RAPID DUST PRODUCTION IN SUBMILLIMETER GALAXIES AT z > 4?

The existence of submillimeter-selected galaxies (SMGs) at redshifts z > 4 has recently been confirmed. Simultaneously using all the available data from UV to radio, we have modeled the spectral energy distributions of the six known spectroscopically confirmed SMGs at z > 4. We find that their star formation rates (average similar to 2500 M-circle dot yr(-1)), stellar (similar to 3.6 x 10(11) M-circle dot) and dust (similar to 6.7 x 10(8) M-circle dot) masses, extinction (A(V) similar to 2.2 mag), and gas-to-dust ratios (similar to 60) are within the ranges for 1.7 < z < 3.6 SMGs. Our analysis suggests that infrared-to-radio luminosity ratios of SMGs do not change up to redshift similar to 5 and are lower by a factor of similar to 2.1 than the value corresponding to the local IR-radio correlation. However, we also find dissimilarities between z > 4 and lower-redshift SMGs. Those at z > 4 tend to be among the most star-forming, least massive, and hottest (similar to 60 K) SMGs and exhibit the highest fraction of stellar mass formed in the ongoing starburst (similar to 45%). This indicates that at z > 4 we see earlier stages of evolution of submillimeter-bright galaxies. Using the derived properties for z > 4 SMGs, we investigate the origin of dust at epochs less than 1.5Gyr after the big bang. This is significant to our understanding of the evolution of the early universe. For three z > 4 SMGs, asymptotic giant branch stars could be the dominant dust producers. However, for the remaining three only supernovae (SNe) are efficient and fast enough to be responsible for dust production, though requiring a very high dust yield per SN (0.15-0.65 M-circle dot). The required dust yields are lower if a top-heavy initial mass function or significant dust growth in the interstellar medium is assumed. We estimate lower limits of the contribution of SMGs to the cosmic star formation and stellar mass densities at z similar to 4-5 to be similar to 4% and similar to 1%, respectively.