Astron. Astrophys. 336, 855-877 (1998)

4. Results and discussion - the unresolved emission

What is very evident from Fig. 3 is that there appears to be a great deal of unresolved emission enveloping the sources associated with the two central galaxies, IC 4329A (P10) and IC 4329 (P8). As seen earlier, this is not seen at first with the HRI (Fig. 1), though becomes strikingly clear when an adaptive filtering technique is applied (Fig. 2). The fact that this emission is very obvious in the unfiltered PSPC data, though this fact is not mentioned in M95, is very encouraging, and to investigate this aspect of the X-ray emission further, a similar filtering technique was applied to the PSPC data. An image (of pixel size , and over the broad 11-235 range) was formed, and sections of the image were smoothed using progressively larger Gaussians for lower intensity pixels. Pixels of amplitude 1 (2,3,4,5,6,7,8) were smoothed with a Gaussian of FWHM . Again, pixels of amplitude greater than 8 remained unsmoothed, to ensure that the bright point sources were not smoothed into the background. The resultant image is shown in Fig. 9. Again, as in the adaptively smoothed HRI image, a great deal of unresolved emission is apparent, extending to a much greater radius than is seen in the equivalent HRI image ( compared to ). This is not too surprising, even taking into account the factor of deficit in exposure times, as the PSPC is far more sensitive to low surface brightness diffuse emission than the HRI.

Fig. 9. ROSAT (0.1-2.4 keV) PSPC map of the IC 4329A field obtained using an adaptive filtering technique (see text), overlayed on a digitized sky survey image. The contour levels are at 2, 3, 5, 9, 15, 31, 63, 127, 255, 511, 1023 and 2047 ( being cts s-1 arcmin-2) above the background ( cts s-1 arcmin-2).

As the PSPC has some spectral resolution, we are able to investigate the spectral properties of this unresolved emission. What was immediately apparent though, before any rigorous spectral fitting (as described later) was performed, was that this unresolved emission appeared markedly two-component. It was decided at first to investigate the spatial properties of these two components, to try and ascertain their distributions.

4.1. The unresolved emission - spatial properties

This two component nature is apparent in Fig. 10. Shown are two images, each essentially the same as Fig. 9, but extracted from two separate energy bands, the first (soft), extracted from channels 8-41, the second (hard) extracted from channels 52-201. In both cases, images with resolution were formed and these images were smoothed using progressively larger Gaussians for lower intensity regions. Pixels of amplitude 1 (2,3,4,5,6,7,8) were smoothed with a Gaussian of FWHM , and pixels of amplitude greater than 8 were not smoothed, ensuring that the bright point sources were not smoothed into the background).

Fig. 10. ROSAT PSPC maps of the IC 4329A field in the (left) soft (channels 8-41) and (right) hard (channels 52-201) bands, obtained using an adaptive filtering technique (see text), and overlayed on optical images. The contour levels are at 2, 3, 5, 9, 15, 31, 63, 127, 255, 511, 1023 and 2047 ( being (soft) and (hard) cts s-1 arcmin-2) above the background ( (soft) and (hard) cts s-1 arcmin-2).

The results of this procedure are really rather striking, and the two components, the soft and the hard components, show remarkably different properties. The soft component is dominated by IC 4329A and the bright H1-P3 source to the south-west. Strong features are also seen associated with IC 4329, with the two outlying sources, H2/4-P4 and H16-P12, and with H15-P11. The unresolved soft emission though, which is what we are here most interested in, appears almost entirely to the south-east of the bright IC 4329A source, in a roughly semi-circular distribution, `centred' on IC 4329A, and extending out past the two outlying sources H2/4-P4 and H16-P12. The emission appears far from uniform, containing much structure, notably in the south-east direction, perpendicular to the IC 4329A disc (and to the line joining the two outlying sources to IC 4329A). Some evidence exists for a similar (though very much smaller) extension to the north-west of IC 4329A, though much of this emission may be due to IC 4329.

As regards the hard emission, many sources, notably IC 4329A, IC4329, H1-P3, and the two outlying sources, appear as strong sources. The unresolved hard emission however, appears markedly different to the unresolved soft X-ray structure. Firstly, it extends not only to the south-east, as the soft component does, but also to the north-west, and appears roughly circular in nature. Furthermore, in contrast to the soft component, it appears rather uniform and smooth, with little in the way of substructure. Interestingly, and as discussed later, the hard component is not centred on IC 4329A, but instead appears to be centred somewhere between the IC 4329A/IC 4329 pair. Also of note is the fact that this emission is seen to envelope two further galaxies within the galaxy group mentioned in the introduction. These two galaxies, IC 4327 to the north-west of the IC 4329A/IC 4329 pair, and NGC 5298 to the south-west, can be seen at the edge of the hard unresolved emission (Fig. 10).

