2 The REFLEX galaxy cluster survey

The survey area of REFLEX is the southern hemisphere below a declination of +2.5$\deg$. A region of $\pm 20$ degrees around the galactic plane is excluded form the study, since clusters are difficult to recognize optically in the dense stellar fields of the Milky Way and the X-ray detection is hampered by the higher interstellar absorption in the inner parts of the galactic band. The region 2.5 degrees above the equatorial equator is included in this project since the COSMOS data extend up to this declination. It provides some overlap with the NORAS survey project (e.g. Böhringer et al. 2000) where both cluster identification programmes can be compared. The total area thus covered is 4.34 ster or 14248 deg². In addition to the region around the galactic plane, the dense stellar fields of the two Magellanic clouds prevent an efficient galaxy search in these regions of the southern sky. In particular the star-galaxy separation technique used in the construction of the COSMOS data base became inefficient in these crowded areas (H.T. MacGillivray, private communication) and therefore no star-galaxy classification is actually provided in the COSMOS data released and used for our project. Therefore we exclude an area of 244.4 deg² for the LMC and 79.8 deg² for the SMC which essentially follows the boundaries of those UK-Schmidt plates without object classification. The areas which are excised from our survey are specified in detail in Table 1.

   

Table 1: Regions of the sky at the LMC and SMC excised from the Survey

The total survey area after this excision amounts to 4.24 ster or 13924 deg² which corresponds to 33.75% of the sky. This survey covers the largest area for which currently a homogeneous combined optical/X-ray survey is possible, since there is no optical survey covering both hemispheres simultaneously. The observational goal of this survey programme is the identification and redshift determination of all galaxy clusters in the study area above a given flux limit. In a first step, within the ESO key programme, we have completed the observations for a sample of 452 galaxy cluster (with redshifts for 449 clusters) above a limiting flux of 3 10^-12 ergs^-1 cm^-2 (0.1-2.4 keV). In addition we have already secured many redshifts at lower fluxes and we plan to extend the redshift survey to flux limit of 1.6-2 10^-12 ergs^-1 cm^-2. This corresponds to a count rate limit in the hard ROSAT band of about 0.08-0.1 cts s^-1. With a typical exposure in the southern part of the RASS of about 330 sec this yields about 25-30 photons for the fainter sources. This is still just enough to determine a flux within uncertainty limits of typically less than 30% and provides some leverage for the determination of some source properties. At this flux limit we expect between 700 and 1000 galaxy clusters in the survey area (based on the number counts of previous surveys e.g. Gioia et al. 1990; Rosati et al. 1998).

For the preparation of the candidate sample we have therefore chosen to start with a source sample with a count rate limit of 0.08 cts s^-1 in the hard ROSAT band (channel 52 to 201 corresponding approximately to an energy range of 0.5 to 2.0 keV). Note that all the fluxes quoted in this paper refer to the total ROSAT energy band (0.1-2.4 keV) in contrast to the more restricted band of pulse high channels chosen for the determination of the count rate. This count rate limit translates into a flux limit for cluster type spectra of 1.55-1.95 10^-12 ergs^-1 cm^-2, a range determined mainly by variations of the interstellar HI column density in the REFLEX area (20% in the range 1-10 10²⁰ cm^-2). Weaker dependences on the cluster temperature (e.g. $\sim$1.4% in the range 3-8 keV, see Fig. 8 in Böhringer et al. 2000) and redshift (in analogy to the optical K-correction, 0.5% in the range z = 0 to z = 0.2) are found. (Below about 2 keV the temperature dependence is stronger, however.) We will be quoting unabsorbed flux values in the ROSAT energy band (defined as 0.1 to 2.4 keV) throughout this paper since the results in this energy band are less dependent on the spectral model assumptions for the sources compared to any other significantly wider band definition. Further assumptions or information on the source spectrum (e.g. intracluster plasma temperatures) are needed to subsequently convert these primary data to other energy bands or to bolometric fluxes and luminosities.

For the calculations of the fluxes, the luminosities, and some other physical parameters in this paper we have made the following assumptions. A first approximate unabsorbed flux is calculated for each X-ray source from the observed count rate, prior to any knowledge about its nature and redshift by assuming a thermal spectrum with a temperature of 5 keV, a metallicity of 0.3 solar (with abundances taken from Anders & Grevesse (1989). A redshift of zero, and an interstellar column density of hydrogen as obtained from Dickey & Lockman (1990) & Stark et al. (1992) for the X-ray source position is adopted. This nominal flux is used to impose the flux limit on the X-ray source sample. After a cluster has been identified and its redshift secured a better temperature estimate is obtained by means of the temperature/X-ray luminosity relation (Markevitch 1998), and a corrected flux and X-ray luminosity is calculated taking the new estimated temperature, the K-correction for the observed redshift, and the dependence on the interstellar absorption into account. The X-ray luminosities are always calculated in the ROSAT band in the cluster restframe, while the fluxes are given in the ROSAT band for the observer frame as unabsorbed fluxes. The calculations are performed within the EXSAS software system (Zimmermann et al. 1994) with the spectral code from John Raymond (Raymond & Smith 1977). Instead of using the standard codes of EXSAS for the count rate-flux conversion we are using our own macros, which have been tested against XSPEC and show a general agreement within less than 3%. For the calculations of the luminosities and other physical properties of the clusters we assume a standard cosmology with H₀ = 50 kms^-1 Mpc^-1, $\Omega_0 = 1$ and $\Lambda = 0$. While the basis of the source detections is the standard analysis of the RASS (Voges et al. 1999), we have reanalysed the source count rates and other source properties as described in Sect. 4 with the growth curve analysis technique. Note that previous comparisons of the results of this technique with deeper pointed ROSAT observations show that the measured flux underestimates the total cluster flux, typically by an amount of 7-10% (Böhringer et al. 2000). The fluxes and luminosities quoted here are the measured values without a correction for the possibly missing flux.