1. INTRODUCTION
The principal baryonic component of clusters of galaxies is
diffuse gas held in hydrostatic equilibrium in the
gravitational potential of the cluster. This gas is
hot (107 - 108 K), relatively dense
(10-4 - 10-2 atom cm-3),
and enriched with heavy elements
(e.g., Fe of 0.3 Solar abundance). This combination results
in significant X-ray emission through thermal bremsstrahlung
radiation. Detailed X-ray observations of clusters can provide
us with accurate total mass measurements, clues to the
merger history of clusters, and a chemical record of the
supernova ejecta that polluted the intracluster medium
during the formation of the stars in the member galaxies.
The X-ray emission from clusters can also be exploited to
select clusters irrespective of their member galaxies. While
the optical selection of clusters is well established and
understood, there are potential problems with projection
and the imperfect scaling of the galaxy population to
total cluster mass that make independent selection methods attractive.
There are four key considerations for any X-ray survey:
-
Spatial resolution To capitalize on the extended nature of
the X-ray emission in clusters, it is important to
have sufficient spatial resolution to differentiate
clusters from most other pointlike X-ray sources
(i.e., stars and AGNs). On the other hand, the most
nearby, diffuse clusters can be missed in the same
way low-surface brightness galaxies may be missed in optical surveys.
- Spectral resolution Each class of X-ray
source has a distinctive spectral signature (e.g., black
body for white dwarfs), so information on the X-ray
spectrum of each source can aid identification. The
thermal nature of cluster spectra (temperatures mostly
greater than 2 keV) give clusters relatively flat
soft X-ray spectra (making them distinctive in the ROSAT
survey), but the overall spectral shape is similar to most
unabsorbed AGNs. Therefore, definitive cluster identification
from spectral data alone requires many photons (> 1,000),
which is only feasible for the brightest detections (see
Nevalainen et al. 2001
for an example using XMM-Newton).
- Flux limit of the survey This is a
particularly important factor for cluster surveys as the flux-limited
nature of X-ray samples translates to a selection
over a wide range in redshifts, since the most luminous
objects are being selected from a very much larger volume than
the least luminous ones. This has its advantages but
requires careful analysis. This is illustrated in
Figure 1,
where the X-ray luminosity is plotted against redshift for
a variety of samples described later in the text.
- Area of sky surveyed In the ideal survey
at any wavelength the aim is all-sky coverage. This has been
achieved several times in X-ray astronomy with the earliest
X-ray satellites scanning with collimated proportional counters
(~ 1° resolution) and ROSAT with a soft X-ray
imaging telescope (~ 1' resolution). This wide coverage
comes at the expense of depth [the ROSAT All-Sky Survey
reaches ~ 10-12 erg s-1cm-2 (0.5-2 keV)].
The alternative survey strategy is to select serendipitous
sources in pointed imaging observations, as pioneered by the
Extended Einstein Medium Sensitivity Survey (EMSS;
Gioia et al. 1990).
This allows much deeper surveys but at the expense of
the area (and hence total volume) covered.
There are a number of cluster properties that can be
used to constrain the nature and evolution of clusters.
- Temperature - For gas in hydrostatic
equilibrium
(which appears to hold for the majority of the volume of a cluster)
the gas temperature and density can be used to directly determine the
cluster mass.
- Elemental abundances The X-ray spectrum of
clusters contains
lines from a number of heavy elements. Most prominent of these is the
iron 6.7 keV line. The inferred abundance ratios from X-ray spectra
of O, S, Si and Fe can be used to determine the dominant supernova type
(Loewenstein & Mushotzky
1996).
- Surface brightness profiles The
distribution of gas in
a cluster has a very significant effect on the total X-ray luminosity
of the cluster, given that the intensity of emission is proportional to
the density squared. Clusters with compact, dense cores (i.e., cooling
flows; see
Fabian 1994)
are much more luminous that more extended clusters of
the same measured X-ray temperature
(Fabian et al. 1994)
and can have an effect on the detection probability in X-ray surveys
(Pesce et al. 1990).
Also, the recent discovery of strong density discontinuities,
termed "cold fronts"
(Markevitch et al. 2000;
Mazzotta et al. 2001)
has highlighted the impact of past mergers on the intracluster
medium. These factors make obtaining high-quality, high-resolution X-ray
imaging a vital element of cluster studies.
Each of these requires either dedicated pointed observations
or a survey drawn from the brightest serendipitous detections in pointed
observations. The former is a relatively slow process requiring
time allocation committees to put substantial resources into
programs to observe "complete" samples. The latter is
very slow given the area covered by sufficiently deep X-ray observations.
For the purposes of this review I will define "low redshift" as
z < 0.5 and treat any paper presenting any new X-ray detection
of cluster as a "survey."
In my talk I used the yardstick of exponential growth to
judge progress in known numbers of X-ray emitting clusters
which I modestly named "Edge's Law." This holds that
for every decade of X-ray astronomy the number of clusters
detected increases by an order of magnitude. I would like
to stress that this was a narrative device and not a
serious bid for future surveys in itself. That said, the
rapid progress in cluster research in the past decades
does require us to stand back and assess it as part of a larger picture.
