The ROSAT-ESO flux limited X-ray (REFLEX) galaxy cluster survey. I. The construction of the cluster sample

We discuss the construction of an X-ray flux-limited sample of galaxy clust ers, the REFLEX survey catalogue, to be used for cosmological studies. This cluster identification and redshift survey nas conducted in the flame of a n ESO key programme and is based on candidates selected from the southern p art of the ROSAT All-Sky Survey (RASS). For the first cluster candidate sel ection from a flux-limited RASS source list, we make use of optical data fr om the COSMOS digital catalogue produced front the scans of the UK-Schmidt plates. To ensure homogeneity of the sample construction process, this sele ction is based only on this one well-defined optical data base. The nature of the candidates selected iii this process is subsequently checked by a mo re detailed evaluation of the X-ray and optical source properties and avail able literature data, The final identification and the redshift is then bas ed on optical spectroscopic follow-up observations. In this paper we document the process by which the primary cluster candidat e catalogue is constructed prior to the optical follow-up observations. We describe the reanalysis of the RASS source catalogue which enables us to im pose a proper flux limit cut to the X-ray source list without introducing a severe bias against extended sources, We discuss th correlation of the X-r ay and optical (COSMOS) data to find galaxy density enhancements at the RAS S X-ray source positions and the further evaluation of the nature of these cluster candidates. Based also on the results of the follow-up observations we provide a statistical analysis of the completeness and contamination of the final cluster sample and show results on the cluster number counts. Th e final sample of identified X-ray clusters readies a flux limit of 3 10(-1 2) erg s(-1) cm(-2) in the 0.1-2.4 keV band and comprises 452 clusters in a n area of 4.24 ster. The results imply a completeness of the REFLEX cluster sample well in excess of 90%. We also derive for the first time an upper l imit of less than 9% for the number of cluster which may feature a dominant contribution to the X-ray emission from AGN. This accuracy is sufficient f or the use of this cluster sample for cosmological tests.