"All things are double, one against another" Ecclesiasticus (xlvii. 24)
"The problem is how much to stress the sameness of twins and how much to emphasise their differences.", Baby and Child Care, Benjamin Spock, 1946
Shortly after the discovery of extragalactic radio sources, their nonthermal spectra and polarisations led to the conclusion that their emission is produced by synchrotron radiation from a relativistic plasma. The large radio luminosities together with the lifetimes of the synchrotron-emitting electrons imply total energies of 1060 erg (e.g. Miley 1980). These huge energies, the collimation on sub-parsec scales and the similarity of orientation between the compact and overall radio structures led to the conclusion that the collimated relativistic beams are produced by rotating supermassive black holes (SMBHs) located at the centres of the host galaxies. The SMBHs are postulated to have rotation axes aligned with the radio source axes. The radio sources are powered through gravitational energy from material accreting onto the SMBHs, that is converted into kinetic energy of the collimated relativistic jets (See Section 5.2).
2.1. HzRGS and low-redshift radio galaxies
Extragalactic radio sources can be classified according to their sizes.
Galaxies and quasars host extragalactic radio sources of all three classes, with extended radio sources predominantly identified with galaxies and compact sources mainly associated with quasars.
HzRGs generally host extended extragalactic radio sources and differ in several properties from low-redshift radio galaxies. At larger redshifts the typical radio luminosities increase, the typical sizes decrease (Section 2.3) and the typical radio spectra steepen (Section 2.4). The host galaxies of HzRGs differ from the low-redshift radio galaxy hosts in (i) the presence of emission-line halos (Section 3.2, (ii) increased clumpiness of the continuum emission (Section 4.3) and (iii) alignments of the radio sources with both the emission-line halos (Section 3.2.3) and the UV/optical galaxy continuum emission (Section 4.3).
Figure 2. Radio structure of 4C 41.17 at z = 3.8. [From van Breugel and Reuland, private communication]. See also (Reuland et al. 2003). Contours obtained at 1.4 GHz with the VLA are superimposed on a Keck narrow-band image in redshifted Ly, showing the warm ionized gas (Section 3.2. The radio angular size of ~ 13 arcseconds corresponds to a projected linear size of ~ 90 kpc. The radio spectrum of this object is shown in Figure 3c. Figure 5 illustrated the radio rotation measure of the brightest radio component and Figure 15 is a higher resolution radio contour map of the central region.
2.2. Radio sizes and morphologies. Size as an evolutionary clock
The radio structures of HzRGs have been studied by Carilli et al. (1997) and Pentericci et al. (2000). Both articles describe observations of 37 radio galaxies with z > 2 using the VLA at 4.7 and 8.2 GHz, at resolutions down to 0.25 arcec. In accordance with their large radio luminosities, most HzRGs are "Fanaroff-Riley Class II" radio sources (Fanaroff and Riley 1974), with double structures, edge-brightened mophologies and one or more hot-spots located at the extremities of their lobes (Miley 1980, Carilli et al. 1994, Carilli et al. 1997, Pentericci et al. 2000). In general HzRGs do not have appreciable flat-spectrum core components at their nuclei, but the use of the ultra-steep spectrum criterion in searching for HzRGs discriminates against finding HzRGs with flat-spectrum cores. Standard minimum energy assumptions (Miley 1980) give typical minimum pressures in these hotspots of a few x 10-9 dyn cm-2 and corresponding magnetic field strengths of a few hundred µG.
There have been a few VLBI observations of fine-scale structure in the lobes of extended radio sources associated with HzRGs (Gurvits et al. 1997, Cai et al. 2002, Pérez-Torres and De Breuck 2005, Pérez-Torres et al. 2006). A component of size ~ 65 pc has been detected several kpc from the nucleus of 4C 41.17 at z = 3.8 (Pérez-Torres et al. 2006). Consideration of the energetics suggests that the radio component is associated with a gas clump of mass MB 1.5 × 108 M. Intriguingly, this is typical for the masses of "predisrupted clumps" invoked as the progenitors of globular clusters (Fall and Rees 1977). Investigating fine scale radio structure in HzRGs and using the radio jet interactions to probe the interstellar medium in the early Universe are likely to be an important field of study for the more sensitive long-baseline interferometers that are presently under construction, such as e-Merlin and the e-EVN.
