies, whereas at lower radio powers spiral morphologies are more prevalent. However, there are exceptions. Table 4 shows a list of radio AGN for which extended disk or spiral components are clearly visible upon cursory inspection of optical images. Note that it is important to distinguish here between objects that show such morphologies, and those that merely harbour dust lanes that cross the nuclear regions of the otherwise elliptical host galaxies (de Koff et al., 2000).
There are several points to make about the radio AGN with clearly visible disk/spiral structures: (a) they are rare (< 5% of the objects in the 3CR and 2Jy samples); (b) they generally fall at the lower power end of the radio AGN population; (c) they are often associated with compact or double-double radio sources; and (d) their host galaxy stellar and black hole masses are comparable with those of the giant elliptical host galaxies of the majority of the radio AGN population (Mstellar > 1011 M⊙ and Mbh > 108 M⊙). The latter point is particularly interesting because it suggests that it is only the most massive disk galaxies that are capable of hosting radio AGN; it also further reinforces the link between black hole mass and radio loudness in the AGN population (see below).
|NGC612||0.0298||2 × 1025||NLRG||FRII||1|
|0313-192||0.0671||1 × 1024||NLRG||FRI/DD?||2,3|
|J0832+0532||0.099||1.5 × 1024||NLRG||FRII||4|
|3C223.1||0.107||5.4 × 1025||NLRG||FRII||5,6|
|J1159+5820||0.054||2.3 × 1024||WLRG||FRII/DD||7,4|
|3C236||0.1007||1.0 × 1026||NLRG||FRII/DD||8|
|3C293||0.0450||2 × 1025||WLRG||FRII?/DD||9,4|
|3C305||0.0416||1.2 × 1025||NLRG||CSS||10, 11|
|Speca||0.1378||7 × 1024||WLRG||FRII?/DD||12|
|J1649+26||0.055||1 × 1024||WLRG||FRII||14,4|
|PKS1814-637||0.0641||1.2 × 1026||NLRG||CSS||13|
|J23345-0449||0.0755||3 × 1024||WLRG||FRII/DD||15|
|Reference key: 1. Emonts et al. (2008); 2. Ledlow et al. (2001); 3. Keel et al. (2006); 4. Singh et al. (2015); 5. de Koff et al. (2000); 6. Madrid et al. (2006); 7. Kozieł-Wierzbowska et al. (2012); 8. O'Dea et al. (2001); 9. van Breugel et al. (1984); 10. Sandage (1966); 11. Heckman et al. (1982); 12. Hota et al. (2011); 13. Morganti et al. (2011); 14. Mao et al. (2015); 15. Bagchi et al. (2014). The radio luminosities in column 3 are in units of W Hz−1. Columns 4 and 5 give the optical spectroscopic and radio morphological classification respectively, with uncertain classifications indicated by a question mark; a DD designation in column 5 indicates a double-double source.|
With the development of sensitive, linear CCD detectors in the 1980s, it became possible to make quantitative studies of the radial light profiles of the hosts down to low surface brightness limits, and to compare them with those of non-active elliptical galaxies. Such studies have the potential to reveal disk-like components in the hosts of radio AGN that might not be immediately apparent upon cursory visual inspection of the images. Many of these studies fit the radial fall off in surface brightness with a Sersic profile:
where r is the radius, µ(r) is the surface brightness in magnitudes arcsec−2 at radius r, n is the Sersic index, Re is the effective radius that contains half the total light, and a and b are constants for a particular profile (Sersic, 1963). Note that n = 4 represents an R1/4 profile characteristic of elliptical galaxies (de Vaucouleurs, 1948) and n = 1 an exponential profile characteristic of the disks of spiral galaxies. Some studies explicitly model a nuclear point source component — especially important in the case of objects with unresolved broad-line nuclei — and/or include more than one Sersic profile component.
The results of the profile fitting have been mixed. On the one hand, some studies have found that the surface brightness profiles of radio galaxies are similar to those of quiescent elliptical galaxies, with a preference for R1/4 outer profiles in the majority of cases, rather than the exponential profiles characteristic of the disks of the spiral galaxies (Heckman et al., 1986, Smith & Heckman, 1989a, Smith & Heckman, 1989b, McLure et al., 1999, Dunlop et al., 2003). On the other hand, a number of studies have presented evidence for significant departures from pure R1/4 profiles. Concentrating first on ground-based optical studies, Colina & De Juan (1995) found that the outer surface brightness profiles depart significantly from an R1/4 law in 65% of their sample of 44 nearby FRI radio galaxies, while Govoni et al. (2000) required an additional exponential component to fit the outer profiles of 29% of their mixed sample of 72 nearby FRI and FRII radio sources. Further evidence for departures from a single Sersic profile is provided by the H-band HST imaging results for a sample of 82 FRI and FRII 3CR sources presented by Donzelli et al. (2007), who require an additional exponential component to fit the outer surface brightness profiles in 45% of their sample.
The departures of the outer surface brightness profiles from R1/4 laws might be taken as evidence for major disk components in a significant fraction of the radio AGN population, but there are other explanations. In particular, given that radio AGN are hosted by massive elliptical galaxies that are sometimes at the centres of rich galaxy clusters, it is possible that the departures from a pure R1/4 or Sersic law are due to the presence of cD-like outer envelopes, since the outer envelopes of cD galaxies often show significant excesses over and above the extrapolations of the inner Sersic or R1/4 profiles (e.g. Seigar, Graham & Jergen, 2007). However, Colina & De Juan (1995) have argued that the hosts of the FRI sources differ from cD galaxies in the sense that the departures from a pure R1/4 profile set in at higher surface brightness levels in the FRI hosts than they do in cD galaxies. Alternatively, the departures might be explained in terms of the light of the accumulated debris of galaxies merging with the radio AGN hosts, with these mergers perhaps triggering the AGN activity (Colina & De Juan, 1995).
It is also important to consider why some of the most recent HST-based studies provide apparently contradictory results. Most notably, McLure et al. (1999) and Dunlop et al. (2003) were able to model the R-band light profiles all 20 intermediate redshift, radio AGN in their sample using single Sersic profiles with n ∼ 4 (range: 3.6 < n < 5.3), whereas Donzelli et al. (2007) found evidence for departures from an R1/4 law in the outer profiles of their larger sample of 3CR sources using H-band observations. At least some of the ambiguities in the results may be due to differences in samples/observations/analysis techniques. For example, it is notable that the detector used by Dunlop et al. (2003) has a much wider FOV than that used by Donzelli et al. (2007) for the objects in their sample in the overlapping redshift range, thus allowing more accurate sky subtraction; the Dunlop et al. (2003) observations are also more sensitive, reaching a magnitude fainter in surface brightness.
