Annu. Rev. Astron. Astrophys. 2005. 43: xxx-xxx Copyright © 2005 by Annual Reviews. All rights reserved

5.4. Spitzer 24 µm Sources

A potential new way to find high-z LIRGs and ULIRGs appeared recently with the launch of the Spitzer observatory. Particularly suited to this goal is the 24 µm channel of the MIPS instrument. The confusion levels in the 70 and 160 µm prevent detection a significant number of high-redshift objects, and the IRAC 3.6 to 8 µm at high redshift probes mostly the old stellar component that is much weaker than the dust emission in starburst galaxies. At the time of writing, the observations are under way, and only a few results are available. Le Floc'h et al. (2004) give the first hint on the 24 µm selected galaxies. They couple deep 24 µm observations in the Lockman hole and extended groth strip with optical and near-infrared data to get both identification and redshift (either spectroscopic or photometric). They find a clear class of galaxies with redshift 1 z 2.5 and with luminosities greater than ~ 5 × 10¹¹ L (see also Lonsdale et al. 2004). These galaxies are rather red and massive with M > 2 × 10¹⁰ M (Caputi et al. 2005). Massive star-forming galaxies revealed at 2 z 3 by the 24 µm deep surveys are characterized by very high star formation rates - SFR 500 M year^-1. They are able to construct a mass of 10¹¹ M in a burst lifetime ( 0.1 Gyr). The 24 µm galaxy population also comprises sources with intermediate luminosities (10¹⁰ L_IR 10¹¹ L) and low to intermediate assembled stellar masses (10⁹ M 10¹¹ M) at z 0.8. At low redshifts, however, massive galaxies are also present, but appear to be building their stars quiescently in long timescales (Caputi et al. 2005). At these redshifts, the efficiency of the burst-like mode is limited to low mass M 10¹⁰ M galaxies. These results support a scenario where star-formation activity is differential with assembled stellar mass and redshift, and proceed very efficiently in massive galaxies (Caputi et al. 2005).

In the Lockman Hole, only one galaxy is associated with an X-ray source. This suggests that these galaxies are mostly dominating by star formation, consistent with the findings of Alonso-Herrero et al. (2004) and Caputi et al. (2005). This is also suggested by SEDs that are best fitted by PAH features rather than by strongly rising, AGN-type continua (Elbaz et al. 2005). The selected sources exhibit a rather wide range of MIPS to IRAC flux ratio and optical/near-infrared shapes, suggesting a possibly large diversity in the properties of infrared galaxies at high redshift as noticed by Yan et al. (2004b). Based on these first analyzes, together with the interpretation of the number counts (e.g., Lagache et al. 2004), it is clear that the 24 µm observations will provide the sample to unambiguously characterize the infrared galaxies up to z 2.5. They should fill the gap between the ISO- and SCUBA-selected galaxies.

Several 24 µm observations have been conducted on selected ERO and SCUBA and MAMBO samples. To our knowledge, LBGs have not been observed at long wavelengths. The MAMBO/SCUBA selected galaxies in the Lockman hole with radio identification have been observed by Spitzer and most of them detected between 3.6 and 24 µm. This allows to get an average SED for these (Egami et al. 2004; Ivison et al. 2004; see Figure 9) Spitzer deep surveys at 24 µm and shallow surveys like the SWIRE legacy (Lonsdale et al. 2004) can easily detect them and are thus a promising new way to find this class of high-z infrared galaxy. Nevertheless, the Early Release Observations from Spitzer have been used to extract their submillimeter flux from a stacking analysis of SCUBA observations in the Lockman hole (Serjeant et al. 2004). In this field, seven SMGs were already known and others were identified by further analysis. For the bulk of the 24 µm sources a marginal detection is found with an S₈₅₀ / S₂₄ ratio (1/20) much lower than that observed for SMGs. This clearly shows that the SMGs are only a fraction of the 24 µm sources, as expected. An interesting challenge is to find if Spitzer color criteria can be found to extract preferentially SMGs, i.e., the galaxies that account for most of the CIB near 1 mm. The SED in the thermal infrared appears quite variable for LIRGs and ULIRGs making this difficult (e.g., Armus et al. 2004).

Figure 9. Rest-frame SED of 15 SMGs (assuming a redshift of 3) with MAMBO and/or SCUBA, Spitzer/IRAC and Spitzer/MIPS 24 µm measurements. Purple diamonds are the galaxies 208, 119, 115, 48, 44 (Frayer et al. 2004), LE850_4, LE850_35 (Egami et al. 2004), and MMJ105201, MMJ105155, MMJ105203, MMJ105216, MMJ105148, MMJ105157, MMJ105207, MMJ105203 (Ivison et al. 2004). Overplotted are the SEDs of M82, normalized at 850 µm (from Chanial 2003), and the SED template of the Lagache et al. (2004) model, for L = 1013 L and a redshift of 3 (no normalization has been applied). Note that this sample of SMGs has a ratio dust/stellar component higher than the template or M82.

Extremely Red Objects (EROs) are usually selected based on their red colors: (R - K_s) 5.3 mag or (I - K_s) 4 mag. This color selection should include early-type galaxies at z ~ 1. However, the color selections are also sensitive to dust-reddened, star-forming systems. Up to now, it remains unclear what fraction of EROs are truly dust-obscured galaxies. Different scenarios of galaxy formation predict very different formation epochs for such galaxies. It is thus interesting to characterize these galaxies, in particular whether they belong to the early-type or dusty star-forming class of objects. Spitzer / MIPS 24 µm observations offer the first opportunity to address this issue because 24 µm observations can clearly discriminate between the two populations. In the N1 field, Yan et al. (2004a) suggest that about 50% of EROs are infrared luminous, dusty starbursts at z 1 (in a similar study, Wilson et al. (2004) show that at least 11% of 0.6 < z < 1.3 EROs and at least 22% of z > 1.3 EROs are dusty star-forming galaxies). Their mean 24 µm flux corresponds to infrared luminosities of about 3 × 10¹¹ and 10¹² L at z ~ 1 and z ~ 1.5, respectively. They are massive galaxies with lower limit M 5 × 10⁹ to 2 10¹⁰ M. The fraction of EROs likely to be AGN is small; about 15%. The link between the two classes of EROs could be that starburst EROs are experiencing, at z > 1, violent transformations to become massive early-type galaxies.