7. FROM ISO TO SPITZER, HERSCHEL, THE JWST AND ALMA

On August 23rd, 2003, NASA's Spitzer space telescope (formerly SIRTF) was launched. Among its first results came the source counts at 24 µm down to ~ 20µJy which confirmed what ISO deep surveys already saw: a strong excess of faint sources indicating a rapid redshift evolution of IR luminous galaxies. When compared to models developped to fit the ISO counts, the faint end of the Spitzer counts are perfectly fitted as shown in the Fig. 17 reproduced from Chary et al. (2004). On the high flux density range, around 1 mJy and above, less galaxies are found than predicted by those models (see also Papovich et al. 2004). This is partly, if not integrally, due to the fact that even ISOCAM-15 µm number counts were initially overestimated above S₁₅ ~ 1 mJy as discussed in Sect. 5.1, since the data reduction techniques were not optimized for the surveys with little redundancy over a given sky pixel. Although some refinement of the template SEDs used in the models might be considered (as suggested by Lagache et al. 2004), the 24 µm number counts appear to be perfectly consistent with the up-to-date 15 µm counts. Moreover, the high flux density range does not strongly affect the conclusions of the models based on previous 15 µm counts since bright objects do not contribute significantly to the CIRB and to the cosmic density of star formation. Hence these refinements are not strongly affecting the conclusions derived on galaxy formation and evolution based on the ISO deep surveys and summarized in the previous sections (see also Dole et al. 2004 for the Spitzer counts in the far IR).

Figure 17. Completeness corrected galaxy counts in the MIPS 24 µm channel from Spitzer observations of the ELAIS-N1 field (from Chary et al. 2004). The error bars reflect the Poissonian uncertainty. The horizontal bars represent the minimum and maximum flux density in that bin. The lines show four models for 24 µm counts: King & Rowan-Robinson (2003, KRR), Xu et al. (2001, Xu), Chary & Elbaz (2001, CE), Lagache, Dole & Puget (2003, LDP). The symbols are plotted at the average of the flux densities of the detected sources in that bin for the data while the lines are plotted at the counts-weighted flux average for the models. The lower plot in the figure shows the histogram of the actual number of sources detected in each flux bin without any completeness correction.

Another hint on the consistency of Spitzer MIPS-24 µm surveys with ISOCAM-15 µm is given by the comparison of the images themselves. Galaxies detected at 15 and 24 µm are clearly visible in both images in Fig. 18, although several 24 µm sources do not have a 15 µm counterpart. This results from the combination of the better sensitivity of MIPS, by a factor 2 or slightly more for the deepest surveys (galaxies are detected down to S₁₅ ~ 40 µJy in the HDFN, see Aussel et al. 1999, without lensing magnification), and of the k-correction. For galaxies above z ~ 1, the PAH bump centered on the PAH feature at 7.7 µm starts to exit to 15 µm broadband while it remains within the MIPS-24 µm band up to z ~ 2. The combination of ISOCAM and MIPS can be used to test whether the library of template SEDs that were used to derive "bolometric" IR luminosities from 8 to 1000 µm for the 15 µm galaxies are correct at least in the mid IR range. One of the most important test is to check whether the 7.7 µm PAH bump is still present at z ~ 1 and whether the 24 over 15 µm flux density ratio is consistent with the template SEDs used by the models from which star formation histories were derived. The template SEDs designed by Chary & Elbaz (2001) provide a very good fit to the combination of both mid IR values for a sample of 16 galaxies detected by ISOCAM and MIPS (crosses on the Fig. 18). The L_IR(8-1000 µm ) derived from either the 15 or 24 µm luminosities or the combination of both to constrain the SED fit present a 1- dispersion of only 20% (Elbaz et al. 2004). For galaxies located around z ~ 1, the relative 15 and 24 µm luminosities clearly suggest the presence of a bump at 7.7 µm as observed in nearby galaxies and due to PAHs.

Figure 18. ISOCAM 15 µm image (left) of the Ultra-Deep Survey in the Marano FIRBACK field (depth 140µJy, 80% completeness) versus Spitzer MIPS-24 µm image (right; depth 110µJy, 80% completeness). The crosses identify 16 galaxies detected at 15 and 24 µm for which VLT-FORS2 spectra were obtained and which SEDs were fitted in Elbaz et al. (2004).