Predictions for the Counts of Faint, High-Redshift Galaxies in the Mid-Infrared

Deep mid-infrared (MIR) observations could reveal a population of faint, high-redshift (z > 3), dusty starburst galaxies that are the progenitors of present-day spheroids or bulges and are beyond the reach of current instruments. We utilize a semianalytic galaxy formation scheme to find a range of models for the MIR galaxy counts, down to a flux level of a few nanojanskys. The models incorporate the formation of heavily dust-enshrouded stellar populations at high redshift and are consistent with existing observations, including faint counts at 1.6 (mu)m in the NICMOS Hubble Deep Field, and the upper limit on the extragalactic MIR background from TeV gamma rays. Our models predict ~0.04~0.4 galaxies arcsec-2 at the threshold of 100 nJy at 6 ~m, with a comparable or larger surface density at longer MIR wavelengths. We conclude that a significant new population of highredshift galaxies could be detected by the James Webb Space Telescope. Such a population would constitute background noise for the Terrestrial Planet Finder (TPF) and could necessitate repeat observations: every TPF resolution element would have a chance of up to ~10% of being contaminated by a background galaxy.