The detection of FIR emission from high-redshift star-forming galaxies in the ECDF-S

We have used the Large Apex Bolometer Camera Survey of the Extended Chandra Deep Field South to investigate rest-frame far-infrared (FIR) emission from typical star-forming systems (Lyman-break galaxies, LBGs, and Lyman alpha emitters, LAEs) at redshift 3, 4, 4.5 and 5 (922, 68, 46 and 20 sources, respectively). We initially concentrate on LBGs at z similar to 3 and select three sub-samples on stellar mass (rest-frame optical brightest, M-* > 10(10.25) M-circle dot), extinction-corrected star formation (assuming beta(UV) = -2 and applying a dust attenuation correction, SFRtot > 6.7 M-circle dot yr(-1)) and rest-frame UV magnitude (representing a typical Lyman-break selection with R < 24.43). We produce composite 870 mu m images of the typical source in our sub-samples, obtaining similar to 4 Sigma detections (0.61, 0.35 and 0.37 mJy, respectively) and suggesting a correlation between FIR luminosity and stellar mass. We apply a similar procedure to our full samples at z similar to 3, 4, 4.5 and 5 and do not obtain detections - a result that is consistent with a simple scaling between FIR luminosity and stellar mass. In order to constrain the FIR spectral energy distribution (SED) of these systems, we explore their emission at multiple wavelengths spanning the peak of dust emission at z similar to 3 using the Herschel Spectral and Photometric Imaging Receiver (SPIRE) observations of the field. We obtain detections at multiple wavelengths of both our stellar mass and UV-magnitude-selected samples, and find a best-fitting SED with dust temperatures in the similar to 33-41 K range. We calculate FIR luminosity, obscured star formation rates (SFRs) and dust masses and find that a significant fraction of star formation in these systems is obscured. Interestingly, our sample selected on extinction-corrected SFR does not display the large FIR fluxes predicted from its red UV spectral slope. This suggests that the method of assuming an intrinsic UV slope and correcting for dust attenuation may be invalid for this sample - and that these are not in fact the most actively star-forming systems. All of our z similar to 3 samples fall on the 'main sequence' of star-forming galaxies at z similar to 3 and our detected sub-samples are likely to represent the high obscuration end of the LBGs population at their epoch. We compare the FIR properties of our sub-samples with various other galaxy populations, finding that our stellar mass selected sample shows some similar FIR characteristics to Submillimeter Galaxies (SMGs) at the same epoch and therefore potentially represents the low FIR luminosity end of the high-redshift FIR luminosity function.