PDF《compact?》

2 deg2 COSMOS ?eld (McCracken et al. 2009). From this catalogue we ex- tracted all the objects with Ks(Vega) <

17.7, whose cor- responding BzK plot is shown in Fig. 1. From this sam- ple we then selected the brightest galaxies satisfying the passive BzK galaxy criterion of Daddi et al. (2004), for which the Ks(Vega) <

17.7 limit roughly corresponds to M? >

2.5*

1011 M⊙ at z ? 1.5 for a Chabrier IMF. This re- sulted in a sample of

22 sources, having retained only galax- Figure 1. The BzK two-colour diagram for all objects with Ks <

17.7 (Vega system) in the COSMOS ?eld. The

12 bona- ?de passive ETGs are shown as red circles. Blue squares with (z ? K) <

1.6 are most likely AGN dominated galaxies, hav- ing point-like morphology and/or X-ray emission. The remain- ing blue squares with redder (z ? K) colours show sBzK galax- ies and pBzK contaminants rejected for being star-forming. The blue solid line and the red dots in the bottom show the expected colours for the stars (from Lejeune et al. 1997), overlapping with the locus of real star sequences. ies with IRAC 3.6 ?m detections with separations <

0′′ .6 from the K-band positions, a criterion required to avoid ob- jects a?ected by blending in the IRAC and MIPS images (see, e.g. Daddi et al. 2007). We notice that

4 star-forming BzK galaxies (sBzK) are found to K <

17.7, after exclud- ing likely AGNs among sBzKs, i.e., the blue squares with (z ? K) <

1.6 in Fig. 1. The colours of these

4 objects are close to the pBzK boundary and in principle some genuine ETG could have been scattered there just by photometric errors noise. However, we ?nd that all

4 objects are well detected at

24 ?m. Since these

22 pBzK-selected objects could include heavily dust-reddened star-forming galaxies at z <

1.4, we carefully checked the properties of the sample to solidly iden- tify the bona-?de ETGs and discard possible contaminants. We requested all of the following criteria for retaining ETG candidates: i) non detection at

24 ?m in the Spitzer+MIPS data (Sanders et al. 2007). We used the most recent deep cata- logue by Aussel &

Le Floc'

h, reaching 5σ completeness levels down to 80?Jy (Le Floc'

h et al. 2009, submitted to ApJ). This corresponds to a limit of Star Formation Rate (SFR) <

50M⊙ yr?1 for z ? 1.5, using the models of Chary &

Elbaz (2001);

ii) elliptical like compact morphology, based on a visual in- spection of the ACS ICband images (Fig. 2);

iii) multicolour SEDs best ?tted with old, passively evolv- ing populations with no dust reddening as showing in Fig. 3.

3 SEDs were constructed using COSMOS photometry in the B-, i-, z-, J, Ks-band data consistently with McCracken et al (2009), expanded to include photometry in the ?rst

3 IRAC bands (from Ilbert, Salvato, &

S-COSMOS collabo- ration, Ilbert et al. (2009). We found that

12 of the

22 pBzK galaxies do satisfy all these criteria and are retained as our sample of bona-?de z 1.4 ETGs. The remaining

10 galaxies that were dis- carded are found to fail multiple criteria in most cases: they tend to be 24?m detected (6/10), have irregular (or point like, unresolved) morphologies in the ACS imaging (7/10), are best ?tted by star forming galaxy templates with large amounts of reddening (7/10), or show evidence for strong AGN contamination of the optical/near-IR light (3/10) as indicated by the '

power law'

shape of the SED and X-ray and/or radio detection. These criteria are very e?cient in se- lecting truly passively evolving ETGs, as con?rmed by their spectroscopic identi?cation in the HUDF by D05. Photometric redshifts and stellar masses were derived from the SEDs using the Hyperz code (Bolzonella et al. 2000), and ?tting photometric data with simple stellar pop- ulations (SSP) as well as with composite populations de- scribed by τ-models (SFR∝ e?t/τ , with τ = from 0.1 Gyr to ∞), and using........