Astron. Astrophys. 322, 66-72 (1997)

2. X-ray data

Cl 0500-24 was observed with the ROSAT/HRI in a pointed observation with an exposure time of 37756 s between February and , 1995.

The positioning of the observation can be checked with a point-like source which coincides with a star of blue magnitude 16.8 at a position , (J2000). From this correspondence we infer that the positional accuracy of the pointing is very good (estimated error 4 arcseconds). The total number of source photons for the cluster Cl 0500-24 is 440.

As the cluster emission is not very strong we follow a two-fold strategy. In addition to the straightforward analysis of the full data set, we try to improve the signal-to-noise ratio by skipping the intervals with enhanced background. In practice, we bin the data to intervals of 100 s. Subsequently, we drop the high background intervals with more than 500 counts / 100 s. The remaining time intervals with the "cleaned data" cover 57% of the original exposure time. The cleaned data set contains 270 source photons.

2.1. Morphology

Because of the limited number of photons from the cluster the observed morphology is influenced by statistical fluctuations. In Fig. 1 we present an image smoothed with a Gaussian filter of arcseconds for a qualitative impression of the morphology. The main maximum of an image smoothed with a Gaussian of arcseconds is at , (J2000). It corresponds to a position between the galaxies N and S (nomenclature according to Giraud (1990)) which are identified by Infante et al. (1994) as the centre of one of two subconcentrations. In an image smoothed with a Gaussian of arcseconds the X-ray maximum is also very close to these galaxies (see Fig. 2). The X-ray emission is elongated southward towards galaxy A. A second maximum of the X-ray emission can be seen in the north-east at , (J2000), close to galaxy # 12 (cf. Infante et al. 1994). There is no extra emission associated with the second optical centre C. All these features are "robust" in that they show up both in the "cleaned" and in the "non-cleaned" data.

Fig. 1. ROSAT/HRI image of the cluster Cl 0500-24. It is smoothed with a Gaussian filter of = 10 arcseconds. The cluster has a clumpy structure with an extension south of the maximum and additional emission in the north-east and in the east. The bright X-ray source in the upper left corner is not associated with the cluster.

Fig. 2. Cluster galaxies assigned to the subclusters N and C by Infante et al. (1994) superimposed on the X-ray contours (smoothed with a Gaussian filter of = 10 arcseconds). Triangles: galaxies assigned to subcluster C, squares: galaxies assigned to subcluster N. The size of the image is the same as in Fig. 1. One tickmark corresponds to 10 arcseconds or 57 kpc. The contour levels have a linear spacing of counts/s/arcmin2. The highest contour corresponds to counts/s/arcmin2, the lowest to counts/s/arcmin2. The central galaxies of the subclusters are marked with N and C, respectively. The X-ray emission is well correlated with the subcluster centred on N, while there is hardly any correlation with the C subcluster. In particular, there is no extra emission at the position of the central galaxy C.

The galaxies are assigned to two subclusters centred on galaxy N and on galaxy C, respectively, by Infante et al. (1994) according to their velocities from Giraud (1990) and Infante et al. (1994). The distribution of "N galaxies" is much better correlated with the X-ray emission than the distribution of "C galaxies" (see Fig. 2). This is a first hint that we see only (or mainly) the N subcluster in X-rays.

In Fig. 3 the X-ray morphology is compared with the galaxy density (Infante, private communication) for all galaxies (while in Fig. 2 only galaxies with measured velocities were included). The contour lines of the luminosity weighted galaxy density show again the two main concentrations around N and C, out of which only N is correlated with X-rays. Another maximum in the galaxy density in the north-east is situated between two X-ray blobs and is probably belonging to the N concentration because there are three "N galaxies" in this region, i.e. again no indication for any emission from the C subcluster.

Fig. 3. Luminosity weighted galaxy densities (contours) from V magnitudes (Infante, private communication) smoothed with a Gaussian of arcseconds are superimposed on the X-ray emission (greyscales, same as the contours of Fig. 2). The size of the image is 2.5x2.5 arcminutes or 860x860 kpc.

About 2.6 arcminutes north-east of the cluster centre is a point-like source (see Fig. 1) at , (J2000). It has a countrate of counts/s. The position of this source corresponds within 10 arcseconds to the radio source PMN0501-2422 which is listed in Griffith et al. (1994) with a 4850 MHz flux density of 154 mJy. Looking for an optical counterpart in the Southern Digitized Sky Survey we find a faint object at about 4 arcseconds distance; it cannot be distinguished whether it is a stellar object or a galaxy.

2.2. Luminosity

The X-ray emission can be traced out to a radius of 2.5 arcminutes (860 kpc at distance of cluster). As summarized in Table 1, we find a countrate of counts/s within this radius. To convert this countrate to a luminosity we assume a typical metallicity of 0.35 solar (Arnaud et al. 1992; Ohashi 1995) and a Galactic hydrogen column density of cm^-2 (Dickey & Lockman 1990). As there is no possibility to derive a gas temperature with the HRI we assume a typical cluster temperature of 4 keV which is also consistent with the relation (Edge & Stewart 1991a; David et al. 1993; White 1996). With these values the luminosity in the ROSAT band (0.1 - 2.4 keV) is erg/s and the bolometric luminosity is erg/s. The errors include the uncertainties of the countrate and an assumed temperature range between 1 and 10 keV.

Table 1. Summary of the X-ray properties of Cl 0500-24 in comparison to other optically rich clusters Cl 0939-4713 and Cl 0016+16. The numbers for Cl 0939-4713 are taken from Schindler & Wambsganss (1996), the numbers for Cl 0016+16 from Neumann & Böhringer (1996). The countrates for Cl 0939-4713 and Cl 0016+16 are converted from PSPC countrates.