Although the current results on the outer surface brightness profiles may not be clear-cut, in other regards the hosts of the radio AGN show much in common with elliptical galaxies. For example, several studies have now demonstrated that radio galaxies follow the correlations between effective radius and the mean surface brightness at the effective radius (the so-called Kormendy relation) for non-active elliptical galaxies (Smith & Heckman, 1989b, Govoni et al., 2000, Dunlop et al., 2003). In addition, although data on the stellar kinematics are currently sparse for radio AGN, due to the observational difficulties in obtaining accurate results for all but the closest objects, the existing results provide evidence that radio galaxies fall on the fundamental plane that relates the kinematic and photometric properties of elliptical galaxies (Smith, Heckman & Illingworth, 1990b, Bettoni et al., 2001).
The final important point to make about the hosts of radio AGN is that their stellar masses are relatively large: optical and near-IR results for low and intermediate redshift samples yield estimates in the range 1011 < Mstellar < 2 × 1012 M⊙ (Dunlop et al., 2003, Inskip et al., 2010, Tadhunter et al., 2011, see Figure 17 below), corresponding to ∼0.7 — 10m*, where m* = 1.4 × 1011 is the characteristic mass of the galaxy mass function (Cole et al., 2001). This result appears to hold even in cases where the hosts show evidence for disk morphologies (see references in Table 4 above), or signs of recent star formation activity (Tadhunter et al., 2011). Given the correlations between black hole mass and host galaxy properties (Kormendy & Ho, 2013), the black hole masses of radio AGN are correspondingly large (Mbh > 108 M⊙). Therefore it has been suggested that, along with black hole spin, the black hole mass is one of the key factors that determine whether an AGN is radio-loud (e.g. Lacy et al., 2001, Dunlop et al., 2003, Sikora, Stawarz & Lasota, 2007, Chiaberge & Marconi, 2011).
3.2. Morphological peculiarities
Despite their predominantly early-type character, it long been noted that the hosts of radio AGN show peculiarities in their detailed optical morphologies (Baade & Minkowski, 1954, Matthews, Morgan & Schmidt, 1964). These peculiarities include: dust lanes, large-scale tidal tails, fans, shells and bridges; double nuclei; and, at a more subtle level, isophotal twists. The first truly systematic studies of these peculiarities were made by Heckman et al. (1986) and Smith & Heckman (1989a, b), using CDD observations of a large sample of radio galaxies of all types, most selected from the 3CR catalogue. Based on isophotal analysis, these studies found that 50% of all the radio AGN host galaxies show departures from elliptical symmetry. Further, it was noted that 50% of the SLRG in the sample show tidal tails, fans, shells and dust lanes at relatively high levels of surface brightness (µv < 25 mag arcsec−2), whereas only 7% of WLRG show similar evidence. The rate of occurrence of tidal features in the SLRG was found to be the higher than the ∼ 10% of quiescent elliptical galaxies that show evidence for shell structures in photographic images (Malin & Carter, 1983). Together, these results appeared consistent with the idea that the SLRG are triggered in galaxy mergers, with perhaps a different triggering mechanism for the WLRG.
In apparent contradiction, some more recent HST-based imaging studies of nearby AGN found less evidence for tidal features. In particular, although some of the 20 radio AGN in the sample of Dunlop et al. (2003) show morphological peculiarities in their R-band HST images, the rate of such features was significantly lower than in the earlier Smith & Heckman (1989a, b) study. Indeed, Dunlop et al. (2003) could find no clear difference between the rate of galaxy interactions in their luminous AGN sample and that of a matched sample of quiescent giant elliptical galaxies at the centres Abell clusters of similar richness to those of the AGN hosts. However, despite their clearly superior spatial resolution, the surface brightness depths achieved in HST observations are sometimes inferior to those of the ground-based studies, because the HST is a relatively small telescope, and any diffuse tidal features are spread over many pixels of the HST detectors. This point is emphasised by the work of Bennert et al. (2008), who repeated the HST imaging of 5 of the Dunlop et al. (2003) objects using longer exposure times (5 orbits rather than 1 orbit) and a more sensitive detector (the ACS rather than WFPC2), and found that 4/5 of the objects show tidal features that were not detected in the earlier study.
Recognising the importance of surface brightness depth in studies of this type, Ramos Almeida et al. (2011a), Ramos Almeida et al. (2012) undertook a deep r'- and i'-band imaging study of the complete intermediate redshift 0.05 < z < 0.7 2Jy sample using the Gemini South telescope, reaching a magnitude fainter in surface brightness depth compared with the earlier Smith & Heckman (1989a,b) study, and with significantly better seeing (0.55 < FWHM < 1.15 arcsec compared with 1 < FWHM < 2 arcsec). Crucially, they also compared their results with those for two comparison samples of quiescent early-type galaxies matched in absolute magnitude and imaged to similar surface brightness depths: the OBEY sample of nearby elliptical galaxies (Tal et al., 2009) and a sample of intermediate redshift early-type galaxy morphologically selected from deep Subaru images of the Extended Groth Strip (EGS).
Example Gemini images of some of the 2Jy sources are shown in Figure 12. Strikingly, 94% of the 35 SLRG in the 2Jy sample show tidal features 18 or close double nuclei (< 10 kpc separation), whereas only 27% of the 11 WLRG in the sample show similar features. This confirms the dichotomy in the incidence of morphological peculiarities between radio AGN with strong and weak emission lines originally noted by Heckman et al. (1986) and Smith & Heckman (1989b), and is consistent with the idea that the AGN triggering mechanisms for the two groups are different (see discussion in section 3.6 below).
Figure 12. Examples of optical r' Gemini-S images of 2Jy radio galaxies from the study of Ramos Almeida et al. (2011a). In each case, the radio source host galaxy is at the centre of the image. Note the diversity in the detailed structures: while PKS2221-02, PKS0349-27, PKS0035-02, PKS1934-63 and PKS1151-34 are pre-coalescence systems that show evidence for tidal interactions with neighbouring galaxies, PKS2314+03, PKS1733-56, PKS1559+02 and PKS0442-28 are post-coalescence systems that show evidence for high surface brightness tidal features (PKS2314+03, PKS1733-56) or more subtle shell structures (PKS1559+02, PKS0442-28). See Ramos Almeida et al. (2011a), Ramos Almeida et al. (2012) for full details.
