A $z=0.9$ supercluster of X-ray luminous, optically-selected, massive galaxy clusters

The cluster RCS231953+0038.0 was discovered in the first Red-Sequence Cluster Survey (RCS-1, Gladders & Yee 2005) and displays spectacular strong lensing features (Gladders et al., 2003). As such, it has been the subject of extensive follow-up, including multi-object spectroscopy (MOS) (Barrientos et al., 2008) and Chandra X-ray imaging (Hicks et al., 2008). The Chandra X-ray observations were targeted so as to include two other cluster candidates (RCS231848+0030.1 and RCS232002+0033.4) which the RCS catalog indicated were consistent with being at the same photometric redshift as the primary cluster target. All three systems were found to be coincident with extended X-ray emission (Hicks et al., 2008). In this Letter we present new follow-up spectroscopy of these three clusters, showing that all three are located at the same redshift and therefore likely physically associated. Hereafter we refer to these collectively as the supercluster RCS2319+00.

Throughout, we assume a concordance cosmology with $\Omega_{m}=0.3$, $\Omega_{\Lambda}=0.7$, $H_{0}=70$ km s^-1 Mpc^-1.

Wide-field MOS spanning a field of 27′ diameter using IMACS on the 6.5-m Baade telescope was obtained on the night of 2006 October 28-29. The observations consisted of 4 $\times$ 1800s exposures of a single mask comprising 725 slits utilizing a custom band-limiting filter, which restricted the wavelength coverage to 6050-8580Å, allowing coverage of the key spectral features from [Oii]$\lambda$3727 to G-band ($\sim 4304$Å) and somewhat beyond, at the redshift of the cluster. The spectra were reduced following a technique similar to that described in Gilbank et al. (2007). Briefly, we used the COSMOS software package¹¹1See http://www.ociw.edu/$\sim$oemler/COSMOS.html to process the 2D spectra. Our procedure required the additional step of masking second and third order bright skyline contamination, which overlapped many of our first order science spectra due to the highly multi-tiered nature of our data. The spectra were extracted to 1D using the apall task in IRAF. Redshifts were measured using the IRAF task rvsao. All the spectra were visually inspected and assigned redshift confidences based on the system described in Gilbank et al. (2007). Hereafter we group redshifts with quality flags 1-3 into a higher confidence category (302 objects) and denote class 4 redshifts as lower confidence (113 objects). The redshift histograms for all galaxies in this field are shown in Fig. 1. Also included are 16 redshifts centered on RCS231953+0038.0 from VLT spectroscopy (Barrientos et al., 2008). A clear peak is seen at z$\sim$0.9. The insets show histograms within 3.2 arcmins (1.5 Mpc) of each cluster center.

The properties of the three clusters are summarized in Table 1 and we refer to the three clusters as A, B, C, as denoted in the table. Fig. 2 shows the sky distribution of all spectroscopically observed galaxies within the IMACS field. Although the field is sparsely sampled, overdensities of spectroscopic members can be seen around the locations of the most significant peaks in the RCS significance map (see Gladders & Yee 2005). The visually-identified BCG in each case has a confident redshift near z$=$0.90 (0.9008, 0.9025 and 0.9005 respectively).

The separations of the components of this system, both in the plane of the sky ($<$ 3 Mpc) and along the line of sight ($\sim$10 Mpc, assuming that the velocity difference is due to Hubble flow) are significantly closer than those seen in other supercluster candidates at these redshifts (Lubin et al., 2000; Swinbank et al., 2007). This invites the question: will these merge to form a more massive cluster? To answer this, we consider a simple estimate (Sarazin, 2002). If we assume that the clusters are not moving apart with the Hubble flow and that their relative velocity is comparable to their restframe LOS velocity difference ($\sim$1200 km s^-1), then the time taken to traverse the $\sim$3 Mpc separation would imply that the individual cluster components would merge by z$\sim$0.5. The merger redshift is also comparable if we assume that the infall velocity is of the order of the velocity dispersion of cluster A. Thus, this system would evolve into a cluster of mass $\gtrsim$ 10¹⁵ M_⊙ at z$\sim$0.5, similar to the z$=$0.54 cluster MS0451.5-0305 (Donahue, 1996; Ellingson et al., 1998).