2.3. Profile

Because of the limited number of photons and the non-spherical appearance of the X-ray contours, the radial profile of the X-ray emission is not very well determined (Fig. 4). Nevertheless, we try to fit a -model to the surface brightness (following Cavaliere & Fusco-Femiano 1976; Jones & Forman 1984)

where is the central surface brightness, is the core radius, and is the slope parameter. The best fit values averaged over a number of fits with or without "cleaning" and with different binning centred on (J2000) are counts/s/arcmin², arcminutes (30 kpc) and , a very small and a very small core radius. But the 1 errors allow for a huge range of about 0.3 to 1.0 for and 0 to 1.5 arcminutes (0 to 500 kpc) for the core radius.

Fig. 4. Radial profile of the X-ray emission fitted with a -model.

2.4. Mass determination

The parameters of the -model can be used to make a deprojection of the 2D image to derive the three dimensional density distribution. The profile of the integrated gas mass is shown in Fig. 5. Within a radius of 1 Mpc the gas mass amounts to 0.46 . As the emissivity of the gas in the ROSAT energy band is almost independent of the temperature (within the temperature range of 2-10 keV it changes only by 6%), the derived gas density distribution is affected very little by the uncertainty of the temperature estimate. The only uncertainty are local unresolved inhomogeneities or substructure which result in an overestimation of the gas mass. Therefore, the value given above is strictly speaking an upper limit.

Fig. 5. Mass profile of Cl 0500-24. The dotted lines show the integrated gas mass, the solid lines the integrated total mass. Profiles for data with/out "cleaning" and with different binning are plotted. For the different profiles the following parameters are used arcsec, ; arcsec, ; arcsec, ; arcsec, (from thin to thick lines). In the inner region the scatter is quite large because of the limited number of photons. But beyond a radius of 100 kpc the profiles agree well. For comparison with mass determinations by the gravitational lens effect, the dashed lines depict the integrated surface mass density.

With the additional assumption of hydrostatic equilibrium, the integrated total mass can be calculated from the equation

where and T are the density and the temperature of the intra-cluster gas, and r, k, , , and G are the radius, the Boltzmann constant, the molecular weight, the proton mass, and the gravitational constant, respectively. For the temperature we use again the cluster temperature of 4 keV from the relation (Edge & Stewart 1991a; David et al. 1993; White 1996). When using the parameters of the various -model fits with varying binning and with/out "cleaning" we find different mass profiles only in the central part (see Fig. 5). Beyond a radius of about 100 kpc the various mass profiles are in good agreement. The error originating from the uncertainty in the temperature is certainly much larger. As the mass is proportional to the assumed temperature (see Eq. 2) we estimate that the error can amount to .

The results are shown in Fig. 5. At a radius of 1 Mpc we find an integrated total mass of . The errors comprise a temperature range from 2 to 6 keV. This total mass yields a relatively high gas mass fraction of %. The gas mass fraction is decreasing when going to smaller radii. But as the gas mass fraction depends on and is not very well defined (see Sect. 2.3) this decrease may not be very significant. For completeness we also check the extreme (and unrealistic) parameters allowed from the surface profile fit kpc and 1.0. The resulting integrated mass profile is steeper than the ones shown in Fig.5 crossing the other profiles at about 350 kpc. At 1 Mpc the total mass for these extreme parameters amounts to .

The total mass (from the realistic parameters) is even smaller than the luminosity-weighted virial mass of of only the N concentration derived by Infante et al. (1994). (The X-ray mass converted to their km/s/Mpc would be .) This is another hint that the X-ray emitting gas traces the potential of only one subcluster.

For a comparison with mass estimates determined from the gravitational lensing effect, we also calculate the total mass of the cluster, as seen inside a certain angle, basically integrating a spherically symmetric three dimensional mass distribution in cylindrical shells with cylinder axis parallel to the line of sight or integrating the cluster surface mass density outward (see Fig. 5). The total mass inside a circle with radius cluster centre-arc is .

From the point lens model, Wambsganss et al. (1989) found a much larger value: (or a velocity dispersion of km/s for an isothermal sphere model); here the now measured arc redshift of (Giraud 1996) and a distance between centre of lens and arc of 22 arcsec was used. Obviously, the mass from the lens model is considerably higher than the mass obtained here when integrating outward the total mass derived from the X-ray emission even when taking into account the large X-ray mass error of 50%. This discrepancy of a factor 4-5 does not necessarily come from a wrong assumption in one of the two mass determination methods. It can have various reasons. The two mass estimates are not directly comparable because they use different centres: the X-ray mass is centred on the X-ray maximum which is located close to galaxy N while the lensing model has an assumed centre close to 'C'. Furthermore, the two methods to determine the mass are sensitive to different things: the gravitational lens effect integrates all matter along the line of sight, while the X-ray mass determination is most sensitive to the (square of the) highest density regions in the centre of the cluster. Qualitatively the discrepancy of a large lensing mass and a small X-ray mass points in the same direction that Cl 0500-24 consists of two subclusters, of which the lensing "feels" the total mass, whereas the X-ray are indicative only of one subclump. But any more detailed comparison would require a much more detailed model of the cluster lens, which could possibly be obtained from a deep exposure of the cluster and subsequent analysis of arclets and the weak lensing effect.

© European Southern Observatory (ESO) 1997

Online publication: June 30, 1998
helpdesk.link@springer.de