Annu. Rev. Astron. Astrophys. 1982. 20: 547-85 Copyright © 1982 by Annual Reviews. All rights reserved

3.2 X-Ray Double Clusters - A Possible Intermediate Dynamical Phase

During the Einstein Observatory survey of clusters of galaxies, four were discovered to have a bimodal X-ray surface brightness distribution (Forman et al. 1981a, see Figure 1 for an example). With the observed small angular separations and velocity differences of the subclusters, Forman et al. showed that these double clusters could not result from random coincidences of a known cluster with another uniformly distributed population. The subclusters have X-ray luminosities of 0.8 to 3.2 x 10⁴⁴ erg s^-1 (within 1 Mpc), core radii of 0.53 to 0.80 Mpc with uncertainties of up to 30%, and central gas densities of approximately 10^-3 cm^-3. For gas in hydrostatic equilibrium, the X-ray surface brightness can be used to trace the distribution of the total cluster mass. Forman et al. computed total masses within 1 Mpc of the subcluster centers to be 4-5 x 10¹⁴ M - typical of the values obtained for bound and virialized clusters. As for other X-ray-emitting clusters, the hot gas contributes only ~ 10% of the total cluster mass.

One of the double clusters (Abell 98) has been optically studied in detail (Faber & Dressler 1977, Dressler 1976, Henry et al. 1981, Beers et al. 1983). Henry et al. found that the optical galaxy counts also display a bimodal distribution. Beers et al. obtained virial mass estimates for the subcluster components of ~ 3 x 10¹⁴ M (with uncertainties of a factor of two), in good agreement with those determined from the X-ray observations (Forman et al. 1981a, Henry et al. 1981). Beers et al. modeled A98 as a two-body system assuming the two subclusters are in a linear orbit (no sheer or rotation) and showed that there is only a 2% probability that the system is unbound. They further argued that the system is most likely infalling (with 95% probability) and will merge in 2.5 x 10⁹ yr.

Although the X-ray observations have focused attention on the double clusters, there is still no precise information available for many of their dynamical tracers - galaxy density, accurate velocity dispersions, or X-ray temperatures - that would independently imply an evolutionary classification for these clusters. In light of the results of numerical cluster modeling, Forman et al. suggested that the X-ray double clusters represent a possible intermediate evolutionary stage (Figure 2c) before the final merger of the subclusters into a relaxed Coma-type cluster. Hierarchical gravitational clustering has been analyzed through numerical simulations by, among others, Peebles (1970), Aarseth (1969), and White (1976a). These numerical experiments have shown that an initially expanding cloud of galaxies will collapse and eventually reach equilibrium with an extended halo surrounding a denser core. White showed that during this evolution, subclustering first occurs about the more massive galaxies, the subclusters then merging into a few large concentrations before their final coalescence (see Figure 2). The observed subcluster separations of 1 Mpc (a minimum separation since projection effects cannot be removed) as well as the core radii agree with those expected from Whites model for an intermediate dynamical phase. However, the observation of X-ray double clusters should not be interpreted as confirmation of the model since only a few simulations exist and they explore a limited range of initial conditions.

Figure 2. The numerical simulation of White (1976a) shows an initially expanding set of galaxies (a) followed by increasing clustering (b) to form small groups. A bimodal subclustering develops (c), which may correspond to the double cluster shown in Figure 1. The final relaxed cluster (d) is seen at the lower right. The bar in each panel corresponds to ~ 1.0 Mpc.

The double clusters differ substantially from typical superclusters (Karachentsev et al. 1976), since supercluster mean densities are about 1000 times smaller. Therefore if superclusters are bound, their collapse times typically exceed the Hubble time.