2006MNRAS.370.1185P

We present spectral energy distributions (SEDs), Spitzer colours, and infrared (IR) luminosities for 850-µm selected galaxies in the Great Observatories Origins Deep Survey Northern (GOODS-N) field. Using the deep Spitzer Legacy images and new data and reductions of the Very Large Array-Hubble Deep Field (VLA-HDF) radio data, we find statistically secure counterparts for 60 per cent (21/35) of our submillimetre (submm) sample, and identify tentative counterparts for another 12 objects. This is the largest sample of submm galaxies with statistically secure counterparts detected in the radio and with Spitzer. Half of the secure counterparts have spectroscopic redshifts, while the other half have photometric redshifts. We find that in most cases the 850-µm emission is dominated by a single 24-µm source, with a median flux density of 241µJy, leading to a median 24-to-850-µm flux density ratio of 0.040. A composite rest-frame SED shows that the submm sources peak at longer wavelengths than those of local ultraluminous infrared galaxies (ULIRGs). Using a basic grey-body model, 850-µm selected galaxies appear to be cooler than local ULIRGs of the same luminosity. This demonstrates the strong selection effects, both locally and at high redshift, which may lead to an incomplete census of the ULIRG population. The SEDs of submm galaxies are also different from those of their high-redshift neighbours, the near-IR selected BzK galaxies, whose mid-IR-to-radio SEDs are more like those of local ULIRGs. Using 24-µm, 850-µm and 1.4-GHz observations, we fit templates that span the mid-IR through radio to derive the integrated IR luminosity (LIR) of the submm galaxies and find a median value of L_IR(8-1000µm) = 6.0x10¹² L_☉. By themselves, 24-µm and radio fluxes are able to predict L_IR reasonably well because they are relatively insensitive to temperature. However, the submm flux by itself consistently overpredicts L_IR when using spectral templates which obey the local ULIRG temperature-luminosity relation. The shorter Spitzer wavelengths sample the stellar bump at the redshifts of the submm sources, and we find that the Spitzer photometry alone provides a model-independent estimate of the redshift, σ[Δz/(1 + z)] = 0.07. The median redshift for our secure submm counterparts is 2.0. Using X-ray and mid-IR data, only 5 per cent of our secure counterparts (1/21) show strong evidence for an active galactic nucleus dominating the LIR.