Introduction

Over the past four years, deep sub-millimetre surveys carried out using SCUBA on the JCMT (Smail et al. 1997, Hughes et al. 1998, Barger et al. 1998 & 1999, Blain et al. 1999, Eales et al. 1999 & 2000), have revealed the extreme importance of dust in the determination of the global star-formation history of the Universe. Observations at optical/UV wavelengths have implied that the star-formation rate (SFR) rises steeply as a function of redshift between z=0 and z=1 (Lilly et al. 1996), peaking at 1 < z < 1.5, and declining to values comparable to the present day at ~z=4 (Madau et al. 1996). However in heavily dust-enshrouded star-forming regions, much of the optical/UV radiation emitted by the young stars is absorbed, leading to the possibility that a significant amount of star formation, particularly at high-redshift, may have been missed in these wavebands.

In order to quantify the star-formation density contributed by such highly-obscured objects, we must observe directly the rest-frame far-infrared (FIR) emission from the heated dust, which at redshifts greater than z~1 is shifted into the sub-millimetre waveband. Furthermore, the steep spectral index of the thermal emission longward of the peak at ~100um results in a sufficiently large negative K-correction that the dimming effect of increasing cosmological distance is effectively cancelled over a wide range in redshift. Consequently, for a galaxy with fixed intrinsic FIR luminosity we would expect to observe approximately the same 850 um flux-density for an object at z=8 as at z=1.

The SCUBA 8-mJy Survey at 850 um

The `SCUBA 8-mJy Survey' (Scott et al. 2001, Fox et al. 2001, Lutz et al. 2001, Almaini et al. 2001, Ivison et al. 2001) is a wider-area, somewhat shallower survey than its earlier counterparts, covering a total area of 260 arcmin², evenly split over two regions of sky (the ELAIS N2 and Lockman Hole East - figs. 1 and 2 respectively), and reaching a typical rms noise level of sigma(850)~2.5 mJy/beam. This survey has three key advantages over previous sub-millimetre surveys. Firstly, the relatively bright flux-density limit facilitates follow-up with existing instrumentation which will eventually yield an accurate astrometric position (via the VLA or IRAM PdB) for every source. Secondly, the lower source-density at ~8 mJy means that source confusion is much less of a problem. Finally, this survey is optimally designed for the detection of the most luminous starburst galaxies at high redshift, with SFRs > 1000 Mo yr^-1. Our extensive follow-up programme with thus allow us to test the suggestion made by several authors (e.g. Dunlop 2001) that such objects are the progenitors of the present-day massive-elliptical population.

We have used a maximum-likelihood technique to measure the statistical significance of each peak in our maps, detecting 19 sources with S/N > 4, 38 with S/N > 3.5, and 72 with S/N > 3 (Scott et al. 2001). Based on these detections and extensive simulations our best estimate of the cumulative source count at S₈₅₀ > 8 mJy is 320⁸⁰_-100 deg^-2, consistent with a range of models involving strong evolution of the dust-enshrouded starburst population. These objects have inferred star-formation rates in excess of 1000 Mo^-1, enshrouded in 10⁸-10⁹ Mo of dust. Assuming that the majority of sources lie at z > 1.5, this result implies that the co-moving number density of high-redshift galaxies forming stars at a rate in excess of 1000 Mo yr^-1 is 10^-5 Mpc^-3, with only a weak dependence on the precise redshift distribution. This is a very interesting number, since it corresponds to the number density of massive ellipticals with L > 3-4 L_* in the present-day universe, and is also the same as the co-moving number density of comparably massive, passively-evolving objects in the redshift band 1 < z < 2 inferred from recent surveys of extremely red objects (EROs). Thus while the brightest sub-millimetre sources uncovered by this survey contribute only ~10% of the sub-millimetre background, they can plausibly account for the formation of all present-day massive elliptical galaxies. A key test of this picture is to determine whether the bright SCUBA population is at least as strongly clustered as are the EROs at intermediate redshifts (Daddi et al. 2001). A first attempt at such a test has proved inconclusive due to the small numbers of significant sources (>3.5 sigma) detected in our survey areas, but there are some initial indications of clustering on scales of 1-2 arcminutes, particularly in the ELAIS N2 field.

Multiwavelength Follow-up

Both the Lockman Hole East and ELAIS N2 have a vast quantity of multi-wavelength data available for follow-up studies. The faintness of the SCUBA population means that spectroscopically-determined redshifts remain a distant goal for all but a handful of the 850 um sources, but we can use the radio-to-infrared spectral energy distribution (SED) to constrain the allowed redshift range. It is then possible to identify potential optical/IR counterparts consistent with both the position of the SCUBA source and with the SED-based redshift constraints and through deep radio or millimetre interferometry establish the correct identification (if any.

