The 1980's were a period of relative calm in X-ray astronomy. The only satellites launched between 1980 and 1990 were the ESA mission EXOSAT and two Japanese missions, Tenma and Ginga. All three of these satellites were designed for targeted follow-up of known objects, and only EXOSAT had any imaging capabilities. EXOSAT and Ginga contributed a significant number of accurate cluster temperature and iron abundance measurements (the data from EXOSAT making my thesis) but very few "new" detections.
This lull in proceedings did allow the previous scanning and Einstein surveys to be collated, and a complete sample of the brightest 55 clusters was compiled (Lahav et al. 1989). This sample has been used to determine the first cluster temperature function (Edge et al. 1990), the first correlation function from an X-ray sample (Lahav et al. 1989), and the fraction of cooling flows (Edge, Stewart, & Fabian 1992; Peres et al. 1998).
This "free-wheeling" in X-ray surveys leaves the number of clusters in 28 AG at 300, well short of the 630 required for my exponential growth. This gap did not last for long....
The German/UK/US satellite ROSAT was launched on the 1st of June 1990 (27.95 AG). The wide-field, soft X-ray imaging telescope of ROSAT was used to conduct a 6-month scanning survey of the whole sky, from August 1990 to February 1991, which detected in excess of 100,000 sources to a flux limit of (0.3 - 1) × 10-12 erg s-1cm-2 (0.1-2.4 keV) (depth depending on position). From March 1991 to December 1998, ROSAT performed a series of pointed observations with the PSPC and HRI detectors. These observations targeted many known and recently detected clusters, as well as detecting a great many clusters serendipitously.
4.1. ROSAT All-Sky Survey
The ROSAT All-Sky Survey (RASS) is a resource that has still yet to fully exploited 12 years after it was completed. A number of coordinated cluster surveys were embarked upon as soon as the RASS ended. The understandably tight control over the RASS data release and the time-consuming nature of the optical follow-up of the clusters has meant a significant lag in the publication of these samples. Table 1 lists a representative set of RASS cluster surveys, both published and unpublished. There are other RASS studies containing clusters (e.g., the RBS, a complete sample of all RASS sources to a count rate limit of 0.2 PSPC count s-1; Schwope et al. 2000), and studies of groups (e.g., RASSCALS; Mahdavi et al. 2000) and Hickson compact groups (Ebeling, Voges, & Böhringer 1994). Table 1 will be added to in the next few years by NORAS-2, REFLEX-2, and eMACS, which will extend each of the existing surveys to lower fluxes, but these will be reaching close to the intrinsic sensitivity limit of the majority of the RASS.
|XBACS||Abell clusters||5.0 × 10-12||All-sky||276|
|Ebeling et al. (1996)||(0.1-2.4 keV)||Y|
|BCS||Abell, Zwicky, extended||4.5 × 10-12||13,578||199|
|Ebeling et al. (1998)||(0.1-2.4 keV)||Y|
|RASS1BS||Abell, extended||3-4 × 10-12||8,235||130|
|de Grandi et al. (1999)||(0.5-2.0 keV)||Y|
|Ledlow||Abell z < 0.09||none||14,155||294|
|Ledlow et al. (1999)||N|
|eBCS||Abell, Zwicky, extended||3.0 × 10-12||13,578||299|
|Ebeling et al. (2000)||(0.1-2.4 keV)||Y|
|HiFLUGS||All||20 × 10-12||27,156||63|
|Reiprich & Böhringer (2002)||(0.1-2.4 keV)||Y|
|NORAS||extended||3.0 × 10-12||13,578||378|
|Böhringer et al. (2000)||(0.1-2.4 keV)||Y|
|NEP||multiple||0.03 × 10-12||80.7||64|
|Gioia et al. (2001)||(0.5-2.0 keV)||Y|
|CIZA||CCD imaging, |b| < 20°||5 × 10-12||14,058||73|
|Ebeling, Mullis, & Tully (2002)||(0.1-2.4 keV)||Y|
|SGP||optical plates scans||3.0 × 10-12||3,322||112|
|Cruddace et al. (2002)||(0.1-2.4 keV)||Y|
|MACS||multiple, z > 0.3||1.0 × 10-12||22,735||120|
|Ebeling et al. (2001)||(0.1-2.4 keV)||N|
|REFLEX||multiple||3.0 × 10-12||13,905||452|
|Böhringer et al. (2001)||(0.1-2.4 keV)||N|
The wide variety of selection criteria, detection methods and areas covered are clear from Table 1. However, each of the larger samples (BCS, 1BS, Ledlow, REFLEX, and NEP) agree in their derived X-ray luminosity functions (Ebeling et al. 1997; de Grandi et al. 1999; Ledlow et al. 1999; Gioia et al. 2001; Böhringer et al. 2002), so these differences do not greatly affect the samples.
One important difference in the principal RASS samples is that one set (XBACS, BCS, and eBCS) is based on a selection using a Voronoi-Percolation-and-Tesselation (VTP) technique (Ebeling & Wiedenmann 1993) and the other (1BS, SGP, NORAS, and REFLEX) is based on a growth curve analysis (GCA) technique (Böhringer et al. 2001). The flux results from both methods agree within the errors, but only VTP acts as a detection algorithm, as the GCA method requires a set of input positions of potential clusters. This difference is relevant only for the most nearby, extended sources, which are not detected by detection algorithms tuned to search for point sources. VTP will reliably detect these, but, through lack of access to the full RASS data set, it was not run over the full sky during the compilation of the BCS. With all RASS data now in the public domain, this is now possible in principle.
