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

4.6. Mid-IR Sources

Mid-IR observations offer the opportunity to detect emission from material heated by star formation or active galactic nuclear (AGN) activity, and thus provide some indication of radiative activity that may be obscured at UV through near-IR wavelengths. The launch of the Infrared Space Observatory (ISO), the first space-based mid-IR telescope capable of such observations, coincided nicely with the availability of the HDF data set, and both deep fields (N and S) were the targets of deep 6.75 µm and 15 µm observations with the ISOCAM imager. The HDF-S ISO observations were made before the HST imaging was carried out, and they are currently being analyzed (e.g., [Oliver et al. 2000]).

The HDF-N ISO observations were carried out using discretionary time allocated by the ISO director to an observing team led by Rowan-Robinson [Serjeant et al. 1997, Goldschmidt et al. 1997, Oliver et al. 1997, Mann et al. 1997, Rowan-Robinson et al. 1997]. Two other groups have subsequently processed and reanalyzed the ISO data using independent methods [Désert et al. 1999, Aussel et al. 1999a]. The 15 µm maps extend well beyond the central HDF to cover a portion of the flanking fields, and have 9" resolution. The high-sensitivity region of the 6.7 µm maps is smaller and roughly matched to the central HDF WFPC2 area, with 4" resolution.

The depth of the ISO image mosaics varies over the field of view, and the data processing procedures are complex and the subject of continuing refinement. Extensive simulations have been used by the three groups to calibrate source detection, completeness and reliability, and photometry. Given these complexities, it is perhaps not unexpected that the groups have produced different source catalogs. A simple, visual inspection of the images presented by [Serjeant et al. 1997] or [Aussel et al. 1999a] demonstrates that the 15 µm data is much more straightforward to interpret - at least ~ 20 sources are readily visible to the eye, whereas the 6.7 µm image has only a few "obvious" sources and much more low-level background structure. Indeed, the source catalogs from the three groups agree reasonably well at 15 µm, commonly detecting most of the brighter sources (~ 20 detections with 200 µJy) with generally similar measured flux densities. Aussel et al and Desert et al, however, claim reliable source detection to fainter flux limits than do Goldschmidt et al, and thus have longer source lists: The complete plus supplemental catalog of Aussel et al. lists 93 sources at 15 µm. At 6.75 µm, the situation is more confused (perhaps literally). All three groups define "complete" lists of six or seven sources, but they are not all the same objects. Differences in the assumptions about ISO astrometry and geometric distortion corrections may be partially responsible. Goldschmidt et al. provide a list of 20 supplemental sources at 6.7 µm, only a few of which are detected by the other groups. There also appear to be systematic differences in the derived source fluxes at this wavelength. Altogether, the 15 µm source detections and fluxes appear to be well established, but only the few brightest 6.7 - µm detections are unambiguous, and the associated fluxes may be uncertain.

The optical counterparts of the HDF-N ISO sources are mostly brighter galaxies at 0 < z < 1.2, although occasionally the identifications may be confused because of the 9 PSF at 15 µm. Many of the 15 µm sources also correspond to radio sources (few or none, however, correspond to the SCUBA sources discussed below). Over this redshift range, the 15 µm measurements primarily sample emission from the "unidentified emission bands," including the strong features at 6.2 - 8.6 µm. These features are considered a characteristic signature of star-forming galaxies (see, e.g., [Rigopoulou et al. 1999]). At z > 1.4, they shift out of the 15 µm ISO bandpass, and therefore more distant objects should become much fainter. Indeed none have been identified yet in the ISO/HDF. Warm dust (T > 150K) may also contribute to the mid-IR emission, and some of the ISO sources are almost certainly AGN (e.g., HDF 2-251, a red early-type radio galaxy at z = 0.960, whose mid-IR emission is very likely not powered by star formation). Most of the HDF/ISO galaxies do not appear to be AGN, however, and if powered by reprocessed stellar UV radiation have obscured star formation rates that greatly exceed those derived from their UV/optical emission [Rowan-Robinson et al. 1997]. Assuming an M82-like spectrum to derive k-corrections, the average HDF/ISO source is roughly 10× more luminous than M82 at 15 µm [Elbaz et al. 1999]. [Rowan-Robinson 1999] estimates that roughly two thirds of the star-formation at z ~ 0.5 is in dust-enshrouded starbursts such as those detected by ISO.