WELL-SAMPLED FAR-INFRARED SPECTRAL ENERGY DISTRIBUTIONS OFz∼ 2 GALAXIES: EVIDENCE FOR SCALED UP COOL GALAXIES

Abstract

We present an analysis of the far-infrared (FIR) spectral energy distributions (SEDs) of two massive K-selected galaxies at z = 2.122 and z = 2.024 detected at 24um, 70um, 160um by Spitzer, 250um, 350um, 500um by BLAST, and 870um by APEX. The large wavelength range of these observations and the availability of spectroscopic redshifts allow us to unambiguously identify the peak of the redshifted thermal emission from dust at ~ 300um. The SEDs of both galaxies are reasonably well fit by synthetic templates of local galaxies with L_IR ~ 10^{11} L_{sun} -- 10^{12} L_{sun} yet both galaxies have L_{IR} ~ 10^{13} L_{sun}. This suggests that these galaxies are not high redshift analogues of the Hyper-LIRGs/ULIRGs used in local templates, but are instead "scaled up" versions of local ULIRGs/LIRGs. For both galaxies the star formation rate determined from the best-fit FIR SEDs (SFR(L_{IR})) agrees with the SFR determined from the dust corrected H-alpha luminosity (SFR(H-alpha)) to within a factor of ~ 2; however, when the SFR of these galaxies is estimated using only the observed 24um flux and the standard luminosity-dependent template method (SFR(24um)), it systematically overestimates the SFR by as much as a factor of 6. A larger sample of 24 K-selected galaxies at z ~ 2.3 drawn from the Kriek et al. (2008) GNIRS sample shows the same trend between SFR(24um) and SFR(H-alpha). Using that sample we show that SFR(24um) and SFR(H-alpha) are in better agreement when SFR(24um) is estimated using the log average of local templates rather than selecting a single luminosity-dependent template, because this incorporates lower luminosity templates. The better agreement between SFRs from lower luminosity templates suggests that the FIR SEDs of the BLAST-detected galaxies may be typical for massive galaxies at z ~ 2, and that the majority are scaled up versions of lower luminosity local galaxies.