Structure in the unresolved emission is evident also in the HRI data. Fig. 11 shows contours of HRI X-ray emission superimposed on an optical image of the two central galaxies. A resolution image was extracted from the HRI channel range 6-11, and smoothed with a Gaussian of FWHM . Firstly, as regards the above discussion, significant emission, in what appears to be in the form of a `bridge' (with a curious northern plume), is seen connecting the two galaxies. Secondly, with regard to the emission surrounding IC 4329A, there appears to be an elongation along the disc, pointing towards the two outlying sources, one of which, H2/4-P4, can be seen at the bottom-right of Fig. 11. Also, there appears to be good evidence for an extension perpendicular to the IC 4329A disc, especially, as seen in the PSPC soft band image above, to the south-east. Lastly, as regards IC 4329, the curious spiral-arm-like structure to the X-ray emission is very intriguing.

Fig. 11. ROSAT HRI map of the IC 4329A field obtained using an adaptive filtering technique (see text), overlayed on a digitized sky survey image. Only channels 6-11 are used. The contour levels are at 2, 3, 5, 9, 15, 31, 63, 127, 255, 511, 1023 and 2047 ( being cts s-1 arcmin-2) above the background ( cts s-1 arcmin-2).

4.2. The unresolved emission - spectral properties

It appears therefore, that the unresolved emission within the IC 4329A/IC 4329 system is two-component, and this two component nature varies across the system, the area to the north-west of the IC 4329A disc being hard, the area to the south-east, being hard also, but with a very significant soft component contribution. This north-west/south-east divide was used to investigate the spectral properties of the unresolved emission. Several unresolved emission spectra were extracted, both from the north-west (NW) and the south-east (SE) side, a line through the centre of IC 4329A, at an angle (anticlockwise from north) of 56^o (so that it approximately followed both the IC 4329A disc and the line joining the two outlying sources to IC 4329A), used to separate the two halves. Two main spectra (NW and SE) were extracted from half-annuli centred on IC 4329A, with inner radii of and outer radii of , and were binned to give a signal-to-noise ratio of approximately six in each channel. These two spectra were each further subdivided into three concentrically-extracted spectra, with inner and outer radii of (NW1 and SW1), (NW2 and SW2) and (NW3 and SW3). These six sub-spectra were each binned to give a signal-to-noise ratio of approximately five in each channel. In the extraction of each of these 8 total spectra, data associated with each of the sources were excluded to a radius of . A background spectrum, free of unresolved emission, was extracated from a annulus, data associated with the sources excluded to a radius of . Once these spectra were corrected for background and exposure time, it was possible to gauge the soft and hard-band contributions within each area. In the north-west, the percentage of counts in the soft (0.1-0.5 keV) band, compared to the total (0.1-2.4 keV) number of counts, is seen to be low (%) and constant (to within %) for each of the three spectra (NW1, NW2, NW3). In the south-east however, the soft band contribution is seen to rise sharply from 19.7% for the inner-extracted spectrum (SE1), through 34.8% for SE2, to 53.6% for SE3.

As in the fitting of the point source spectra, spectral models were fitted to these unresolved emission spectra. Dealing first with the north-west spectra, power-law, and Raymond & Smith hot plasma models were first attempted, and the results of the best fits are summarized in Table 5. As can be seen, power-law models (columns 2-6) are able to fit not only the NW spectrum, but also the three separate NW1, NW2 and NW3 spectra, very well, with reduced s of less than unity (in fact more like 0.5 in every case, apart from NW2). Though the resulting parameters at first appear rather different, the errors are rather large, and the spectral fit parameters are quite consistent with one another, i.e. the north-western spectrum appears to remain constant with radius. Fig. 12 (left) shows the 99%, 95% and 68% confidence contours in the absorbing column-spectral index plane for the power-law fit to the full NW spectrum (i.e. Table 5, row 1).

Fig. 12. Model spectrum fits to the NW spectrum. (Left) Gaussian contour levels of 1, 2 and 3 in the spectral index-absorption column plane for the power-law fit to the NW spectrum (Table 5, column 1). (Right) Gaussian contour levels of 1, 2 and 3 in the temperature-absorption column plane for the Raymond & Smith plasma fit to the NW spectrum (Table 5, column 1).

Table 5. Results of fitting power-law and Raymond & Smith hot plasma models to the north-western unresolved emission spectra (see text), as follows: Column 1 gives the spectrum, whether the full NW spectrum or one of NW1, NW2, or NW3, the radially extracted spectra. Columns 2-6 give the results of the best power-law fits; fitted (column 2), fitted spectral index, , where (column 3), and the reduced and number of degrees of freedom (column 4). Columns 7-9 give the results of the best Raymond & Smith hot plasma model fits (with the metallicity frozen at the solar value); fitted (column 7), fitted temperature kT, in keV (column 8), and the reduced and number of degrees of freedom (column 9). Where parameters are seen to `peg' at the highest or lowest allowable values, 2 upper or lower limits are given. Two values for the (0.1-2.4 keV) X-ray luminosity (scaled to account for the emission lost in the `holes' left after the source subtraction procedure) are each given (corresponding to a distance of 64 Mpc) in columns 5 and 6 (for the power-law fits) and columns 10 and 11 (for the hot plasma fits). Cols. 5 and 10 give the `intrinsic' luminosity (correcting both for Galactic and intrinsic absorption), while columns 6 and 11 give the `emitted' luminosity (i.e. correcting merely for the Galactic ).