We explore the expected space density of such systems by utilizing the cluster catalogs from the light cone output of the Virgo Consortium Hubble Volume simulation²²2See http://www.mpa-garching.mpg.de/NumCos (Evrard et al., 2002). Since the abundance of superclusters depends sensitively on the mass threshold chosen, we adopt the conservative approach of searching for clusters of 3.3$\times$10¹⁴ M_⊙ (the 1$\sigma$ lower limit for the cluster with the lowest X-ray mass) within similar separations. In one octant of the sky, we find only one such system in the redshift range 0.8 to 1.0, corresponding to a surface density of $1.9\times 10^{-4}$ per sq. deg. We should therefore have a probability of $<$2% of finding such an object in the 92 deg² of RCS-1, if the Hubble Volume simulation gives an accurate description of the abundance of structure. Relaxing the maximum separation requirement slightly to 10 Mpc (which is the maximum inferred spatial separation if the LOS velocity difference is due to cosmic expansion), we find five such triple clusters, corresponding to a surface density of $9.5\times 10^{-4}$ per sq. deg. or a probability of $<$9% of finding such an object in the RCS-1 survey area. If the clusters are in fact less massive than our mass estimates suggest, then the probability of finding such a supercluster structure increases. We repeat our analysis using different mass thresholds for the component clusters until the surface density of such structures approaches that of one system per RCS-1 survey area. To do this requires the individual clusters to have masses of only $\sim 2.3\times 10^{14}$ M_⊙, i.e., all three would have to be at least a factor of two less massive than our best estimates, which corresponds to a $>$2$\sigma$ deviation. The Hubble Volume simulation was performed with a value for the amplitude of fluctuations, $\sigma_{8}$, of 0.90, whereas recent measurements such as the WMAP 3-year results (Spergel et al., 2007), find a lower value ($\sigma_{8}=0.761^{+0.049}_{-0.048}$). Lowering $\sigma_{8}$ lowers the abundance of massive structures, particularly at higher redshift, making the chances of finding such a supercluster more unlikely in this cosmology.

We have presented spectroscopic confirmation of three clusters, selected from a large optical survey and found to be X-ray luminous from Chandra observations. Two independent mass estimators for one cluster (including a strong-lensing estimate from a newly confirmed z$\sim$4 Lyman break galaxy behind the cluster) support the X-ray mass estimates.

The major axes of the three components of supercluster RCS2319+00, as traced by the red-sequence significance (Gladders & Yee, 2005) contours, all point towards the nearest-neighbor cluster (Fig. 2). This alignment is also largely mirrored by the X-ray contours (fig. 1, Hicks et al. 2008). The tendency for neighboring clusters to point toward each other was first noticed by Binggeli (1982) and is further independent evidence that these clusters are associated. In addition, this may give some clues as to the 3D alignment of the clusters’ haloes. One of the interpretations for the high incidence of strong-lensing clusters in RCS-1 is that the strongly-lensing systems may have their major axes directed along the line-of-sight (Gladders et al., 2003). The alignment with the neighboring cluster, coupled with the small line-of-sight separation, might suggest that this is not the explanation in this case. Detailed lens modeling will be required to try to constrain halo orientation and is beyond the scope of this Letter.

The supercluster RCS2319+00 comprises three clusters, each massive in their own right ($\sim 5\times 10^{14}$ M_⊙), which are likely to merge by low-redshift to form a system $>1.5\times 10^{15}$ M_⊙. Thus, this system gives us a unique opportunity to study one of the most massive clusters in the universe in the process of assembly. Two other groups (Lubin et al., 2000; Swinbank et al., 2007) have reported finding superclusters at similar redshifts, but these represent somewhat different systems to that presented here. The Lubin et al. (2000) supercluster (Cl1604) comprises at least four components (Gal et al., 2005) spanning $\sim 10$ Mpc by $\sim 100$ Mpc ($0.865<z<0.921$). The clusters range in velocity dispersion from $<500$ km s^-1to $\sim$900 km s^-1. However, the X-ray temperature for the main cluster of Cl1604 is only 2.5 keV (Lubin et al., 2004), suggesting that at least this component is less massive than its velocity dispersion would imply, possibly due to the cluster being dynamically unrelaxed. The lower masses and larger separations suggest that Cl1604 is less likely to merge into as massive a cluster as RCS2319+00. The Swinbank et al. (2007) supercluster comprises five systems with approximate velocity dispersions $\lesssim 500$ km s^-1 ($M\lesssim 8\times 10^{13}$ M_⊙) spread across 30 Mpc. The lower masses of these systems is not surprising given the low space density of massive clusters and the 1 sq. deg. size of their survey.

The chances of finding a structure as massive as RCS2319+00 within the RCS-1 survey are $<9\%$ with $\sigma_{8}=0.90$ and less with a lower value, using the currently favored cosmological parameters. This may indicate that refinements to the current cosmological model are still to be made, but we are, of course, cautious about using the existence of a single system to make statements about cosmology. The order-of-magnitude greater area of RCS-2 (Yee et al., 2007) will allow us to place stronger constraints on the abundance of such massive structures.

A program of detailed wide-field spectroscopy to fully explore the 2319+00 structure is underway. In conjunction with our comprehensive on-going IMACS spectroscopy of a core sample of RCS clusters, selected to uniformly sample mass and redshift space, we will be able to probe the processes which drive galaxy evolution in clusters as a function of cluster mass. As an unusually massive cluster in the early stages of merging, and through comparison with lower-redshift $\sim 10^{15}$ M_⊙ clusters, RCS2319+00 may give us a unique opportunity to identify processes which occur in massive clusters prior to their final assembly.