Note that, in terms of firmly establishing that radio AGN are triggered in galaxy mergers, it is not sufficient simply to demonstrate a high incidence of tidal features. This is because a large proportion of the quiescent elliptical galaxy population show tidal features at faint surface brightness levels (van Dokkum, 2005, Tal et al., 2009, Duc et al., 2015). Therefore comparisons with control samples are important. Concentrating first on the incidence of morphological features with surface brightnesses µv < 26.2 mag arsec−2 — this limit includes all the tidal features detected in the 2Jy radio galaxies — Ramos Almeida et al. (2012) show that such features are present in 93% and 95% of the SLRG at redshifts z < 0.2 and 0.2 ≤ z < 0.7 respectively, but in only 67% of the 55 quiescent elliptical galaxies in z < 0.01 OBEY sample, and 55% of the 109 early-type galaxies in the 0.2 < z < 0.7 EGS sample. Moreover, the tidal features detected in the 2Jy radio galaxies have surface brightnesses that are 1 - 2 magnitudes brighter on average than those detected in the comparison sample galaxies.
Overall, the deep optical studies provide compelling evidence that SLRG are triggered in galaxy mergers. This is consistent with the recent results for radio-quiet AGN with quasar-like nuclei (Lbol > 1038 W: Bessiere et al., 2012, Treister et al., 2012) and high redshift radio galaxies (z > 1: Chiaberge et al., 2015), but in contrast with the results for samples of low-to-moderate luminosity AGN (Lbol < 1037.5 W: Grogin et al., 2005, Cisternas et al., 2011). The dichotomy between high and low luminosity AGN suggests that, while luminous AGN favour galaxy mergers as a triggering mechanism, low/moderate luminosity AGN are triggered via secular processes. Interestingly, we see evidence for this dichotomy within the radio AGN population through the morphological differences between SLRG and WLRG.
The other striking feature of the deep imaging results is that radio galaxies are diverse in their detailed morphological properties: Ramos Almeida et al. (2011a) found that 37% of the SLRG in the full 2Jy sample are pre-mergers in the sense that the host galaxies are tidally interacting with companion galaxies (e.g. have bridges) or have close double nuclei, whereas 57% show tidal features consistent with the hosts being observed at or after the time of coalescence of the merging nuclei.
Despite the strong evidence they provide for the triggering of SLRG in galaxy mergers, the images alone provide only limited information about the types of mergers involved in the triggering events. Fortunately there are other facets of the triggering events that can provide further information, including the star formation properties (section 3.3), the cool gas contents (section 3.4), and the large-scale environments (section 3.5). For example, if the radio AGN were triggered at the peaks of major, gas-rich mergers we would expect them to appear similar to ultra luminous infrared galaxies (ULIRGs: Sanders & Mirabel, 1996), with large masses of cool ISM, prodigious star formation activity, and relatively low density galaxy environments; detailed spectral synthesis modelling of the optical spectra also has the potential to provide information about the timing of the triggering of the AGN relative to the peak of the merger-induced starburst (e.g. Canalizo & Stockton, 2001, Tadhunter et al., 2005, Tadhunter et al., 2011). These aspects are considered in the following sections.
3.3. Star formation
Measuring the level of star formation activity in the host galaxies of radio AGN is challenging. Figure 13 illustrates the problem: as well as the direct starlight and emission from dust heated by recent star formation activity, several AGN-related components can contribute to the UV, optical and infrared continuum emission, even in cases where the broad-line AGN nucleus is obscured. Apart from the components shown in Figure 13, non-thermal radiation from compact synchrotron-emitting components (e.g. cores, jets, and lobes) also has the potential to contribute at all wavelengths in radio AGN.
Figure 13. Schematic showing various AGN-related continuum components that contribute at UV, optical, mid-IR and far-IR wavelengths. Blue lines indicate UV/optical components and red lines indicate mid- and far-IR components. By diluting the starlight (UV/optical wavelengths) and the emission of dust heated by recent star formation activity (mid- and far-IR wavelengths), these component make it challenging to determine the star formation properties of the host galaxies, especially for the objects with the most luminous AGN.
Taking full account of the potential for AGN contamination, I now review the star formation properties of host galaxies of radio AGN at UV/optical, mid-IR, and far-IR wavelengths.
3.3.1. UV/optical diagnostics
The earliest studies of star formation in radio AGN were based on the optical or optical/near-IR colours provided by deep photometric observations. For example, Smith & Heckman (1989b) found that a subset of their sample of z < 0.4 3CR radio galaxies show blue (B-V) colours relative to quiescent elliptical galaxies. Interestingly, the bluer colours are associated with the SLRG in their sample rather than the WLRG, most of which tend to show red, elliptical-like colours. These results are broadly consistent with those obtained using spectroscopic measurements of the 4000Å break (D4000) 19 by Tadhunter et al. (2002) and Herbert et al. (2010), and HST photometric measurements of the UV/optical colours by Allen et al. (2002) and Baldi & Capetti (2008). However, it is important to emphasise that the presence of blue or UV excesses does not necessarily imply high levels of star formation. For example, in their detailed study of the UV emission in 3C236 — one of the nearby radio galaxies with the brightest off-nuclear UV structures — O'Dea et al. (2001) show that the star formation rates in the UV knots are relatively modest: typically a few solar masses per year.
The higher incidence of blue/UV excesses in SLRG than in WLRG might be taken as evidence for distinct star formation histories for the two classes, perhaps related to differeces in the triggering mechanisms (e.g. Smith & Heckman, 1989b, Allen et al., 2002, Baldi & Capetti, 2008, Herbert et al., 2010). However, at UV/optical wavelengths there is a strong potential for contamination by AGN-related continuum components. These include: scattered AGN light (e.g. Tadhunter, Scarrott & Rolph, 1990, Tadhunter et al., 1992) and nebular continuum (Dickson et al., 1995) 20 from the NLR; and direct AGN emission from weak or partially obscured broad-line AGN (Shaw et al., 1995). Since the luminosities of the nebular continuum, scattered AGN continuum, and narrow emission lines all depend strongly on the intrinsic AGN luminosity and the covering factor of the NLR, the absolute level of the AGN-related continuum components is expected to strongly correlate with the emission line luminosity, with the highest degrees AGN contamination relative to the level of the stellar continuum found in objects with the highest equivalent width emission lines. Therefore, at least some of the correlation between the optical spectroscopic class and blue/UV excess could be due to AGN contamination.
Recognising the potential for AGN contamination, Tadhunter et al. (2002) used a combination of spectroscopy and polarimetry observations to directly quantify the contribution of various AGN continuum components to the optical and near-UV emission of the integrated light of a complete sample of 22 0.15 < z < 0.7 2Jy radio galaxies. Their polarimetry measurements show that scattered AGN light makes a significant contribution (up to 10 — 30% in some cases) to the UV continuum below 3600Å in 32% of the full sample (50% of the 11 NLRG and 20% of the 9 BLRG). Moreover, based on measurements of the Balmer emission lines, they show that nebular continuum is important in all the SLRG objects in the sample, contributing as much as 30% of the UV continuum. In a further 40% of objects, emission from the direct light of weak or partially obscured broad-line AGN also contributes at UV wavelengths.
