2.8. Submm observations of known high-redshift galaxies and QSOs

The advent of SCUBA and MAMBO has also provided the opportunity to study the submm properties of large samples of interesting high-redshift galaxies, including almost all types of previously known distant galaxies. Isolated detections of high-redshift AGN-powered radio galaxies and QSOs were made in the mid 1990's using single-element bolometer detectors (for example Dunlop et al., 1994; Isaak et al., 1994); however, the compilation of statistical samples, and the secure rejection of contamination from fluctuating atmospheric noise have only been possible more recently, using SCUBA and MAMBO, and the 350-µm one-dimensional bolometer array SHARC at the 10.4-m aperture Caltech Submillimeter Observatory (CSO) on Mauna Kea. A key advantage of observing these sources is that both their redshifts and some of their astrophysical properties are already known, in contrast with the submm-selected galaxies discovered in blank-field surveys. Some of the targeted galaxies - very faint non-AGN radio galaxies, mid-IR-selected ISO galaxies, and X-ray selected AGNs - have only been detected very recently. As the relationship between these populations of galaxies and submm-selected galaxies is still unclear, many of the limits will be discussed in the context of following up submm surveys in Section 4.

Targeted surveys include a search for submm-wave continuum emission from high-redshift AGN-powered radio galaxies (Archibald et al., 2001), and observations of various samples of optically-selected QSOs (for example Carilli et al., 2001; Isaak et al., 2002). In these observations a single bolometer is aimed at the position of the target. While this does not lead to a fully-sampled image of the sky, it provides a more rapid measurement of the flux density at a chosen position. The results have been the detection and characterization of the dust emission spectra for a range of luminous high-redshift galaxies and QSOs, including APM 08279+5255 (Lewis et al., 1998), the galaxy with the greatest apparent luminosity in the Universe. Barvainis and Ivison (2002) have targeted all the known galaxies magnified into multiple images by the gravitational lensing effect of foreground galaxies from the CASTLES gravitational lens imaging project, ¹⁰ significantly expanding the list of high-redshift galaxies magnified by a foreground mass concentration with a submm detection. The SEDs of several of these galaxies are shown in Fig. 2.

Archibald et al. (2001) find evidence for significant evolution in the properties of dust emission with increasing redshift in a carefully selected sample of AGN radio galaxies, whose radio properties were chosen to be almost independent of the redshift of the observed galaxy. The results perhaps indicate that more intense star-formation activity, as traced by the submm emission, takes place alongside the radio source activity at higher redshifts, and so provide a possible clue to the formation and evolution of the massive elliptical galaxies thought to host radio galaxies. Hughes et al. (1997), Ivison et al. (1998b) and Omont et al. (2001) discuss the consequences of finding large masses of dust at high redshifts, in terms of the limited cosmic time available for the formation of the stars required to produce the metals and dust required to generate sufficiently intense submm emission from the host galaxy.

The radio galaxies detected by Archibald et al. (2001) in pointed single-bolometer SCUBA observations were followed up by imaging observations of the surrounding 5-arcmin² fields, to search for submm-loud companions. Ivison et al. (2000b) found that the surface density of submm galaxies in some of these fields is about an order of magnitude greater than that in a typical blank field, indicating a significant overdensity of sources. This is likely due to some radio galaxies being found in high-density regions of biased high-redshift galaxy formation, which are possibly `protoclusters' - rich clusters of galaxies in the process of formation.

A similar targeted approach has been taken to try to detect submm-wave emission from optically-selected LBGs at redshifts between 2.5 and 4.5 (Steidel et al., 1999). The Lyman-break technique (Steidel et al., 1996) detects the restframe 91.2-nm neutral hydrogen absorption break in the SED of a galaxy as it passes through several broad-band filters. Large samples of candidate LBGs can be gathered using multi-color optical images from 4-m class telescopes. The efficiency of the selection method is of order 70% after spectroscopic confirmation of the candidates using 8/10-m class telescopes. The LBGs are the largest sample of spectroscopically confirmed high-redshift galaxies, with a well-defined luminosity function (Adelberger and Steidel, 2000), a surface density of order 10 arcmin^-2, and inferred star-formation rates between 1 and 10 M yr^-1. They appear to be typical of the population of distant galaxies, and their spectra provide useful astrophysical information.

