HST, radio and infrared observations of 28 3CR radio galaxies at redshift z ∼ 1 — II. Old stellar populations in central cluster galaxies

P. N. Best M. S. Longair H. J. A. Rttgering 0 Cavendish Laboratory , Madingley Road, Cambridge CB3 OHE 1 Sterrewacht Leiden, Huygens Laboratory , PO Box 9513, 2300 RA Leiden , the Netherlands Hubble Space Telescope images of 3CR radio galaxies at redshifts 0.6 < z < 1.8 have shown a remarkable variety of structures, generally aligned along the radio axis, indicating that the radio source strongly influences the optical appearance of these galaxies. In this paper we investigate the host galaxies underlying this aligned emission, combining the HST data with ground-based infrared images. An investigation of the spectral energy distributions of the galaxies shows that the contribution of the aligned blue component to the K-band light is generally small (- 10 per cent). The radial intensity profiles of the galaxies are well matched at radii - 35 kpc by de Vaucouleurs' law, demonstrating that the K-band light is dominated by that of an elliptical galaxy. There is no evidence for a nuclear point source, in addition to the de Vaucouleurs profile, with a contribution 15 per cent of the total K-band flux density, except in two cases, 3C 22 and 3C 41. We conclude that the K-band emission of the distant 3CR galaxies is dominated by starlight. The magnitudes, colours and location of the distant 3CR galaxies on the projected fundamental plane indicate that their stellar populations formed at high redshift and have since been evolving passively. Large characteristic radii are derived for the 3CR galaxies, indicating that they must be highly evolved dynamically, even at a redshift of 1. At radii larger than ---35 kpc, a combined galaxy profile clearly shows an excess of emission as compared with de Vaucouleurs' law, indicating that at least some of the galaxies possess cD-type haloes. This supports other independent evidence for the hypothesis that the distant 3CR galaxies lie in moderately rich (proto-)clusters. Since the nearby FR II galaxies in the 3CR catalogue lie in more diffuse environments and do not possess cD haloes, the galactic environments of the 3CR galaxies must change with redshift. The K-z relation of the 3CR galaxies cannot, therefore, be interpreted using a standard 'closed-box, passively evolving stellar population' model, whereby the galaxies that host distant 3CR sources will evolve into the galaxies that host nearby 3CR FR 11 sources. At redshifts z -- 1, the absolute K magnitudes of the stellar populations of the 3CR galaxies are brighter than those of the lower radio power 6C galaxies, indicating that the 3CR galaxies contain a greater mass of stars; this is consistent with them lying towards the centres of clusters. Powerful high-redshift radio galaxies possess radio beams the kinetic power of which is close to the Eddington limiting luminosity of a central supermassive black hole. Since the mass of the black hole is likely to scale in proportion to the mass of the host galaxy, the 3CR galaxies will contain more massive central engines than the 6C galaxies, which accounts for their more powerful radio emission. At redshifts z < 0.6, the beam power of the radio sources is limited by the availability of fuel for the central engine rather than by the black hole mass, and so no correlation is expected between the radio power and the mass of the host galaxy. 1 INTRODUCTION The 3CR sample of radio sources defined by Laing et al. (1983) consists of the brightest extragalactic radio sources in the northern sky, selected at 178 MHz, and contains radio galaxies and radio quasars with redshifts up to z -- 2. The host galaxies of the lowredshift sources in the sample are identified with giant elliptical galaxies containing old stellar populations and, if the host galaxies of the high-redshift sources are also giant ellipticals, their stellar populations can be used to study the evolution of this class of galaxy with cosmic epoch and consequently to constrain models of galaxy formation and evolution. Lilly & Longair (1982, 1984) obtained infrared K magnitudes for an almost complete sample of 83 3CR galaxies with redshifts 0 < z < 1.6, and constructed the K magnitude versus redshift relation for these objects. The resulting relation showed remarkably little scatter and was interpreted as indicating that the 3CR host galaxies at redshift z - 1 are indeed giant elliptical galaxies. Lilly & Longair (1984) also showed that, unless the deceleration parameter were as large as q0 3.5, the shape of the K-z relation would not be consistent with non-evolving stellar populations, but that at least passive evolution is required. The K-z relation suggests that the host galaxies must have formed at large redshift, zf > 3; this age is consistent with the red colours of the infrared emission of some of these galaxies (Lilly 1989). Support for such an early formation epoch comes from a very deep spectrum of 3C 65, at redshift z = 1.176, which is well-matched using a stellar population of at least 4 Gyr in age (Stockton, Kellogg & Ridgway 1995), and also from the recent results of Dunlop et al. (1996). In 1987, McCarthy et al. and Chambers, Miley & van Breugel discovered that the optical emission of these powerful radio galaxies tends to be aligned along the axis of the radio source. Many models have been proposed to account for this alignment effect, the most promising being massive star formation induced by the passage of the radio jets (e.g. Rees 1989), scattering of light from an obscured active galactic nucleus (AGN) by electrons or by dust (Cimatti et al. 1996, 1997; Dey & Spinrad 1996, and references therein) and nebular continuum emission from warm line-emitting regions (Dickson et al. 1995). This complicates the use of these galaxies as cosmological probes. A number of observations have also suggested that a proportion of the K-band light of these galaxies may not be associated with starlight: (i) the discovery of a weaker alignment effect at nearinfrared wavelengths (e.g. Eisenhardt & Chokshi 1990; Rigler et al. 1992; Dunlop & Peacock 1993), which may be produced by the long-wavelength tail of a flat-spectrum component responsible for the optical alignment effect; (ii) the detection of broad Ha emission from the distant radio galaxy 3C 22 at z = 0.938 (Rawlings et al. 1995; Economou et al. 1995) and a reported detection of an unresolved central component in the K-band emission of 3C 65 at z = 1.176 (Lacy et al. 1995), suggesting that the infrared emission may contain light directly from the central AGN; (iii) the observation that infrared emission lines such as [S iii] 9532 can contribute a significant percentage of the K-band flux (Rawlings, Eales & Lacy 1991). In addition, the K-z relation of the 6C radio galaxies, which are roughly a factor of 5 lower in radio luminosity than the 3CR sources, tends to track that of the 3CR galaxies at low redshift, but at higher redshift the 6C galaxies tend to be fainter, lying closer to the `no evolution' line (Eales & Rawlings 1996; (...truncated)