Taking full account of the AGN-related components, Tadhunter et al. (2002) found that the spectra of 85% of their complete sample could be adequately modelled using a combination of quiescent elliptical template and a power law (PL), with the latter included to represent scattered or direct AGN light. However, in 3 objects (14% of the full sample) an additional young stellar population (YSP) with age tYSP < 2 Gyr was required to model the continuum in the region of the Balmer break; all three of these objects also show higher order Balmer lines in absorption, providing direct confirmation of the presence of a YSP. Subsequent deeper spectroscopy observations reported in Holt et al. (2007) have detected high order Balmer lines in one further object in the sample. Less directly, in a further 7 objects the PL contributions to the UV continuum are larger than expected on the basis of the polarimetry results if all the PL represents scattered light; since a very young YSP can mimic a PL, potentially this might indicate a YSP contribution to the UV continuum in these objects. Overall, between 18% (objects with detected Balmer absorption lines) and 55% (including objects with less direct evidence for YSP) of the 0.15 < z < 0.7 2Jy sample show evidence for recent star formation activity in their UV/optical spectra. This rate of detection of YSP components is similar to the ∼30 — 40% deduced by Aretxaga et al. (2001) and Wills et al. (2002) based on spectral synthesis modelling of low redshift 3CR sources. Note, however, that a large proportion of radio AGN may have lower levels of star formation activity that have not so far been detected due to the strength of light of the old stellar populations and/or AGN-related continuum components.
The detection of significant YSP in some radio AGN opens the prospect of deducing their detailed properties (masses, ages, reddening) in order to further investigate possible triggering mechanisms. This is challenging work because of the difficulty of adequately accounting for the AGN-related continuum components, as well as the degeneracies inherent in modelling multiple stellar components of different ages, some of which may be heavily reddened. Only the small proportion of radio AGN host galaxies that are already known to have significant YSP are suitable for detailed spectral synthesis modelling studies. The results of such studies reveal a substantial diversity in the YSP properties. On the one hand, in some nearby radio AGN the YSP have intermediate ages (0.1 < tysp < 2 Gyr) and relatively large masses, suggestive of post-starburst populations (Tadhunter et al., 2005, Emonts et al., 2006, Holt et al., 2007, Tadhunter et al., 2011). In these cases, the ages of YSP are older than the typical lifetimes of the radio sources, consistent with the idea that the radio AGN have been triggered after the peaks of the merger-induced starbursts. On the other hand, in other radio AGN the YSP are much younger (tysp < 0.1 Gyr), suggesting that the starbursts and AGN are triggered quasi-simultaneously. Interestingly, younger YSP ages tend to be associated with radio AGN with quasar-like AGN luminosities, whereas older YSP ages are more frequently found in objects with lower AGN luminosities. In the case that the AGN with detected YSP are triggered in mergers, this difference is consistent with the idea that the lower luminosity radio AGN are triggered later in the merger sequence (Tadhunter et al., 2011).
3.3.2. Mid-IR diagnostics
A major advantage of the mid-IR wavelength region (5 < λ < 30µm) is that it is much less affected by dust extinction than the UV/optical. Potentially, this property allows the detection of regions of star formation that suffer high levels of dust extinction. However, contamination by the AGN is particularly pronounced in the mid-IR, where the continuum is dominated by emission of the warm dust in the circum-nuclear torus (mainly SLRG), or by the synchrotron emission of the inner jets (most WLRG). Only a few radio AGN with particularly strong starbursts show evidence that the emission from dust heated by regions of recent star formation makes a significant contribution to their continuum emission at 24µm (Dicken et al. 2012). Otherwise, the main diagnostics of star formation activity at these wavelengths are the broad polyaromatic hydrocarbon (PAH) features — emitted by large, sheet-like organic molecules — which are strong in starbursts, but might be destroyed by the hard radiation fields of AGN (Siebenmorgen, Krügel & Spoon, 2004).
The Spitzer Observatory made a major impact this field, because its high sensitivity allowed for the first time deep mid-IR spectroscopy observations to be made of large samples of radio AGN (e.g. Ogle et al., 2006, Cleary et al., 2007, Leipski et al., 2009). Dicken et al. (2012) exploited these capabilities to make a Spitzer/IRS survey of the complete 0.05 < z < 0.7 2Jy sample, combining the results with those for a complete sample of 3CRR FRII sources with redshifts z < 0.11 (see Figure 4 for example spectra).
The principal mid-IR PAH features are emitted at 6.6. 7.7 and 11.3 µm. Unfortunately, the strongest PAH blend at 7.7 µm is particularly broad, is potentially contaminated by narrow fine-structure emission lines, and falls at the blue edge of the 10 µm silicate absorption feature. Therefore, the 11.3 µm PAH feature, which is considerably narrower than the 7.7 µm feature, is preferred when attempting to establish the occurrence rate of PAH features, and hence the incidence of recent star formation activity, in the radio AGN population. Dicken et al. (2012) detected this feature in only 30% of their combined 2Jy and 3CRR sample. This immediately suggests that the hosts of most radio AGN do not harbour major starbursts, although the dilution effect of the strong AGN continuum at mid-IR wavelengths would make it difficult to detect low levels of star formation activity based on the PAH features alone.
Another technique involves using the mid- to far-IR (MFIR) colours (e.g. L70 µm / L24 µm), since an increase in the contribution from cool dust heated by young stars will tend to boost the far-IR relative to the mid-IR emission. Indeed, a number of studies have shown that the MFIR colours correlate well with a range of other diagnostics of the relative contributions of AGN and star formation activity (e.g. Veilleux et al., 2009, Dicken et al., 2009). In the case of the combined 2Jy and 3CRR sample studied by Dicken et al. (2010), Dicken et al. (2012), 4% of the 55 objects with the requisite photometric data were found to have red MFIR colours L70 µm / L24 µm ≥ 5, consistent with a significant contribution at far-IR wavelengths from dust heated by young stars.
3.3.3. Far-IR diagnostics
The strength of far-IR (30 < λ < 500µm) continuum is often considered to provide a clean diagnostic of star formation activity. This is because the dust heated by starbursts is relatively cool, such that its thermal emission peaks in the far-IR, whereas the AGN-heated dust in the compact, circum-nuclear torus is much warmer and radiates predominantly at shorter, mid-IR wavelengths. However, this division is not as clear-cut as it might at first seem. Mirroring the controversy surrounding the nature of the UV excess in radio AGN host galaxies (see section 3.3.1 above), the main issue here is the extent to which AGN-related continuum components contaminate the far-IR continuum: although the warmer dust components in the torus will radiate mostly at mid-IR wavelengths, clumpy torus models and those with large outer torus radii allow the possibility of significant cool dust components in the torus that radiate in the far-IR; AGN illumination of the larger-scale NLR may also power cool dust that radiates at far-IR wavelengths (Tadhunter et al., 2007); and, for radio AGN, contamination by synchrotron-emitting jets and lobes may also be an issue in some sources (e.g. Cleary et al., 2007, Dicken et al., 2008, Leipski et al., 2009, van der Wolk et al., 2010). Most attempts to investigate the degree of AGN contamination of the far-IR continuum have adopted the approach of fitting the mid- to far-IR continuum spectral energy distributions using a combination of starburst and AGN templates (e.g. Netzer et al, 2007, Mullaney et al., 2011).