Observing the LBGs at submm wavelengths is an important goal, as an accurate determination of their submm-wave properties will investigate the link (if any) between the large well-studied LBG sample and the more enigmatic submm galaxy population (Blain et al., 1999c; Lilly et al., 1999; Adelberger and Steidel, 2000; Granato et al., 2001). At present, the typical very faint limits to the optical counterparts of the subset of submm galaxies with accurate positions (Smail et al., 1998a, 2002; Downes et al., 1999; Dannerbauer et al., 2002), and the detection of Extremely Red Object (ERO) galaxies (with R - K > 6) as counterparts to a significant fraction of submm galaxies (Smail et al., 1999, 2002; Gear et al., 2000; Frayer et al., 2002; Ivison et al., 2001; Lutz et al., 2001), argue against a large overlap between the two populations. The direct submm detection of LBGs using SCUBA has been largely unsuccessful at the 0.5-mJy RMS level: a single galaxy out of 16 was detected by Chapman et al. (2000; 2002c), while Webb et al. (2002b) describe a low significance of overlap between LBGs and SCUBA galaxies in a wide-field survey. The LBG cB58 at z = 2.72 (Ellingson et al., 1996; Frayer et al., 1997; Seitz et al., 1998; Pettini et al., 2000), which is magnified strongly (by a factor of 10-20) by a foreground cluster of galaxies at z = 0.37, and is at least ten times brighter than a typical LBG, was detected in the mm and submm by Baker et al. (2001) and van der Werf et al. (2002). However, after correcting for lensing, its 850-µm flux density is only about 0.1 mJy, below the level of confusion noise in SCUBA images, and similar to the flux density level of the statistical detection of high-redshift LBGs in the 850-µm SCUBA image of the HDF-N (Peacock et al., 2000). In the field surrounding an overdensity of LBGs at z = 3.09, Chapman et al. (2001a) were successful in detecting bright submm emission that appears to be associated with diffuse sources of Lyman- emission at the redshift of the overdensity, but were not included in the Lyman-break catalog. A key point to note is that the limits on submm-wave emission from LBGs are typically lower than expected if the relationship between UV spectral slope and far-IR luminosity observed for low-redshift low-luminosity starburst galaxies (Meurer et al., 1999) continues to high redshifts. Goldader et al. (2002) indicate that the relationship does not appear to hold for the most luminous galaxies.

The required sensitivity for successful submm observations of typical LBGs seems to be deeper than can be achieved using existing instruments. Observations using future very sensitive, high-resolution interferometers certainly ALMA, and perhaps CARMA and SMA, will shed more light on the submm-LBG connection.

The advent of the current generation of very sensitive X-ray observatories, Chandra and XMM-Newton, is generating a large sample of faint, hard X-ray sources, the luminosity of which is assumed to be dominated by high-redshift AGN (Fabian, 2000). Absorption and Compton scattering in large column densities of gas preferentially depletes soft X-rays, hardening the X-ray SEDs of gas-rich AGN. Such a population of hard, absorbed X-ray sources is required in order to account for the cosmic X-ray background radiation spectrum, which is harder than the typical SEDs of individual low-redshift AGN (Fabian and Barcons, 1992; Hasinger et al., 1996). Observations of the limited areas of the sky where both submm and X-ray data are available (Fabian et al., 1999; Hornschemeier et al., 2000; Mushotzky et al., 2000; Almaini et al., 2002) have tended to show little direct overlap between the X-ray and submm galaxies, although there are examples of X-ray-detected submm-wave galaxies (Bautz et al., 2000). The combined results of Bautz et al. and Fabian et al. reveal that 2 out of 9 SCUBA galaxies are detected by Chandra. In the larger-area brighter 8-mJy survey, Almaini et al. (2001) identify only 1 out of 17 SCUBA galaxies using Chandra. Page et al. (2002) discuss further the submm properties of X-ray sources. Perhaps of order 10% of known submm galaxies have faint hard X-ray counterparts that would be typical of dust-enshrouded AGN. There is also a statistical detection of excess submm-wave emission from the positions of faint high-redshift hard X-ray sources (Barger et al., 2001) and a positive submm-X-ray galaxy correlation function (Almaini et al., 2002). The lack of strong X-ray emission from a majority of submm galaxies lends circumstantial support to the idea that much of their luminosity is derived from star formation and not from AGN accretion. However, some submm galaxies may have hydrogen column densities, and thus optical depths to Compton scattering, that are sufficiently great to obscure soft X-ray radiation entirely (> 10²⁴ cm^-2). Even if they contained powerful AGN, these submm galaxies would be very faint in Chandra surveys, which reach detection limits of order 10^-17 erg cm^-2 s^-1 at soft 0.5-2 keV wavelengths (Giacconi et al., 2002). They may be found in very deep observations using the the greater collecting area of XMM-Newton for hard X-ray photons. However, note that the 15-arcsec resolution of XMM-Newton leads to confusion due to unresolved faint sources in the beam that is likely to impose a practical limit of order 10^-15 erg cm^-2 s^-1 to the depth of a survey in the hard X-ray 2-8 keV band (Barcons et al., 2002). Deconvolution of the images from joint XMM-Newton/Chandra deep fields, exploiting the sub-arcsec positional information from Chandra, will perhaps allow this limit to be exceeded.

Finally, galaxies detected in far-IR surveys using ISO (for example Puget et al., 1999) out to redshifts z ~ 1 have been targeted for SCUBA submm observations (Scott et al., 2000). The large arcmin-scale observing beam in 170-µm ISO surveys makes identification of submm counterparts difficult, but progress has been made by combining sub-arcsec resolution deep radio images. The results include some sources with apparently rather cool dust temperatures of order 30 K (Chapman et al., 2002d), and are generally consistent with redshifts less than unity and dust temperatures of less than about 50 K for most of the galaxies.

¹⁰ http://cfa-www.harvard.edu/castles. Back.