In Fox et al. (2001) we consider 19 of our most significant sources in this way. Exploiting the parallel SCUBA 450um observations, in combination with radio and ISO data, our derived SED-based redshift constraints suggest that all of these sources are at z > 1, with at least 50% at z > 2. Deep K-band imaging currently exists for 8 of these 850um sources, of which 5 have revealed a potential ERO counterpart to the SCUBA source. In the case of our most significant Lockman Hole East source, Lockman850.1 (Lutz et al. 2001), an accurate astrometric position yielded by the IRAM PdB interferometer has allowed us to make a solid K-band identification. The near-infrared counterpart (fig. 3) is an extended (20-30 kpc), clumpy, and extremely red object (I - K > 6.1). The SED suggests it to be a dusty star forming object at a redshift of about 3 (2-4). Its SFR and near-infrared properties are consistent with Lockman850.1 being a massive elliptical in formation.

Over the next two years we aim to obtain comparably detailed redshift and morphological information on a substantial fraction of our brightest sources. In particular, very deep VLA imaging is already providing greatly-improved redshift constraints (shortly to be published by Ivison et al. 2001), while deep near-infrared imaging of our sources with Gemini is currently scheduled as the top-ranked UK proposal with NIRI on Gemini for the current semester.

The relationship of SCUBA and Chandra sources

The relationship between the hard X-ray and sub-millimetre populations is explored in Almaini et al. (2001) using deep Chandra observations of the ELAIS N2 field. In agreement with other recent findings (eg Barger et al. 2001), we confirm that the direct coincidence is small; 1/17 SCUBA sources with S/N > 3.5, or 2/36 with S/N > 3 having a significant X-ray detection consistent with active galactic nucleii (AGNs). In one other case where sub-arcsecond VLA and K-band positions are available, a weak X-ray flux is measurable but consistent with starburst activity. Unless the central engine is obscured by Compton-thick material with a low (< 1%) scattered component, this implies that the majority of SCUBA sources are not powered by AGN. Furthermore, the detection of only ~5% of the SCUBA sources by Chandra would require that the typical obscuration is almost isotropic, inconsistent with a unified AGN scheme.

References

Almaini, O., et al., 2001, MNRAS, submitted (astro-ph/0108400).
Barger, A.J., et al., 1998, Nat, 394, 248.
Barger, A.J., et al., 1999, ApJ, 518, L5.
Barger, A.J., et al., 2001, AJ, submitted (astro-ph/0106219).
Blain, A.W., et al., 1999, ApJ, 512, L87.
Daddi, E., et al., 2000, A&A, 361, 535.
Dunlop, J.S., 2001, in: Deep Sub-millimetre Surveys, eds. Lowenthal, J. & Hughes, D.H., World Scientific, in press (astro-ph/0011077).
Eales, S.A., et al., 1999, ApJ, 515, 518.
Eales, S.A., et al., 2000, AJ, 120, 2244.
Fox, M.J., et al., 2001, MNRAS, submitted (astro-ph/0107585).
Hughes, D.H., et al., 1998, Nat, 394, 241.
Ivison, R.J., et al., 2001, in prep.
Lilly, S.J., et al., 1996, ApJ, 460, L1.
Lutz, D., et al., 2001, A&A, submitted (astro-ph/0108131).
Madau, P., et al., 1996, MNRAS, 283, 1388.
Smail, I., 1997, ApJ, 490, L5.
Scott, S.E., et al., 2001, MNRAS, submitted (astro-ph/0107446).

Figure 1: The 850um image of the ELAIS N2 field, smoothed with a beam-size Gaussian (14.5 arcsec FWHM). The numbered circles highlight those sources found at a significance of >3.5sigma in order of decreasing significance.

The 850um image of the Lockman Hole East field, smoothed with a beam-size Gaussian (14.5 arcsec FWHM). The numbered circles highlight those sources found at a significance of >3.5sigma in order of decreasing significance.

The figure shows a 30 x 30 arcsec region of a deep (6-hour integration on UKIRT + UFTI) K-band image of the field in the vicinity of the S_850umum = 10.5 mJy source Lockman850.1. The position of the SCUBA source is marked by the large (10-arcsec diameter) circle, while the position of the 1.3mm source detected in follow-up observations with the IRAM PdB interferometer is marked by the small (2-arcsec diameter) circle (Lutz et al. 2001). The SCUBA/IRAM source can be unambiguously identified with the brightest of a group of (apparently connected) compact peaks with K~21.4. All of these faint peaks are too red to be detected in the deepest available I-band image of this field reaching I~27.5.
 

Click here for printable version.