The sensitive mid- and far-IR continuum observations of radio AGN made possible by Spitzer and Herschel have helped to illuminate this debate. First, analysis of the mid-IR to radio SEDs has shown that contamination by synchrotron emitting components is unlikely to be a serious issue at mid- to far-IR wavelengths for most SLRG, but may be important for WLRG (Cleary et al., 2007, Dicken et al., 2008, Leipski et al., 2009, van der Wolk et al., 2010, Dicken et al., 2016). Second, Dicken et al. (2009, 2010) have found strong correlations between the optical [OIII] emission line luminosity — often taken as a proxy of the overall level of AGN activity — and both the mid-IR 24µm and the far-IR 70µm continuum luminosities for their combined sample of 2Jy and 3CRR radio galaxies (see Figure 14), although the scatter is larger for the L70 µ m vs L[OIII] correlation. A partial rank correlation analysis shows that these correlations are not simply the result of the well-known correlations between redshift, radio power and emission line luminosity for flux limited samples (Dicken et al., 2009, Dicken et al., 2010).
Figure 14. Correlation plots for L24 µm vs L[OIII] (top) and L70 µm vs L[OIII] (bottom) for the combined 0.05 < z < 0.7 2Jy and z < 0.11 3CRR FRII sample of (Dicken et al., 2010). The 24µm and 70µm monochromatic continuum luminosities have units of W Hz−1, while the [OIII] luminosities have units of W. Objects showing evidence for recent star formation based on optical spectroscopy and/or PAH detection are indicated by blue stars, whereas those that show no such evidence are indicated by black, filled circles. Note the larger scatter in the L70 µm vs L[OIII] correlation. In the lower plot, the horizonal dashed lines show the equivalent star formation rates, based on the calibration of Calzetti et al. (2010) and assuming that all the far-IR emission is due to dust heated by young stars.
The strength L24 µm vs L[OIII] correlation is not surprising given that the 24µm and [OIII] emission are both produced by AGN illumination of the circum-nuclear structures: the dusty torus and/or the NLR in the case of the 24µm emission, and the NLR in the case of the [OIII] emission; as the bolometric luminosity of the AGN increases, the level of illumination of both the torus and the NLR increases, hence the correlation. The ratio of the mid-IR to the [OIII] luminosity depends on the relative covering factors of the torus and the NLR, and Dicken et al. (2009) have shown on energetic grounds that the observed correlation is consistent with the expected torus/NLR covering factor ratios.
There are two interpretations of the L70 µm vs L[OIII] correlation. First, the L70 µm far-IR continuum luminosity measures the rate of star formation, while the [OIII] luminosity measures intrinsic power of the AGN, and the correlation then arises because the star formation rate and AGN power are correlated (Schweitzer et al., 2006, Netzer et al, 2007). Second, the far-IR continuum is produced by AGN illumination of the extended dust in the NLR or circumnuclear torus, and is therefore strongly correlated with the [OIII] luminosity, which is produced by AGN illumination of the NLR (Tadhunter et al., 2007, Dicken et al., 2009). Support for the latter interpretation is provided by the fact that the slopes of the L24 µm vs L[OIII] and L70 µm vs L[OIII] are similar. Moreover, Dicken et al. (2009) have shown that luminosity of the far-IR continuum in most SLRG can be reproduced if the covering factors of the dust structures emitting the far-IR continuum are similar to those of the NLR. An important caveat on the latter argument is that, in order for the dust in the NLR to radiate significantly at far-IR wavelengths, it must be distributed on sufficiently large scales to avoid being heated too highly by the AGN. Whether the latter condition is fulfilled is currently uncertain, but it is notable that some SLRG show NLR that are extended on radial scales of kpc.
If the far-IR and mid-IR continuum are both produced by AGN illumination, why is the scatter larger for the L70 µm< vs L[OIII] correlation? Clues are provided by Figure 14, where the objects showing independent evidence for recent star formation activity are highlighted in the correlations. It is clear that all of the objects that form the upper envelope of the L70 µm vs L[OIII] correlation show independent evidence for star formation activity, whereas the main correlation is made up of the majority of radio AGN that show no such evidence. Therefore much of the enhanced scatter in the L70 µm vs L[OIII] correlation can be explained in terms of the boosting of the far-IR emission in a minority of the sources by starburst-heated dust, with the degree of boosting reaching a factor of 10 or more in some cases. It is also plausible that the far-IR continuum of most of the SLRG on the main correlation is dominated by the emission of cool, AGN-illuminated dust in the NLR. In this case, the far-IR results provide evidence for a wide range of star formation properties in radio AGN, with star formation rates varying by more than an order of magnitude for a given intrinsic AGN luminosity.
Uncertainties about the degree of contamination by AGN heated dust make it difficult to use the far-IR observations to derive accurate star formation rates (SFR) for individual objects; however, upper limits on the SFR can be obtained by assuming that all the far-IR continuum is associated with star formation activity. The horizontal dashed lines in Figure 14 show the star formation rates corresponding to different 70µm luminosities, as determined using the calibration of Calzetti et al. (2010). The two most far-IR luminous radio AGN plotted in Figure 14 — 3C459 and PKS2135-20 — have IR luminosities that qualify them as ULIRGs with star formation rates SFR > 100 M⊙ yr−1. However, for the majority of SLRG on the main correlation in Figure 14 the upper limiting star formation rates are much lower, and many have SFR <10 M⊙ yr−1 (see also Figure 16 below).
3.3.4. Star formation summary
Although the UV/optical, mid-IR and far-IR results on the star formation properties have been presented separately, there is a strong correlation between the results from the different diagnostics: objects showing clear signs of star formation activity based on their UV/optical spectra generally also show PAH features at mid-IR wavelengths, as well as a far-IR continuum excesses Dicken et al. (2012). This is demonstrated by the Venn diagram shown in Figure 15. Individually, each of the techniques gives a relatively low incidence of radio AGN with clear evidence for recent star formation activity (∼20 - 30%), and it is notable that the proportion of objects in the combined 2Jy and 3CRR sample of Dicken et al. (2012) that show any evidence for recent star formation activity is only 33%. Clearly the star formation properties of radio AGN are diverse. A major implication of the low detection rate of star formation activity is that a large fraction of radio AGN in the local universe cannot have been triggered at the peaks of major, gas-rich mergers, given that such mergers lead to prodigious star formation activity around the time of coalescence of the merging nuclei which would be detectable using at least one, but probably all, of the techniques considered above.
