Annu. Rev. Astron. Astrophys. 2000. 38: 761-814 Copyright © 2000 by Annual Reviews. All rights reserved

4.2. Far-Infrared Surveys

The SED of actively star-forming galaxies peaks at ~ 60-80 µm. For observations at 200 µm, the "negative" k-correction is advantageous for detecting high-redshift galaxies; at the same time, such measurements give the total luminosity directly, without a model-dependent bolometric correction. For these reasons, several surveys were undertaken with the ISOPHOT at 175 µm. A 1-square-degree field in the Lockman hole was surveyed as part of the Japanese ISO cosmology project (Kawara et al 1998, 1999; see Figure 16, see color insert). Two fields in the southern Marano area and two fields in the northern ELAIS fields with a combined area of 4 square degrees constitute the FIRBACK program (Puget et al 1999, Dole et al 1999). An additional "serendipity" survey used the slews between pointed observations to survey the sky at ~ 200 µm at the 2 Jy level (Stickel et al 1998, 1999). This survey extends IRAS far-IR SEDs to longer wavelengths for a few thousand nearby galaxies. The depth of the two deep surveys is ~ 100 mJy (5). The key problems for the far-IR detection of faint individual sources are detector noise and large, spatially varying backgrounds: Galactic cirrus, extragalactic backgrounds, and (to a lesser extent) zodiacal dust emission. To minimize the impact of cirrus, the deep survey areas were chosen in regions known to have low HI-column densities and IRAS 100 µm surface brightness. In the best areas, the surveys are limited by the fluctuations in the extragalactic background (Lagache & Puget 1999). In the case of the Lockman hole and ELAIS regions, a comparison with a second map at 90 µm gives additional useful information. In the final source lists, only detections with flux densities 120 mJy were taken into account. At that level, reliability and completeness are good and far above the expected fluctuations in the cirrus and extragalactic backgrounds. The sources detected, therefore, are likely individual galaxies. At the 100 mJy cutoff level, Kawara et al reported 45 175-µm sources (and 36 95-µm sources) in the Lockman field, and Dole et al found 208 175-µm sources in the combined Marano and ELAIS fields. In a 0.25-square-degree Marano field, Puget et al detected 22 sources. Figure 16 shows the counts from all three references, in comparison with models A and E of Guiderdoni et al (1998). At the lowest flux densities, the number counts are a factor of 4 to 10 above the predictions of non-evolving models. The counts are consistent with those that include strong evolution as well as a significant population of z 1 ULIRGs (e.g. model E of Guiderdoni et al 1998).

Figure 16. Deep far-infrared surveys. Left: ISOPHOT 90 µm (top) and 175 µm (bottom) maps of a 44' × 44' field LHNW in the Lockman hole (Kawara et al 1998). Right: Integrated 175 µm counts from the FIRBACK (Marano and ELAIS fields) (diamonds: Pug et al 1999, filled circles: Dole et al 1999) and Lockman field (green asterisk, Kawara et al 1998). The blue line is the model 'E' of Gideroni et al (1998) that includes a strong ULIRG component at high z. The red dotted curve is model 'A' of Guiderdoni et al (1998) that does include strong source evolutuion but does not include ULIRGs (from Dole et al 1999).

The integrated counts are an order of magnitude greater than the 100 µm IRAS counts at the same flux level. In the ELAIS area, 31 sources have a 90 µm counterpart. The average 175/90 flux density ratio is 2, as compared with ~ 1 for a sample of local sources (Stickel et al 1998). One part of the Marano ISOPHOT field overlaps with the deep ISOCAM survey area. At most, 50% of the ISOPHOT sources have 15 µm counterparts (Elbaz et al 2000). For those, the 175 µm / 15 µm flux density ratios correspond to what one would expect for relatively nearby (z ~ 0.5) galaxies with Arp 220-like spectra. Scott et al (1999) observed ten FIRBACK sources with SCUBA. The average 850 (and 450) µm / 175 µm flux density ratios also favor a low mean redshift (~ 0.3) although the degenerate dust temperature/redshift relation would also permit in principle solutions with z up to 2. The far-IR sources not detected by ISOCAM could be at higher redshift. The interplay of the k-corrections is such that ISOCAM cannot see galaxies with an Arp 220 SED located at z ~ 2.

The integrated counts constitute less than 10% of the COBE far-IR background (Hauser et al 1998). This suggests that ISOPHOT is detecting the tip of the iceberg of a new population of very luminous, moderate to high redshift galaxies. Most of the background is in sources with 175 µm flux densities between a few and 100 mJy.