Among the variety of objects discovered so far at high-redshift, a special class is represented by the so called Extremely Red Objects (EROs hereafter), characterized by moderately faint near-IR magnitudes (), and extremely red optical-infrared colors (e.g, , see for example Elston et al. 1988; McCarthy et al. 1992; Hu & Ridgway 1994). The observed colors and luminosities place this class of objects at , an hypothesis confirmed in a few cases by a direct spectroscopic measurement of the redshift (Graham & Dey 1996; Spinrad et al. 1997; Stanford et al 1997 - S97 hereafter; Liu et al. 2000). A twofold interpretation of such observational properties is possible: EROs can be either high-redshift starburst galaxies reddened by a large amount of dust, or passively evolving high-z ellipticals characterized by old stellar populations ( Gyr).
The importance of assessing the ERO nature and determining their space density is clear: the epoch of formation of massive elliptical galaxies is a crucial test for the standard hierarchical models for structure formation (e.g.: White & Rees 1978; Kauffmann et al. 1993), which predict such objects to have formed relatively late from the merging of smaller-size objects (presumably disk galaxies). A large density of high redshift evolved ellipticals would imply severe revision to the hierarchical theories. The other relevant question is the global star formation history: calculations based on the observed rest frame UV flux (e.g. Madau et al. 1996; Connolly et al. 1997) might be significantly underestimated, if a large fraction of the overall star formation at high redshift takes place in highly obscured starburst galaxies (e.g. Steidel et al. 1999; Barger et al. 2000).
One way to disentangle this ambiguity is provided, in some cases, by near-infrared spectroscopy, in particular if the ERO spectrum exhibits features revealing star-formation activity, such as the redshifted H line; this is the case, for example, for the galaxy HR10 (Graham & Dey 1996; Dey et al. 1999). More recently, deep near-infrared spectroscopy allowed to classify two more galaxies as likely starburst - although their spectra lack of spectral features - from the amount of reddening required to explain their overall spectral energy distribution (Cimatti et al. 1999). A different kind of test is provided by observations in the submm waveband, which traces the thermal emission by dust in the starbursts; this method was successful in the case of HR10 (Cimatti et al. 1998; Dey et al. 1999) whose detection allowed its non-ambiguous classification as a dust reddened starburst, a result furtherly confirmed by the observation of its CO emission (Andreani et al. 2000). Other objects, first detected in the submm, have afterwards turned out to be EROs (Smail et al. 1999; Gear et al. 2000).
When images of sufficient spatial resolution are available, however, the most direct way to distinguish between the two classes is their morphology: elliptical galaxies are compact, regularly-shaped objects, whereas we expect starburst galaxy to look much more irregular (in particular, if the starburst is triggered by a merger, or if a large amount of dust irregularly distributed is present in the galaxy). HR10, imaged by HST, is consistently characterized by a clearly disturbed morphology (see Dey et al. 1999).
For what concerns the total number density of EROs and their link with passively evolving ellipticals, the works by Cowie et al. (1994) and Hu & Ridgway (1994) suggest that at most a fraction of the present day ellipticals ( 10%) could have its progenitors among EROs, but at present the question is far from being settled. Thompson et al. (1999) and Barger et al. (1999), for example, raise this estimate by a factor 4 5; Benítez et al. (1999) claim that the density of luminous galaxies is comparable with the local value up to ; Eisenhardt et al. (2000), finally, argue that the fraction of red galaxies at might be significantly higher than previously thought, and consistent with a pure luminosity evolution scenario. Finally, Daddi et al. (2000) recently showed that EROs are strongly clustered and that such a clustering can explain the origin of the previous discrepant results on the surface density of elliptical candidates as due to strong field-to-field variations. It has also been noted that even a small amount of star formation would drive a high-redshift elliptical galaxies towards bluer colors, so that it would be missed by a sample selection based on photometric properties only (for example, see Schade et al. 1999). The problem of identifying high-redshift evolved galaxies, therefore, is not restricted to the ERO population alone; in this perspective, color-based selection criteria appear insufficient. Again, a different diagnostic tool (Franceschini et al. 1998; Schade et al. 1999) is provided by a quantitative analysis of morphological characteristics. A local elliptical galaxy is an evolved system from the point of view of both its stellar population and its internal dynamics; we may presume that, for some objects at high-z, a residual small star formation activity (and in general the overall stellar content of the galaxy) could affect the global colors but leave the shape of the brightness distribution more or less unchanged, so that such galaxies could be easily identified on a morphological basis. Of course this kind of approach requires imaging at high angular resolution, such as can only be obtained by space observatories (namely, by the Hubble Space Telescope - HST hereafter).
As a first effort to investigate the morphology of EROs, we present here a quantitative analysis carried out on deep HST archive images both in the optical red and in the near infrared. Our aim is to identify elliptical galaxies using their morphological characteristics, and establish their fractional abundance with respect to the overall ERO population, assuming that their surface brightness distributions at are similar to the ones observed in the local universe. In particular, at the resolution provided by HST, a first classification can be performed visually between compact and irregular objects; among the former ones, different distributions can then be distinguished by fitting different models to the data (for example, exponential and de Vaucouleurs profiles, typically associated to disk galaxies and ellipticals respectively). To this purpose, we have implemented a code for the analysis of the surface brightness distributions of such objects as observed by HST, tested its accuracy on a large number of simulated galaxies, and applied it to a sample of 41 EROs.
This paper is organized as follows: we start describing our sample and the data available for every galaxy, turning afterwards to discuss in detail the techniques developed for the final steps of the data reduction, and for the data analysis; the discussion of the resulting parameters and a morphological classification of the sample are carried out in Sects. 6 and 7; the conclusions follow in Sect. 8.
© European Southern Observatory (ESO) 2000
Online publication: December 15, 2000