Figure 15. Venn diagram comparing the detection of recent star formation activity in individual objects in the combined 2Jy and 3CRR FRII sample of Dicken et al. (2012) using three different techniques: far-IR excess, UV/optical spectroscopy and PAH detection. In the case of the far-IR excess, objects lying more than 0.5 dex (i.e. a factor of 3) above the main correlation shown in Figure 14 are considered to show a far-IR excess that indicates recent star formation activity. For each technique, the detection rate of recent star formation activity using that technique across the full sample is shown in brackets.
This lack of evidence for major star formation activity in the majority of radio AGN is further demonstrated by Figure 16, which plots the maximum star formation rates derived from the far-IR photometry against the total stellar masses for a complete sample of nearby 2Jy sources with redshifts in the range 0.05 < z < 0.5. Apart from one clear starburst object, the majority of sources fall on or well below the “main sequence” of star formation derived for nearby star forming galaxies, and towards the part of the diagram normally occupied by “red and dead” galaxies. It is also notable that, while all but one of the WLRG fall well below the main sequence, a significant proportion of the SLRG (∼ 50%) fall close to or above the main sequence. However, this difference may in part reflect an enhanced level of contamination of far-IR light by AGN heated dust for the SLRG (see section 3.3.3), given their generally higher levels of AGN activity.
Although the majority of the local radio AGN population lacks evidence for the moderate-to-high levels of star formation that would be readily detected using existing techniques, this does not necessarily imply that their star formation rates are zero. Indeed, based on the typical cool ISM masses estimated from the Herschel results (108 < Mgas < 2 × 109 M⊙: see section 3.4 below), and the relationships between total gas mass and SFR deduced for the star forming galaxies at both low and high redshifts (Daddi et al., 2010), we would expect typical radio AGN to form stars at a rate of ∼ 0.5 – 30 M⊙ yr−1 if the radio AGN hosts have the type of efficient star formation that is characteristic of starbursts. However, the star formation rates would be at least a factor of ten lower than this if the star formation efficiency were more typical of the disks of spiral galaxies. Such low levels of star formation would be difficult to detect because of AGN-related components that effectively “hide” low levels of star formation activity, particularly in the objects with the most luminous AGN. Dilution by the strong continuum emission of the old stellar populations in the giant elliptical galaxy host galaxies also disfavours the detection of low levels of star formation activity at optical wavelengths.
Figure 16. Star formation rate vs total stellar mass for a complete subset of 30 sources from the Dicken et al. (2009) 2Jy sample (see Table 1) with redshifts in the range 0.05 < z < 0.5, excluding quasars and BLRG whose K-band light has a major contribution from AGN emission. Blue circles, green triangles and red squares represent SLRG/FRII, WLRG/FRII and WLRG/FRI sources respectively. The solid line shows the main sequence for nearby star forming galaxies (z < 0.2) derived by Elbaz et al. (2007) from SDSS data, with the dashed lines representing the 1σ scatter about this relationship, while the dotted line shows a more recent determination of the main sequence for z = 0.5 star forming galaxies taken from Schreiber et al. (2015). The stellar masses were estimated using K-band photometry under the assuption of a constant mass-to-light ratio: (LK / M*)⊙ = 0.9. The 64 kpc aperture K-band magnitudes of Inskip et al. (2010) were used for the majority of objects, but K-band magnitudes from 2MASS were used for the four sources not included in Inskip et al. (2010) study. The star formation rates were estimated from the Spitzer 70µm luminosities of Dicken et al. (2009) by using the calibration of Calzetti et al. (2010). Note that the star formation rates derived for the radio AGN are likely to represent upper limits because they have not been corrected for contributions to the 70 µm luminosities by AGN heated dust and synchrotron emission.
3.4. Dust and gas contents
Most triggering mechanisms for the AGN in SLRG involve the accretion of cool gas in a particular event such as a galaxy merger. Large reservoirs of cool gas are required in the triggering event to fuel the AGN and simultaneously grow the bulge of the host galaxy: a cool ISM reservoir of at least ∼109 M⊙ is required for an AGN at the lower end of the quasar luminosity range if the quasar has a lifetime of 107 yr (Tadhunter et al., 2014). Therefore, measurements of the cool gas contents of radio AGN have the potential to provide key information about the nature of the triggering events.
Until recently, most studies of the cool gas in radio AGN involved observations of the mm-wavelength CO lines (Evans et al., 2005, Smolcić & Riechers, 2009, Ocaña Flaquer, 2010). Unfortunately, even using long integration times, the sensitivity of such observations was limited. Moreover, the CO observations tended to involve relatively small, heterogeneous samples of radio AGN, or samples of objects that are unusually bright at far-IR wavelengths. Overall, the CO detection rates are low (∼20 – 60%), and the upper limits on the gas masses derived for the undetected sources are often large — well above the predicted minimum mass of the total reservoir required to trigger a quasar event.
Fortunately, the high sensitivity of the Herschel Observatory at the longer far-IR wavelengths (≥ 100 µm) has provided an alternative way of estimating the cool ISM masses using the emission of the associated dust, under the assumption that the dust radiates as a modified black body, and that the gas-to-dust ratio is constant. The deep Herschel observations of the complete 0.05 < z < 0.7 2Jy sample reported in Dicken et al. (2016) detect 100% of the sources at 100 µm and 85% at 160 µm. Considering first the 35 SLRG objects in the sample, the derived dust and cool ISM masses cover a wide range: 7 × 105 < Mdust < 3 × 108 M⊙ and 108 < Mgas < 4 × 1010 M⊙ (see Tadhunter et al., 2014, for details). For reference, the median cool ISM mass for the SLRG in the 2Jy sample (1.2 × 109 M⊙) is a factor ∼4 × lower than the total cool ISM mass of the Milky Way, and remarkably close to the predicted minimum total mass of the gas reservoir required for a quasar triggering event.
Figure 17 compares the distribution of dust masses for the 2Jy SLRG with those measured for samples of local ULIRGs and quiescent elliptical galaxies using similar techniques. The dust masses of the SLRG are typically a factor of 10 higher than those of quiescent elliptical galaxies, but a factor of 10 lower than those of ULIRGs; however, the high mass tail of the elliptical galaxy distribution overlaps with lower end of the SLRG distribution, and some SLRG have dust masses comparable with those of ULIRGs. It is notable that many of the latter group belong to the rare subset of radio galaxies that show evidence for recent star formation activity (see section 3.3 above).