The optical follow-up of clusters at redshifts above 0.3 requires additional optical imaging, as archival photographic plate material is too shallow to reliably detect cluster members. At the brighter flux limits (5 × 10-12 erg s-1cm-2) there are relatively few of these distant clusters [e.g., two in the BCS and RXJ1347-11 (z = 0.45) in RASS1BS], but this number increases with decreasing flux limit (e.g., there are seven in the eBCS). With these higher redshift, X-ray luminous clusters in mind, Harald Ebeling and I have searched the RASS-BSC sample (Voges et al. 1999) for z > 0.3 clusters using the UH 2.2m telescope to a flux limit of 10-12 erg s-1cm-2, creating the MAssive Cluster Survey (MACS; Ebeling, Edge, & Henry 2001). To date, the sample contains 120 clusters with very few candidates left for imaging. The MACS sample has been extensively followed up at all wavelengths, including complete VRI imaging with the UH 2.2m, multi-object spectroscopy with Keck, Gemini and CFHT, deep, wide-area imaging with Subaru/SUPRIMECAM, VLA imaging (Edge et al. 2003), Sunyaev-Zel'dovich observations (LaRoque et al. 2003), Chandra observations, and Cycle 12 HST/ACS imaging. This sample represents more than an order of magnitude improvement in the number of distant, X-ray luminous clusters known (i.e., two EMSS clusters with z > 0.4, Lx > 1045 erg s-1, compared to 39 in MACS).
4.2. ROSAT Pointed Observations
The large field of view of the PSPC detector made it very efficient at detecting serendipitous X-ray sources within 15' of the pointing position of the telescope. Given the substantial numbers of relatively deep observations during the ROSAT Pointed Phase an area of well over 500° at high Galactic latitude has been covered by the central region of the PSPC with more than 10 ks exposure. This area is significantly reduced, as some targets are not suitable for serendipitous searches (e.g., nearby clusters, nearby galaxies, globular clusters, etc.), but the majority has been used in a series of surveys (listed in Table 2). As with the RASS samples, the selection strategies differ between surveys, but the results from each survey agree. For instance, the requirement of significant source extent used by SHARC certainly eases source selection, but at the potential loss of the most distant and/or compact, cooling flow clusters. These effects do not appear to have any great impact on the results.
|Paper||(10-14 erg s-1cm-2)||(°)||Published?|
|Mason et al. (2000)||(0.5-2.0 keV)||Y|
|Perlman et al. (2002)||(0.5-2.0 keV)||Y|
|Vikhlinin et al. (1998)||(0.5-2.0 keV)||Y|
|Collins et al. (1997)||(0.5-2.0 keV)||N|
|Romer et al. (2000)||(0.5-2.0 keV)||Y|
|Borgani et al. (2001)||(0.5-2.0 keV)||N|
|Donahue et al. (2002)||(0.5-2.0 keV)||Y|
|BMW||extent||~ 10||~ 300||~100|
|Lazzati et al. (1999)||(0.1-2.4 keV)||N|
|Gilbank et al. (2003)||(0.5-2.0 keV)||N|
|Jones et al., in prep.||(0.5-2.0 keV)||N|
The majority of the clusters selected in these surveys are relatively nearby (z = 0.15 - 0.3) and of low X-ray luminosity (Lx = 1043 - 44 erg s-1; 0.5-2 keV), but a few distant, luminous clusters are found (Ebeling et al. 2000, 2001), and in the deepest of the ROSAT serendipitous sample, RDCS (Rosati et al. 1998), there are several candidate clusters at z > 1 (Borgani et al. 2001).
The PSPC instrument eventually ran out of gas in mid-1994, but ROSAT continued to make observations with the HRI. While this instrument was less sensitive than the PSPC and covered a smaller area of sky, the excellent spatial resolution it provided has been used very effectively by the Brera Observatory group in the BMW survey (Lazzati et al. 1999; Panzera et al. 2003). While the combination of sensitivity and total area covered by the BMW survey will never match that of PSPC surveys (e.g., the 160 square degree survey; Vikhlinin et al. 1998), it does provide an important reliability test.
The full potential of the ROSAT pointed phase has yet to be tapped as all existing surveys have been restricted to the central 15'-20' radius where the point-spread function is best. While the flux sensitivity is poor in the outer parts of the detector, the brighter (fx > 10-13 erg s-1 cm-2) sources can easily be detected. As noted above, the most time consuming part of the follow-up of X-ray selected cluster candidates at z > 0.3 is the deeper optical imaging required. This is exacerbated at lower X-ray fluxes by the fainter optical counterparts of all X-ray counterparts. The combination of the low-resolution X-ray imaging with deeper multicolor panoramic surveys such as SDSS, UKIDSS, and RCS2 will provide a "free" resource to identify cluster counterparts to these ROSAT sources.
So, how successful has ROSAT been overall in harvesting clusters? The RASS samples published or about to be published account for a total of around 1,200 new clusters, with a further 500-1,000 at lower fluxes. Add these to the serendipitous detections, ~500 in the central region of the PSPC (of which ~ 250 are published), > 1,500 in the outer regions, and ~ 300 clusters in the HRI. Therefore, the total after the ROSAT mission is ~ 4,000 (when in 36.5 AG 4,500 would be required). It is worth noting that the majority of these clusters have yet to be identified and many may never be.