Finding Galaxy Clusters at z > 0.5

Michaeal D. Gladders, H. K. C. Yee

Dept. of Astronomy, U. of Toronto
60 St. George St., Toronto, ON
Canada, M5S 3H8
Electronic-mail: gladders@astro.utoronto.ca, hyee@astro.utoronto.ca

Abstract:

Galaxy clusters are the most massive collapsed systems in the Universe. As such, they are extremely important as both cosmological probes and as laboratories for studying galaxy evolution in dense environments. In this paper, we suggest an efficient technique for selecting galaxy clusters from a two-colour, wide-field survey. We demonstrate a test of this new technique on photometric data from a redshift survey, and show that it effectively selects clusters and groups and is less affected by projection effects than other techniques.

Introduction

  Galaxy clusters represent the collapsed state of the most extreme fluctuations in the primordial density field, and are thus critical testing grounds for theories of structure formation, galaxy evolution and ultimately, cosmological models. In general, in a present-day low density universe, clusters must have formed at relatively early times. This is because the collapse of large scale structure into strong over-densities generally occurs when $\Omega_{local}\geq1$. The 'freeze-out' of structure in a low-density universe at relatively early times requires that the structures we see at the present epoch must have been essentially in place at high redshifts (e.g. Press & Schechter 1974). Conversely, in a present-day critical universe ($\Omega_{0}=1$), significant collapse can still be occurring. Additionally, the formation history of clusters is dependent on $\sigma_{8}$ (Frenk et al. 1990). In a universe with biased galaxy formation ($\sigma_{8}<1$), the underlying mass distribution is smoother than would be implied by the galaxy distribution, and so the requirement that the observed galaxy structure formed at high redshifts is somewhat relaxed. Thus the density of clusters as a function of z is strongly dependent on both the overall matter density, $\Omega$, and the the normalization of the perturbation spectrum, $\sigma_{8}$ (Figure 1).

 

Figure 1:  The expected cumulative counts of clusters per deg² for two cosmologies, for Abell Richness 1 and 2. Note the large difference in cluster abundances between the two cosmologies, and that these differences are more pronounced at higher z.

.

Numerous systematic searches have been made for clusters since Abell's original definition of a comprehensive low-redshift catalog. The low redshift galaxy clusters are becoming well understood, with many comprehensive surveys, at many wavelengths (e.g. Henry et al. 1992; Lopéz-Cruz 1997). However, at high redshifts (z>0.5) we know much less about galaxy clusters. This is in large part due to the lack of a well defined cluster sample at z>0.5. Previous attempts to define such a sample have either been too small (i.e. the 5.1 sq. degrees of the Palomar Distant Cluster Survey [Postman et al. 1996]), so that they do not statistically sample the rare richer objects, or extremely wide but too shallow (e.g. the EMSS: Henry et al. 1992), so that only the truly unusual objects (i.e. ultra-rich clusters) are detected at high z. Progress has been made in recent years, but there is still an almost total dearth of known clusters at z>1. Detections of clusters at these redshifts have almost exclusively been a result of pointed searches around AGN or radio galaxies (e.g. Deltorn et al. 1997), and are likely not representative. A large sample of rich galaxy clusters at z>0.5 is badly needed.

Cluster Finding

Two major techniques for finding galaxy clusters exist in the literature. The first, X-ray selection, relies upon the detection of thermal bremsstrahlung emission from the intra-cluster medium. This technique has been shown to be quite effective to $z \approx 0.75$(e.g. Rosati et al. 1998), but may be hampered at higher redshifts by the negative evolution of the bright end of the X-ray luminosity function at high z (e.g. Bower et al. 1997; Gioia et al. 1990). The other major selection technique relies upon direct optical/IR detection of the clusters, using imaging data in one or more passbands. Numerous techniques have been used to identify clusters in such imaging surveys, the most recent being the matched filter technique of Postman et al. (1996).

One of the most striking recent observational facts about galaxy clusters is the apparent uniformity and great age of their core elliptical galaxy populations e.g. Lopéz-Cruz 1997; Gladders et al. 1998). These galaxies appear to be coeval (both within individual clusters [Bower et al. 1992] and between clusters [Lopéz-Cruz et al. 1998; Smail et al. 1998]) and appear to have formed at z>3 (Figure 2). Given this, it seems appropriate to search for rich clusters using the core elliptical galaxies as tracers, as these galaxies: 1) exist in all rich clusters studied to z=1.3 (e.g. Stanford et al. 1997; Stanford et al. 1998); 2) populate a highly homogeneous red sequence (e.g. Ellis et al. 1997; Lopéz-Cruz 1997); 3) are extremely old (e.g. Bower et al. 1992; Gladders et al. 1998); 4) are the dominant luminous galaxies in clusters; 5) show positive luminosity evolution with increasing z; 6) are redder than almost all field galaxies at z>0.7; 7) are intrinsically compact, and can thus be morphologically selected (c.f. Abraham et al. 1994) and, 8) are the most strongly clustered galaxies within a cluster (Dressler et al. 1997). Thus, a 2-colour imaging survey with filters straddling the 4000Å break is sensitive to the presence of clusters, as these will appear as concentrations of galaxies in both angular space and colour. Furthermore, such a survey is not prone to the the projection effects common to single filter surveys, as a random projection of field galaxies or groups does not exhibit the necessary red sequence in a colour-magnitude diagram which typifies a rich cluster. We have tested such an algorithm on a 25'$\times$27' patch of the CNOC2 (Yee et al. 1998) redshift survey data (Figure 3) and find that we can reliably locate real overdensities in the survey, and that we can disentangle projection effects. Though we find no rich clusters in the CNOC2 data (as is expected, given the small area), the success of this technique in finding poor clusters and groups over this small area bodes well for finding rich clusters in a large 2-colour survey.

 

Figure 2:  A combined colour-magnitude diagram for 41 clusters at z<0.15. The data are from [12], and show only galaxies in the interior 0.5 Mpc. The individual CMDs have been k-corrected to the mean redshift, and then stacked. Note the impressive cluster-to-cluster homogeneity of the red sequence.

 

Figure 3:  Cluster finding test on a section of a CNOC2 patch (27'$\times$25' in g,I). The 'All' panel shows the smoothed galaxy surface density, with 2 significant peaks. The remainder show density maps at color slices with corresponding redshifts indicated. Several more significant peaks appear and the S/N of the peaks is increased by $\geq$3 over the 'All' map. The most prominent peak in the 'All' map is resolved as 2 peaks at different z. All peaks (including the projection) are confirmed by the redshift data. These density enhancements are not rich clusters; rather, they are poorer than Abell Richness 0.