Figure 17. Dust emission as a probe of the ISM contents of radio AGN host galaxies: (a)example radio to mid-IR SEDs of the 2Jy NLRG/FRII objects PKS0806-10 and PKS0349-27; (b) dust masses derived for the SLRG in the 0.05 < z < 0.7 2Jy sample of (Dicken et al., 2012) (top) compared with those for nearby ULRGS (middle) and elliptical galaxies (bottom). The vertical dashed and dotted lines indicate the dust mass of the Milky Way and LMC respectively. See Tadhunter et al. (2014) for details.
Analysis of the long-wavelength SEDs demonstrates that the far-IR emission of 5/6 of the WLRG/FRI sources in the 2Jy sample is likely to be dominated by non-thermal synchrotron emission (see Dicken et al., 2016). However, even if all the far-IR emission were entirely due to dust emission in these WLRG/FRI sources, their dust dust masses would fall at or below the lower end of the dust mass distribution of the SLRG. In contrast, all 5 WLRG/FRII sources in 2Jy sample show evidence for cool dust emission, with a dust mass distribution similar to that of the SLRG. This apparent difference between the dust properties of WLRG/FRI and WLRG/FRII sources has important implications for our understanding of the nature of the WLRG/FRII sources, potentially supporting the idea that the SLRG/FRII represent objects in which the AGN has recently switched off (see section 2.5). However, observations of a larger sample will be required to put this result on a firmer footing.
3.5. Large-scale environments
A further important property that might provide clues to the triggering mechanism is the large-scale environment of the host galaxy. For example, a high density cluster environment might favour fuelling of the radio AGN via the accretion of the hot ISM that is expected to be relatively dense in such environments, whereas a lower density group environment is more likely to be consistent with galaxy mergers as the main triggering mechanism, since major mergers are most common in such environments.
Several studies have sought to measure the environments of radio AGN using number counts derived from optical images, with much of the analysis based on the spatial clustering amplitude (Bgg). The most recent study by Ramos Almeida et al. (2013) for the 0.05 < z < 0.7 2Jy sample found clear differences between the environments of WLRG and SLRG: while the 82% of the 11 WLRG in the 2Jy sample are in relatively rich cluster environments (400 < Bgg < 1600, corresponding to Abell class 0, 1, 2 and 3 clusters), only 31% of the 35 SLRG in the same sample are in similarly dense environments; however, there is a significant overlap between the environmental properties of the two groups. Perhaps not surprisingly given the correlations between optical and radio classifications, similar results are obtained when comparing the environments of FRI and FRII galaxies, in the sense that the FRI sources inhabit significantly denser environments on average than FRII sources. All of these results for the 2Jy sample agree with those obtained for other samples of nearby radio AGN using number count techniques (Longair & Seldner, 1979, Prestage & Peacock, 1988, Zirbel, 1997).
An alternative method involves using X-ray observations to measure the luminosities of the extended hot gas surrounding the host galaxies. Such luminosities can be used as a proxy for the large scale galaxy environments, since richer galaxy environments tend to be associated with denser, more massive hot gaseous halos. Ineson et al. (2015) used this technique to make an extensive study of the environments of 2Jy and 3CRR sources at z < 0.2, finding a clear dichotomy between the extended X-ray luminosities of SLRG and WLRG, in the sense that WLRG show significantly higher extended X-ray luminosities for a given radio power; notwithstanding the uncertainties inherent in both techniques, they also found reasonable concordance with the number count analysis of Ramos Almeida et al. (2013) for 2Jy objects in the overlapping redshift range.
To summarize, there is now clear evidence from various studies at low and intermediate redshifts that WLRG occupy higher density environments than SLRG on average, but there is clearly some overlap between the two groups in terms of their environmental properties.
3.6. The triggering of radio AGN
Although the majority of radio AGN hold in common the feature that they are hosted by giant elliptical galaxies, the host galaxies are surprisingly diverse in their detailed properties. Together, the detailed information on the optical morphologies, star formation properties, gas contents and environments allows us to develop a coherent picture of the dominant triggering mechanisms for the different sub-types of radio AGN. Table 5 summarises the typical properties of the main SLRG/FRII and WLRG/FRI classes.
|Host galaxy||Giant ellipticals||Giant ellipticals|
|Environment||Group or weak cluster||Modest or rich cluster|
|Star formation rate||Low or modest||Low|
|Cool gas content||Modest||Low|
|AGN accretion mode||Geometrically thin, radiatively||Radiatively inefficient|
|thick accretion disk||accretion flow|
|Triggering mechanism||Modest merger||Hot ISM/ICM accretion|
The hosts of SLRG/FRII sources frequently show optical peculiarities such a tidal tails, fans, shells and bridges that have higher surface brightnesses than the tidal features detected in comparison samples of quiescent elliptical galaxies matched in absolute magnitude. They also show a tendency to be associated with modest, but not rich, galaxy environments. These two features are fully consistent with the idea that their AGN have been triggered in galaxy mergers. On the other hand, most SLRG/FRII hosts lack evidence for high levels of recent star formation activity, and contain relatively small amounts of cool ISM compared with ULIRGs, even if the cool ISM masses are larger than those of typical quiescent elliptical galaxies. Therefore, for the majority of SLRG/FRII sources, the triggering mergers are likely to have been relatively modest: a merger between a giant elliptical and a galaxy with a cold gas mass twice that of the Large Magellanic Cloud (LMC) would suffice; in these objects we are likely to be witnessing the late-time re-triggering of radio-AGN activity in mature elliptical galaxies with existing super-massive black holes.
The majority of SLRG/FRII sources could therefore represent a fleeting active phase in the evolution of a subset of general population of early-type galaxies with high stellar masses Mstar > 1011 M⊙, which formed the bulk of its stellar mass at z > 1 and has not shown major evolution in stellar mass or co-moving number density at more recent epochs (e.g. Pozzetti et al., 2010, McDermid et al., 2015). Interestingly, there is evidence that minor mergers have contributed most of the growth in stellar mass of this population since z = 1 (Kaviraj et al., 2011).
However, a minority of SLRG/FRII objects (< 20% of the full population) have cool gas masses and levels of star formation activity that are more comparable with those of ULIRGs. In these objects, major, gas-rich mergers are likely to have been involved in the triggering of the activity, and we may be witnessing the main event in terms of the formation of both the host galaxy bulges and the supermassive black holes.
Of course, these conclusions for SLRG/FRII objects refer to the redshift and radio power ranges covered by the bright radio AGN samples in the local universe (z < 0.7), and it is possible that the situation changes at higher redshifts. Indeed, a recent Herschel study of a sample of 3CR radio galaxies at high redshifts (1 < z < 2.5) found evidence for higher rates of star formation and larger dust masses than those of the local samples (Podigachoski et al., 2015), thus suggesting that major, gas-rich mergers may become a more common triggering mechanism for radio AGN at higher redshifts and/or radio powers.
