X-ray morphology

Figure 1: 4-$\sigma$ adaptively smoothed ACIS-S image (contours) overlaid on the Digitized Sky Survey image of A4059. See text for discussion.

For each of the two Chandra observations, the 0.3-8keV data have been background subtracted and corrected for detector and vignetting effects using weighted exposure maps (see §2). The resulting processed images were then combined. Fig. 1 shows contours of the adaptively smoothed Chandra image overlaid on the optical image (from the Digitized Sky Survey) of the cluster. The adaptively smoothed image was derived by smoothing the raw image with a minimum significance of 4-$\sigma$ using the CIAO tool csmooth.

It can be seen that the cluster core has a complex X-ray morphology. The principal morphological features present in these images were previously noted by Huang & Sarazin (1998) and Heinz et al. (2002). The cluster within about $30''$ radius has an hour-glass like structure (or bar) with two broad peaks. The strongest peak, at the center of the cluster, contains further sub-structure with 3 bright regions. The brightest region coincides with the optical nucleus (ESO 349-G010 from the Digitized Sky Survey), although it is clearly not a point source. In fact, we do not detect any pointlike source coincident with the nucleus of PKS 2354$-$35. The second of the principal X-ray peaks is $\sim~15''$ south-west from the center has no optical or radio counterpart. The SW edge of this feature is so sharp as to be unresolved in our adaptively smoothed map; more precisely, inspection of the smoothing length map produced by csmooth indicates that the SW edge of this feature must be less than 3-4arcsec (3-4kpc) across. On larger scales, the X-ray emission is elongated and aligned along almost the same position angle as the major axis of the cD galaxy.

Furthermore, there are two cavities in the X-ray emission to the NW and SE of the cluster center (see Heinz et al. 2002 for a detailed discussion of the statistical significance of these cavities). The NW cavity seen in our Chandra data can clearly be identified with the NW cavity seen in the ROSAT-HRI data. For the more subtle SE cavity, however, the Chandra cavity appears in a different position by about $20''$ from the ROSAT-HRI cavity. In order to directly compare the Chandra image with the ROSAT image, we obtained an image in the $0.3-2$ keV, which is approximately the band covered by ROSAT. This does not change our conclusions regarding the position of the cavity and the discrepancy with ROSAT. Considering the unprecedented high spatial resolution and throughput of Chandra, we conclude that the SE cavity given by Chandra is likely to be real, not the result of statistical fluctuation (see Heinz et al. 2002). After an examination of the raw ROSAT data, we suggest that a $3\sigma$ fluctuation in the photon statistics of the ROSAT image may have led to an incorrect determination of the SE cavity's location. We also note that it is difficult to check the correctness of the ROSAT-HRI aspect solution given the lack of identifiable sources in this short observation.

As pointed out in Heinz et al. (2002), the axis connecting these two cavities lies perpendicular to the central hour-glass like structure but the center of the axis does not coincide with the radio galaxy (see Fig. 2 in § 3.2). Prompted mainly by that fact, Heinz et al. argued that the radio galaxy had interacted with a moving ICM and, hence, that the cavities had been ``blown'' in the north-east direction. As we show in § 3.4, spatially resolved X-ray spectroscopy, as well as the HST-WFPC2 image, provide further support for this hypothesis.