Figure 4.12. Evolution of the comoving SFR density as a function of redshift including a compilation of results at z<6, estimates from the lensing-cluster survey of Richard et al. (2006) for the redshift ranges 6-10 and 8-10, and the values derived by Bouwens and collaborators from the Hubble Ultra-Deep Field (labelled "UDF"). Thick solid lines: SFR density obtained from integrating the LF of the first-category candidates of Richard et al. down to L1500 = 0.3L*z=3; dotted lines: the same but including also second-category candidates with a detection threshold of <2.5a in H. From Richard et al. (2006).
going from z ~ 4 to z ~ 6, whereas beyond this the results are quite controversial, as we will discuss; see e.g. a recent update by Hopkins (2006).
The properties of individual galaxies will be discussed in Section 22.214.171.124.
Going beyond redshift 7 requires the use of near-IR observations, as the Lya break of such objects moves out of the optical window. Given the different characteristics of such detectors and imagers (lower sensitivity and smaller field of view), progress has been less rapid than for lower-redshift observations.
In the NICMOS Ultra-Deep Field, Bouwens et al. (2004, 2006) have found 1-4 z-dropouts detected in J and K, compatible with redshift z ~ 7 starbursts. From this small number of objects and from the non-detection of J-dropouts by Bouwens et al. (2005) they deduce a low SFRD between z ~ 7 and 10, corresponding to a significant decrease of the SFRD with respect to lower redshift (see Figure 4.12, symbols labelled "UDF"). The properties of these and other z > 7 galaxies will be discussed below.
As an alternative to the "blank fields" usually chosen for "classical" deep surveys, the use of gravitational-lensing clusters - i.e. galaxy clusters acting as strong gravitational lenses for background sources - has over the last decade or so proven very efficient at finding distant galaxies (e.g. Hu et al. 2002; Kneib et al. 2004). Using this method, and applying the Lyman-break technique plus a selection for blue UV-rest-frame spectra (i.e. starbursts), our group has undertaken very-deep near-IR imaging of several clusters to search for z > 6-10 galaxy candidates (Schaerer et al. 2006). Thirteen candidates whose SEDs are compatible with that of star-forming galaxies at z > 6 have been found (Richard et al. 2006). After taking into account the detailed lensing geometry and sample incompleteness, correcting for false-positive detections, and assuming a fixed slope taken from observations at z > 3, their LF was computed. Within the errors the resulting LF is compatible with that of z > 3 Lyman-break galaxies. At low luminosities it is also compatible with the LF derived by Bouwens et al. (2006) for their sample of z > 6 candidates in the Hubble Ultra Deep Field and related fields. However, the turnover observed by these authors towards the bright end relative to the z > 3 LF is not observed in the Richard et al. sample. The UV SFRD at z > 6-10 determined from this LF is shown in Figure 4.12. These values indicate a similar SFRD to that for z > 3-6, in contrast to the drop found from the deep NICMOS fields (Bouwens et al. 2006).1 The origin of these differences concerning the LF and SFRD remains unknown. In any case, recent follow-up observations with the HST and Spitzer telescope undertaken to constrain the SEDs of these candidates better or to exclude some of them as intermediate-z contaminants show that data for the bulk of our candidates are compatible with their being truly at high-z (Schaerer et al. 2007a).
One of the main ways to clarify these differences is by improving the statistics, in particular by increasing the size (field of view) of the surveys. Both surveys of more lensing clusters and wide blank-field near-IR surveys, such as UKIDSS, are going on. The first z > 7 candidates have recently been found by UKIDSS (McLure 2007, private communication).
In this context it should also be remembered that not all optical dropout galaxies are at high-z, since a simple "dropout" criterion relies exclusively on a very red colour between two adjacent filters. As discussed for the ¿-dropouts above, extremely red objects (such as EROs) at z > 1-3 can be selected by such criteria. See Dunlop et al. (2007) and Schaerer et al. (2007b) for such examples. This warning is also of concern for searches for possible massive (evolved) galaxies at high redshift as undertaken by Mobasher et al. (2005) and McLure et al. (2006).
Let us now review the main properties of individual z > 6 LBGs, i.e. continuum-selected galaxies and discuss implications thereof. Lya emitters (LAEs), such as the z = 6.56 lensed galaxy found by Hu et al. (2002), have already been discussed (Section 4.4.2). Determinations of stellar populations (ages, SF history), extinction and related properties of such distant galaxies have really been possible only recently with the advent of the Spitzer space telescope providing sensitive-enough imaging at 3.6 and 4.5 |im. These wavelengths, longwards of the ^-band and hence not available for sensitive observations from the ground, correspond to the rest-frame optical domain, which is crucial to constrain properly stellar ages and stellar masses.
A triply lensed high-z galaxy magnified by a factor of >25 by the cluster Abell 2218 has been found by Kneib et al. (2004). Follow-up observations with the Spitzer telescope allowed the authors to constrain its SED up to 4.5 |im and revealed a significant Balmer
1 The SFRD values of Bouwens have been revised upwards, reducing the differences with our study (Hopkins 2007).
Abell 2218 KESR - all filters + spectroscopy
Abell 2218 KESR - all filters + spectroscopy
Was this article helpful?