It is important to draw the distinction here between modest or minor mergers 21 and major, gas-rich mergers. The debate about the triggering mechanism for luminous AGN (whether radio-loud or radio-quiet) has often been framed in terms of major, gas-rich mergers, with quasars perhaps representing a late, post-coalescence phase in the evolution of ULIRGs (Sanders et al., 1988a, Sanders et al., 1988b). Therefore, the hypothesis that luminous AGN are triggered in mergers has sometimes been rejected on the grounds that the host galaxies do not appear as major, ULIRG-like mergers. However, the results for the nearby radio AGN clearly demonstrate that luminous AGN activity can be triggered in more modest mergers, and also that the triggering occurs at a variety of merger stages, including pre-coalescence. Since the relatively subtle morphological, star formation and gas content signs of modest mergers may be difficult to detect in surveys of high redshift AGN, clearly some caution is required when interpreting the results from such surveys in terms of triggering mechanisms.
In stark contrast to the SLRG/FRII objects, the WLRG/FRI sources tend to favour more massive host galaxies, inhabit richer large-scale environments, show less evidence for recent star formation activity and tidal features, and have lower masses of cool ISM. These properties are consistent with the idea that, rather than being triggered by a sharp accretion event such as merger, the WLRG/FRI sources are fuelled by the more gradual accretion of the hot ISM from the host galaxy or cluster; certainly analysis of the LEGs in the SDSS sample suggests a high duty cycle for this type of radio AGN activity (Best et al., 2005). However, it is always dangerous to generalise, and a subset of the WLRG/FRI objects have host galaxy and/or environmental properties that are similar to those of SLRG/FRII sources. An obvious example of this is the closest radio AGN, the WLRG/FRI Centaurus A, which contains massive reservoir of cool gas in its prominent dust lane (∼ 3 × 109 M⊙: Parkin et al., 2012), is located in a group environment (e.g. Côté et al., 1997), and shows large-scale tidal features in deep imaging observations (Malin, Quinn & Graham, 1983). In such objects is possible that the original galaxy merger delivered a large reservoir of cool gas to the system and perhaps triggered a luminous, SLRG/FRII event at the merger peak, but subsequently the gas reservoir has settled into a more stable dynamical configuration, reducing the rate of accretion of cool gas into the nuclear regions and leading to lower-level WLRG/FRI activity. This would be consistent with the evidence for a delay between the merger-induced starburst and the triggering of the current phase of AGN activity based on analysis of the stellar populations in lower-luminosity radio AGN with YSP (Tadhunter et al., 2005, Emonts et al., 2006, Tadhunter et al., 2011).
One further possible fuel supply, which may be particularly relevant to the WLRG objects, is gas cooling from the hot X-ray haloes of the host galaxies and clusters and falling into the nuclear regions: the so-called “cooling flows”. Indeed, it has been argued that the kinematics of the extended emission line gas in some radio AGN, particuarly WLRG in cluster environments, are consistent with a cooling flow origin for the warm gas (Tadhunter, Fosbury & Quinn, 1989a, Baum, Heckman & van Breugel, 1992).
Considering more luminous AGN, cooling flows could be particularly effective under the conditions of chaotic cold accretion (Gaspari et al. 2013: see discussion in section 2.4 above) in the centres of rich clusters of galaxies. In this context, it is notable that a cooling flow has been implicated in the triggering of the luminous, quasar-like AGN and starburst at the centre of the Phoenix cluster (z = 0.596: McDonald et al., 2012). However, the Phoenix cluster is extreme in its X-ray properties — it is one of the most X-ray luminous galaxy clusters with one of the highest cooling rates. Therefore, it is not clear whether this mechanism is significant for the majority of SLRG, which are in relatively low density galaxy environments.
This discussion of triggering/fuelling mechanisms feeds into the question of whether it is the rate or the mode of accretion that is most important for determining the optical spectral properties of radio AGN (e.g. whether WLRG/LEG or SLRG/HEG: see section 2.4). Broadly, the results on the host galaxies and environments of the radio AGN are consistent with cold mode accretion via galaxy mergers for the SLRG objects, and hot mode accretion of gas from the gaseous haloes of the host galaxies/clusters for the WLRG. However, the modes are not independent of the rates. For example, because of the rarefied nature of the hot ISM at the centres of galaxies and clusters, hot mode will generally tend be associated with low rates to accretion. Moreover, it is likely that the types of events associated with cold mode accretion (e.g. galaxy mergers) tend to lead to higher rates of accretion and Eddington ratios above that required to produce a SLRG. However, the link between rate and mode is not one-to-one, and cases like Centaurus A suggest that cold accretion at low rates may be possible following a merger, once the gas has settled to a stable configuration.
Finally, it is interesting that the WLRG/FRII objects, which can be considered misfits in terms of their radio/optical classifications, also appear to show hybrid host/environment properties: while on average they tend to inhabit relatively rich environments that are similar to those WLRG/FRI sources (Ramos Almeida et al., 2013, Ineson et al., 2015), evidence is emerging that their cool gas contents are more similar to those of SLRG/FRII sources (Dicken et al., 2016); and one of the WLRG/FRII objects in the 2Jy sample (PKS0347+05) appears to be involved in a major, gas-rich galaxy merger (Tadhunter et al., 2011). Although intermittency/switch off may help to explain some of their properties (see section 2.5), the WLRG/FRII sources clearly present a challenge to our current understanding of radio AGN phenomenology.
18 Dust lanes have not been counted as tidal features. Note that only two SLRG in the intermediate redshift 2Jy sample — 3C105 (PKS0404+05) and PKS0252-71 — show no evidence for tidal features or close double nuclei. It is significant that one of these objects (3C105) is affected by an unusually high level of dust extinction at optical wavelengths (∼1 magnitude in the r'-band), which makes it harder to detect faint morphological features, while the other (PKS0252-71) is at higher redshifts and was observed in relatively poor seeing conditions. Therefore the true rate of occurrence of tidal features or close double nuclei in the 2Jy SLRG could approach 100%. Back.
19 The 4000Å break measures the ratio of fluxes measured in wavelength bins above and below the metal line blanketing break for old stars at 4000Å. The version of D4000 defined by Tadhunter et al. (2002) uses continuum bins that avoid strong emission lines (3750 < λ < 3850Å and 4150 < λ < 4250Å ). Back.
20 The nebular continuum comprises recombination, 2-photon and free-free emission from the warm ionised gas in the NLR: see Dickson et al. (1995) for details. Back.
21 Minor mergers are generally defined to be those in which one object involved in the merger has a mass of 25% or less of that of the other. Back.