The arrival of the third decade of X-ray astronomy midway through the ROSAT mission coincided with advances in CCD detector technology that have allowed a vast increase in the power of X-ray spectroscopy. These advances, coupled with nested-mirror systems, mark a clear maturing in the field and a move away from large samples of objects with limited information to limited samples with very detailed information. This is an inevitable progression that emerging disciplines experience, radio astronomy being a prime example. With the advance of aperture synthesis, radio astronomers in ~ 35 AJ (Anno Jansky) could obtain insights into the nature of individual sources and the focus moved away from surveys. In the past decade, radio astronomy has turned back to surveys (NVSS, FIRST, WENSS, 4MASS), and this will happen in X-ray astronomy (but hopefully in less than 25 years time!).
This trend for more detailed study has had a huge impact on cluster research, and the need for spatially resolved spectroscopy of clusters was apparent from the first X-ray detections of clusters. The nature of cluster surveys has also changed with a greater emphasis on understanding complete samples in many different wavelength regimes (e.g., Crawford et al. 1999; Giovannini, Tordi, & Ferretti 1999; Pimbblet et al. 2002). A sample of 200 clusters is a great resource but of little use without some information about the X-ray temperature, iron abundance, X-ray surface brightness profile, optical photometry and spectroscopy, or radio imaging. The availability of new optical, near-infrared, and radio surveys (e.g., SDSS, UKIDSS, NVSS, FIRST) will make the multiwavelength aspects of these studies much easier, but the need for further X-ray observations is hard to avoid.
5.1. ASCA Observations
The first step in this progression was the Japanese-US satellite ASCA. The nested, foil-replicated mirrors of ASCA resulted in a relatively asymmetric, broad point-spread function (2' FWHM), but the excellent performance of the SIS CCD detectors provided some very high-quality spectra for clusters (Mushotzky & Scharf 1997; Markevitch 1998; Fukazawa et al. 2000; Ikebe et al. 2002).
Over the course of the seven-year pointed phase, ASCA provided accurate temperatures and iron abundances for most of the 350 clusters observed. While few complete samples were observed, the ASCA data are an excellent complement to archival ROSAT observations. The notable exceptions to this are the flux-limited sample of 61 ROSAT-selected clusters (Ikebe et al. 2002) and the complete sample of 0.3 < z < 0.4 EMSS clusters (Henry 1997) from which limits of the evolution of the cluster temperature function can be derived.
5.2. The Unfulfilled Potential of ABRIXAS
One of the most disappointing events in X-ray astronomy was the unfortunate failure of the German satellite ABRIXAS in June 1998. Its simple design and the track record of the team behind ROSAT meant the planned 3-year, all-sky survey ABRIXAS would have had a huge impact on X-ray astronomy. The survey depth envisioned of 1.5 × 10-13 erg s-1cm-2 (0.5-2.0 keV) and 9 × 10-13 (2-12 keV) would have detected in excess of 20,000 clusters (i.e., more than the number required to keep pace with exponential growth).
5.3. Chandra and XMM-Newton
The launch of Chandra and XMM-Newton in 1999 has seen X-ray astronomy reach full maturity. The sub-arcsecond imaging of Chandra and unprecedented throughput of XMM-Newton have had a profound impact of our understanding of clusters (e.g., McNamara et al. 2000; Peterson et al. 2001; Allen, Schmidt, & Fabian 2002). The potential for surveys with both satellites is largely through serendipitous detections, but several important pointed surveys are being undertaken.
The only large Chandra serendipitous survey is CHamP (Wilkes et al. 2001), which will cover 14° in 5 years and identify 8,000 X-ray sources of all types, of which 150-250 will be clusters (which will all be spatially resolved). The relatively small number of clusters makes this sample unlikely to set any strong cosmological constraints, but it will act as an excellent control sample for past and future samples to test how spatial resolution affects detection statistics.
XMM-Newton has a program similar to CHamP, the XID program that has three tiers: faint (10-15 erg s-1cm-2, 0.5°), medium (10-14 erg s-1cm-2, 3°), and bright (10-13 erg s-1cm-2, 100°). Again, like CHamP, the number of clusters detected in the XID program will be small (< 50), so from a purely cluster view-point is not particularly relevant. There are currently two dedicated serendipitous cluster surveys. One expands on the XID programme (Schwope et al. 2003) and the other (the X-ray Cluster Survey, XCS, Romer et al. 2001) aims to extract all potential cluster candidates from the XMM-Newton archive and compile a sample of > 5,000 clusters from up to 1,000° over the full lifetime of the satellite. The contrast of XCS to CHamP and XID illustrates the huge increase in efficiency when one class of objects is chosen over the study of "complete" X-ray samples or contiguous area X-ray surveys, such as the XMM-LSS (Pierre et al. 2003), where the number of detected cluster is relatively small.
The principal pointed cluster surveys with Chandra and/or XMM-Newton target a sample of MACS clusters (Ebeling et al. 2001) with Chandra using GTO and GO time (PIs Van Speybroeck and Ebeling), a sample of REFLEX clusters with XMM-Newton in GO time (PI Böhringer) and a sample of SHARC clusters with XMM-Newton in GTO time (PI Lumb). Each of these projects is designed to determine the cluster temperature function, but will clearly have may other potential uses. These projects are all based on sub-samples of ROSAT-selected clusters to minimize the number of observations required. The reluctance of time allocation committees to devote time to complete samples in preference to the "exotica" (e.g., most distant, strongly lensing, etc., which predominate in successful proposals) is a hindrance to this "targeted" survey approach.