2008ApJ...686..251D

Observations at submillimeter wavelengths have revealed a population of sources thought to be at relatively large redshifts. The position of the 850 µm passband on the Rayleigh-Jeans portion of the Planck function leads to a maximum redshift estimate of z∼4.5, since sources will not retain their redshift-independent brightness close to the peak of the Planck function and thus drop out of surveys. Here we review evidence that ice absorption is present in the spectra of local ultraluminous infrared galaxies, which are often taken as analogs for the 850 µm source population. We consider the implication of this absorption for ice-induced spectral structure at far-infrared wavelengths and present marginal astronomical evidence that amorphous ice may have a feature similar to crystalline ice near 150 µm. Recent corroborative laboratory evidence is supportive of this conclusion. It is argued that early metal enrichment by pair-instability supernovae may lead to a high ice content relative to refractory dust at high redshift, and a fairly robust detection of ice emission in a z=6.42 quasar is presented. It is further shown that ice emission is needed to understand the 450 µm sources observed in the GOODS-N field. We are thus encouraged to apply far-infrared ice emission models to the available observations of HDF 850.1, the brightest submillimeter source in the Hubble Deep Field. We suggest that a redshift as large as 13 may need to be considered for this source, nearly a factor of 3 above the usual top estimate. Inclusion of the possibility of far-infrared ice emission in the spectral energy distributions of model sources generally broadens the range of redshifts to be considered for submillimeter sources compared to models without ice emission.