There are several correlations between the sizes of radio sources and and other properties (Röttgering et al. 2000). First, for 3C sources at lower redshifts, there is a relation with optical morphology. Smaller radio sources consist of several bright optical knots aligned along the radio axes, while larger sources are less lumpy and aligned (Best et al. 1998). Secondly, there is a relation with emission line properties. Smaller radio sources have generally lower ionisations, higher emission line fluxes and broader line widths than larger sources (Best et al. 1999). Thirdly, there is a connection with Ly absorption. HzRGs with smaller radio sizes are more likely to have Ly absorption than larger sources (Section 3.3.2) (van Ojik et al. 1997). Taken together, these correlations are consistent with an evolutionary scenario in which radio size can be used as a "clock" that measures the time elapsed since the start of the radio activity.
Note that the large sizes of radio sources (usually several tens of kpc) imply that the nucleus of host galaxies has been undergoing activity for at least > 105 y (light travel time across the source) and up to > 108 y (assuming that the jets advance at a few hundred km/s). Hence we know that AGN associated with extended radio sources must be long-lived. This is not the case for radio-quiet or compact radio quasars, that may only have been active quasars for a few × 102 y.
2.3. Radio size vs redshift correlation
It has been known since the nineteen sixties that there is a statistical decrease of the angular size of radio sources with redshift [Miley 1968]. More recent work includes measurement of the angular size-redshift relation for luminous extended radio sources (Nilsson et al. 1993, Neeser et al. 1995, Daly and Guerra 2002), quasars (Buchalter et al. 1998) and compact radio sources (Gurvits et al. 1999).
Over the years there have been many valiant attempts to use the angular-redshift relation to derive information about the geometry of the Universe (e.g. determination of q0 and 0) and even to set constraints on dark energy (Podariu et al. 2003). However, there are many observational selection effects involved. Furthermore, it is difficult to disentangle varying geometry of the Universe from effects due to physical evolution of the radio sources, their host galaxies and the surrounding ambient medium. For example, the sizes of radio sources can be expected to decrease at larger redshifts due to a systematic increase in density of the ambient medium. Also the energy density of the cosmic microwave background increases as (1 + z)4, substantially enhancing inverse Compton losses. This will tend to extinguish radio sources earlier in their lives, making them on average smaller.
In summary, the many non-cosmological effects that influence the angular size-redshift relation have made it impossible to draw robust conclusions about cosmology from such considerations.
2.4. Radio spectral index vs redshift correlation
One of the most intriguing properties of the relativistic plasma in HzRGs is the strong correlation that exists between the steepness of radio source spectra and the redshift of the associated radio galaxies (Tielens et al. 1979, Blumenthal and Miley 1979). Radio sources with very steep spectral indices at low frequencies 1 GHz tend to be associated with galaxies at high redshift (e.g. Figure 3). This empirical correlation between radio spectral steepness and redshift has proved to be an efficient method for finding distant radio galaxies. Most known HzRGs (Table 3) have been discovered through following up those radio sources with the steepest ten percentile of radio spectra (spectral index, ~ -1).
The conventional explanation for the z ~ correlation is that it is the result of a concave radio spectrum (see Cygnus A in Figure 3), coupled with a radio K-correction. For higher redshifts, observations of sources at a fixed observed frequency will sample emission at higher rest-frame frequencies where the concave spectrum becomes steeper. The most important mechanisms for making the radio spectra concave are synchrotron and inverse Compton losses at high frequencies (e.g. Klamer et al. 2006) and synchrotron self-absorption at low frequencies.
Although there is evidence from radio colour-colour plots that systematic concave curvature is present in the radio spectra (e.g. Bornancini et al. 2007), such an explanation alone is insufficient to explain the observed correlation. The radio spectra of many distant luminous radio galaxies are not concave at the relevant frequencies. The radio source with the most accurately determined spectrum over a wide frequency range is 4C 41.17 at z = 3.8 (bottom right in Figure 3). This source has an extremely straight spectrum between 40 MHz and 5 GHz, the relevant frequency range for the z ~ correlation (Chambers et al. 1990). Although the spectrum steepens above 5 GHz, this is too high a frequency to contribute to the z ~ correlation. Furthermore, a recent study by Klamer et al. (2006) showed that 33 of 37 sources in their SUMSS-NVSS sample have straight and not concave spectra between 0.8 and 18 GHz.
Figure 3. Left. Plot of radio spectral index versus redshift, showing that more distant sources have steeper spectra. [From De Breuck et al. (2000)]. Above right. Radio spectrum of the luminous radio galaxy Cygnus A at z = 0.05, showing spectral curvature. Bottom right. Radio spectrum of the HzRG 4C 41.17 and its various components. [From Chambers et al. (1990)]. This is still one of the most well determined spectra of a HzRG at low frequencies. Note the absence of significant spectral curvature.
