Annu. Rev. Astron. Astrophys. 2000. 38: 667-715
Copyright © 2000 by . All rights reserved

Next Contents Previous

5.4. Obscured populations

ISO and SCUBA observations of the HDF and other fields have revealed an energetically important population of dust-obscured objects. Interpretation of these results has been the subject of considerable debate, because of ambiguities in source identification and in distinguishing starbursts from AGN. The ISOCAM (6.7 and 15 µm) and SCUBA (850 µm) observations bracket but do not sample the wavelength regime 100 < lambdap < 200 µm near the peak of the far-IR emission, making it difficult to assess reliably the source contribution to the global emissive energy budget of galaxies. Although strong mid-IR emission accompanies vigorous star formation in many nearby galaxies, the unidentified IR emission bands carry most of the energy in the wavelength range sampled by ISOCAM at z ~ 1. The bulk of re-emitted radiation, however, emerges near lambdap, and there is considerable diversity in f (10 µm) / f(100 µm) flux ratios among nearby luminous and ultraluminous IR galaxies. Therefore, deriving star formation rates from mid-IR measurements alone requires a substantial extrapolation and is quite uncertain. SCUBA 850 µm observations sample the re-radiated thermal emission directly, but at a wavelength well past lambdap, again requiring an extrapolation to total far-IR luminosities assuming a dust temperature and emissivity that are almost never well constrained by actual, multi-wavelength measurements. Furthermore, SCUBA and ISO do not in general detect the same sources. ISO detects objects out to z ~ 1, whereas most of the SCUBA sources could be at much higher redshifts. For both mid-IR and sub-millimeter sources, AGN-heated dust may also play a role. Several active galaxies (including radio ellipticals) in the HDF are detected by ISO, and some SCUBA sources (not yet in the HDF, however) have been identified with AGN. Chandra x-ray observations [Hornschemeier et al. 2000] do not detect with high significance any of the sub-millimeter sources in the HDF.

Difficulties in identifying the optical counterparts to the mid-IR and sub-millimeter sources are a second source of ambiguity. The mean separation between galaxies in the HST images of the HDF is about 3". In comparison the ISO 15 µm PSF has full-width at half max (FWHM) approx 9", whereas the SCUBA 850 µm beam size is approx 15". Nevertheless, the HDF mid-IR sources have plausible counterparts among the brighter galaxies in the survey. SCUBA sources, on the contrary, often seem to have have extremely faint, and sometimes entirely invisible, counterparts in the optical and near-infrared. Sub-mm objects sometimes correspond to microjansky radio sources at centimeter wavelengths, and sometimes have faint, very red near-IR counterparts. Many, however, do not, even with the deepest radio and near-IR data available with current instrumentation (see < ahref="Ferguson4_7.html">Section 4.7).

Other than the occasional ultra-red optical counterpart (of which there are none in the HDF-N), the non-AGN counterparts to sub-millimeter and far-IR sources rarely have particularly unusual photometric or spectral features which highlight them as the remarkable objects they must be. Counterparts to 6.7 and 15 µm ISO sources show a predominance of "post-starburst" Balmer absorption spectra ([Flores et al. 1999a, Flores et al. 1999b, Aussel et al. 1999b, Cohen et al. 2000]); apart from the occasional AGN, few have strong emission line spectra, which suggests the most intense star-forming regions are highly obscured. There are hints that the Lyman-break galaxies represent the faint tail of the sub-millimeter source population. [Chapman et al. 1999] detect one out of 18 galaxies in a targeted study of z ~ 3 Lyman-break galaxies. [Peacock et al. 1999] find a significant statistical correlation of Lyman-break galaxies with sky fluctuations in the HDF-N SCUBA map. A straightforward analysis suggests that the ratio of hidden star formation to star formation directly measured in the UV is about 6:1, and that Lyman-break galaxies account for at least 25% of the 850 µm background.

Given the present status of faint sub-millimeter and mid-IR surveys, the most telling information comes not from individual source identifications but from the ensemble statistics using the combined data from many surveys. The ISO 15 µm number counts show good agreement between various surveys and a strong excess over no-evolution models ([Elbaz et al. 1999]), as do 850 µm SCUBA counts ([Barger et al. 1999]; [Blain et al. 1999]). As with the "faint blue galaxy" excess [Kron 1980] that helped spark the boom in the optical study of distant field galaxies in the 1970s and 1980s, the ISO and SCUBA counts point toward strong cosmological evolution, probably manifesting the star formation history of the galaxy population. And like the faint blue galaxy problem, the robust interpretation of this number count excess will undoubtedly require extensive follow-up observations to characterize the source population and its properties. Currently, given the uncertainties in source identification, in extrapolating the current measurements to "true" far-IR luminosities, and in balancing the roles of AGN vs. star formation, it seems premature to attempt a detailed revision of our picture of galaxy evolution and cosmic star formation. But it is also abundantly clear that a true understanding will have to account for the important and perhaps dominant energetic role played by this obscured population.

Next Contents Previous