11.2. Old, Confined (Frustrated) Sources
One explanation for the large relative numbers of GPS and CSS sources compared with the large classical doubles is that they have similar ages. This requires that the GPS and CSS sources are kept compact for a significant fraction of their lifetimes. Partly because of radio morphologies, which suggested that the sources were undergoing strong interaction, it was suggested that the CSS sources were confined by interaction with dense gas in their environment (see, e.g., Wilkinson et al. 1984a; van Breugel et al. 1984b). O'Dea et al. (1991) suggested that the GPS sources could be confined by interaction with dense clouds in their environment. Gopal-Krishna (1995) suggested that distortions of the central gaseous torus during a merger could cause the jet to collide with dense gas. As discussed in section 10, there are now multiple pieces of evidence for dense gas in the environments of GPS and CSS sources.
Gopal-Krishna & Wiita (1991) suggested that CSS sources are members of an intrinsically less radio-luminous source population, and that their radio luminosity is enhanced to its observed values because of its dense environment. They note that work by Eilek & Shore (1989) shows that the efficiency of conversion of jet kinetic energy into radio luminosity can vary depending on age and environment of the source. Gopal-Krishna & Wiita suggest that the efficiency is a factor of about 6 higher in the CSS sources and that since they are intrinsically weaker by this factor, their environments are sufficient to confine them.
There is evidence that the jet kinetic energy and the flux of ionizing photons from the nucleus are proportional (see, e.g., Baum & Heckman 1989; Rawlings & Saunders 1991; Baum, Zirbel, & O'Dea 1995). This would suggest that the CSS sources should have a ratio of optical line to radio luminosity, which is low compared to the large-scale classical doubles. However, this is not seen (section 8.8). In addition, the ratio of MFIR to radio luminosity should also be low compared to the large-scale classical doubles, and again this is not seen (section 7). Thus, the hypothesis that they are intrinsically fainter but more efficient (by factors of more than a few) is not supported by the current data. However, if the optical line emission and MFIR emission can also be enhanced in the CSS sources (e.g., through interaction with the radio source or the presence of additional gas and dust in the host galaxy), such an increase in efficiency would be permitted.
De Young (1993) carried out simple numerical hydrodynamical simulations of a jet propagating in a smooth medium and in a medium with a fine "mist" of dense clouds (cloud radius of 1 pc). Both cases gave similar results, and De Young found that most CSS sources could be confined if the average density in the ISM were ~ 1-10 cm-3. The most powerful sources would require even higher densities for confinement. For the case of a smooth medium with constant density, the total gas mass required is
where R is the radius, and n0 is the density. And for the case of a density varying with radius as R-2, which describes steady infall, the enclosed mass is
where n1 is the density at 1 kpc. It seems unlikely that a uniform dense medium of mass 1012 M extends out to the 10 kpc required to confine CSS sources, though the CSS sources may well interact with clouds of gas that are infalling into the center of the galaxy.
Carvalho (1994, 1998) has presented a simple analytic model of jet interactions with dense clouds and concludes that interaction with dense clouds would slow down the jet more than propagation through a smooth medium with the same average density. He finds that interaction with clouds could result in GPS sources having lifetimes as old as that of the large-scale doubles. Carvalho suggests that the GPS sources could be confined by a clumpy medium with a mass of ~ 109-1010 M on the scale of hundreds of pc. Such masses are in general consistent with the small number of observations in Table 9 and are similar to those observed in the ULIRGs. Thus, it appears that interactions with dense gas could confine the radio sources. However, it is not yet clear whether sufficient gas generally exists in these sources to achieve confinement.
If the presence of large amounts of dense gas in the host galaxies is confirmed via sensitive observations of a large number of objects, this would support the "frustrated" source hypothesis. A scenario would be that an encounter with a gas-rich companion has resulted in large amounts of gas being accreted by the host galaxy. As shown in numerical simulations (Hernquist 1989; Barnes & Hernquist 1991; Bekki & Noguchi 1994; Hernquist & Mihos 1995) and observed in the ULIRGs (see, e.g., Scoville et al. 1994), the gas sinks to the center of the galaxy (to subkiloparsec scales). The gas can then trigger/feed the nuclear activity and confine the resulting radio source. The accumulation of dense gas in the nucleus should also trigger a large starburst (see, e.g., Sanders et al. 1988; Norman & Scoville 1988; Mihos & Hernquist 1994). The clouds will confine the radio source for the shorter of the following times: (1) the time for the jet to push aside or drill through the clouds (see, e.g., De Young 1991; Carvalho 1994, 1998) or (2) the time it takes for a starburst-driven superwind to disperse the clouds (see, e.g., Heckman, Armus, & Miley 1990). These timescales can both be of order 107 yr and will result in confinement of GPS sources to the subkiloparsec scale for most if not all of their lifetime. One prediction of this scenario is that there should be a large starburst in GPS and CSS sources. GPS galaxies have r - i colors consistent with passively evolving ellipticals and do not show significant amounts of blue light from young stars. High-resolution near-IR imaging is needed to detect dust-obscured starbursts.