Two alternative effects have been proposed to explain the observed z ~ relation. The first possibility is that the z ~ relation is an indirect manifestation of a luminosity, L ~ effect (Chambers et al. 1990, Blundell and Rawlings 1999).
Classical synchrotron theory predicts that continuous particle injection will result in a spectrum with a low-frequency cut-off, l, whose frequency depends on the source luminosity, L according to l ~ L-6/7. For a flux-limited sample, Malmquist bias will cause sources at higher redshift to have preferentially larger radio luminosities. Over the relevant frequency range, the L-6/7 vs l effect would therefore result in the observation of an z ~ relation. However, it is unlikely that this luminosity - spectrum relation is the correct explanation, or at least the whole story. Athreya and Kapahi (1998) showed that a z ~ correlation still persists even when samples are restricted to a limited range of L.
A second explanation is that some physical effect causes the spectral index to steepen with higher ambient density and that the ambient density increases with redshift (Athreya and Kapahi 1998, Klamer et al. 2006). For example, putting the radio source in a denser environment would cause the upstream fluid velocity of the relativistic particles to decrease and a first-order Fermi acceleration process would then produce a steeper synchrotron spectrum. Recently Klamer et al. (2006) pointed out that such a mechanism would (i) result in both z ~ and L ~ correlations and (ii) provide a natural physical link between high-redshift radio galaxies and nearby cluster halos.
However, it is difficult to produce the observed z ~ relation from such a simple density-dependent effect alone. The clumpy UV/optical morphologies (e.g Section 6.1) indicate that the density of gas around HzRGs is highly non-uniform and that the density is larger close to the nucleus than in the outer regions. Furthermore, the internal spatial variations of spectral index within individual source is observed to be smaller than the source to source variations of the integrated spectral indices (e.g. Carilli et al. 1994). If the ambient medium is highly non-uniform, how can one side of a radio source "know" that the other side has an uncommon ultra-steep spectrum? It is therefore likelier that the ultra-steep spectra are produced by some mechanism in which the spectral index is determined within the galaxy nucleus rather than by the environment at the locations of the radio lobes.
In summary, the origin of the z ~ effect is still unclear and more detailed information is needed about the dependence of the radio spectrum on redshift. In the near future accurate measurements of low-frequency radio spectra of HzRGs with new facilities such as LOFAR will be important for such studies.
2.5. Nonthermal X-ray emission
Although X-ray measurements of HzRGs are sensitivity-limited, significant progress in the field has been made in the last decade using the Chandra and XMM X-ray telescopes. Extended X-ray emission has been detected from about a dozen high-redshift radio galaxies and radio-loud quasars (Carilli et al. 2002, Fabian et al. 2003, Fabian et al. 2003, Scharf et al. 2003, Yuan et al. 2003, Belsole et al. 2004, Overzier et al. 2005, Blundell et al. 2006, Erlund et al. 2006, Johnson et al. 2007). The extended X-ray emission is typically elongated in the direction of the radio source, indicating that there is some physical link between the X-ray emission and the relativistic plasma.
Several mechanisms have been proposed for producing the extended X-ray emission. The one that is most widely suggested inverse Compton scattering (IC) of the cosmic microwave background (Fabian et al. 2003, Scharf et al. 2003, Belsole et al. 2004, Erlund et al. 2006, Johnson et al. 2007). Because the density of CMB photons increase as (1 + z)4, IC scattering of the CMB becomes increasingly important at high redshift. Under the assumption that the X-ray emission is due to this process, comparison of the radio and X-ray luminosities (Felten and Rees 1969) yields magnetic field strengths consistent with equipartition (Belsole et al. 2004, Overzier et al. 2005, Johnson et al. 2007). Because the radiative lifetimes of radio synchrotron-emitting electrons are shorter than the lifetimes of the X-ray emitting IC electrons, the IC emission traces older particles. In 4C 23.56 at z = 2.48, the X-ray emission is observed to extend by ~ 500 kpc (Johnson et al. 2007), implying an energy in both relativistic and IC-emitting electrons of 1059 erg, an energy reservoir equivalent to ~ 108 supernovae.
Other processes that have been invoked to produce the extended X-ray emission include inverse-Compton up-scattering of synchrotron photons in the jet (synchrotron self-Compton emission - SSC) (Scharf et al. 2003) and thermal emission from shocks (Carilli et al. 2002, Belsole et al. 2004) (see Section 3.1).