One should note here the bottom four power law fits, given in Table 5, the implications of which are discussed more fully later. Here, an attempt has been made to see how consistent the NW spectra are with the spectrum of the bright central source IC 4329A. Upon freezing the absorption column at the value obtained in the fitting of the IC 4329A spectrum (22.7 cm^-2, see Table 4), the quality of each fit is seen to remain very good. The values obtained for the spectral index however, though consistent with each other, are somewhat higher than the IC 4329A value.

Moving on now to the Raymond & Smith hot plasma model fits to the NW spectra (columns 7-11), the quality of the fits is again good, in the majority of cases. The absorbing column appears to be small and is consistent with the Galactic value in every case. Unfortunately, as regards the temperature, little definite information can be gleaned from these results, whether the absorbing column is left to optimize or is frozen at the Galactic value. Though the fitted temperatures appear to be high, the fitting procedure often `pegging' the temperature at the highest allowable value, the errors are large. The size of this error region can be seen in Fig. 12 (right), where 99%, 95% and 68% confidence contours are shown in the absorbing column-temperature plane for the Raymond & Smith hot plasma model fit to the full NW spectrum (i.e. Table 5, row 1).

Finally, it is worth noting that usage of a thermal bremsstrahlung model in the spectral fitting gives very similar results to the Raymond & Smith results. The equivalent absorbing column-temperature confidence grid appears essentially identical to Fig. 12 (right). Only in the fitting to the total NW spectrum was a best fit realised - cm^-2, kT KeV, with a reduced slightly worse than for the Raymond & Smith case (0.54, with 9 degrees of freedom).

The emission to the south-west, as mentioned previously, is considerably more complicated, requiring a two-component model to fit the spectra, even adequately (the best one-component power-law and Raymond & Smith hot plasma fits to the SE spectrum result in reduced s of over 5 and 7, respectively). Freezing of a number of the components was necessary, as too many free parameters led to fits that either refused to settle at consistent values or had huge error regions. Following the line of thought touched upon above, the hard component was assumed to be identical to the IC4329A spectrum. The variation of the spectral parameters can be seen in Fig. 13. Here, the 99%, 95% and 68% confidence contours in the soft-component absorption column-temperature plane are shown for the full SE spectrum, i.e. the hard component is frozen to that of IC4329A, and the metallicity of the soft component is frozen at the solar value.

Fig. 13. Gaussian contour levels of 1, 2 and 3 in the soft-component temperature-absorption column plane for the (hard) power-law plus (soft) Raymond & Smith plasma fit to the SE spectrum.

As can be seen, the soft component appears to be very soft, and rather unabsorbed. Fitting of the three sub-spectra SE1, SE2 and SE3, failed to settle at consistent values, and it is only when the absorption column of the soft component is fixed at the Galactic value (4.40 cm^-2), as seems completely reasonable from Fig. 13, that good fits are obtained. Table 6 summarizes these best two-component (power-law plus Raymond & Smith hot plasma) model fits to the SE spectra. In actuality, whatever (within reason) the hard component is frozen at (for instance, the higher index fit to the NW spectra - see Table 5), or whether the soft-component metallicity is frozen or left free, makes very little difference to the resultant fits. Assuming therefore that a hard component somewhat similar (if not identical) to the IC4329A spectrum is present, then a soft component, consistent with a very cool, relatively (if not completely) unabsorbed hot plasma, appears to be present. It is also worth noting here, that the equivalent confidence contour plots for the three sub-spectra SE1, SE2 and SE3, appear more-or-less identical, though with larger-spaced contours reflecting the reduction in statistics.

Table 6. Results of fitting two-component (power-law + Raymond & Smith hot plasma) models to the south-eastern unresolved emission spectra (see text), as follows: Column 1 gives the spectrum, whether the full SE spectrum or one of SE1, SE2, or SE3, the radially extracted spectra. Columns 2 and 3 give the parameters of the power-law component to the best fit; fitted (column 2), and fitted spectral index, , where (column 3). Columns 4 and 5 give the parameters of the Raymond & Smith hot plasma component of the best fit (with the metallicity frozen at the solar value); fitted (column 4), and fitted temperature kT, in keV (column 5). The reduced and number of degrees of freedom are given in (column 6). Two values for the (0.1-2.4 keV) X-ray luminosity (scaled to account for the emission lost in the `holes' left after the source subtraction procedure) are given (corresponding to a distance of 64 Mpc) in columns 7 and 8. column 7 gives the `intrinsic' luminosity (correcting both for Galactic and intrinsic absorption), and column 8 gives the `emitted' luminosity (i.e. correcting merely for the Galactic ), the figures in brackets indicating the percentage contribution to this luminosity from the Raymond & Smith hot plasma (i.e. the soft) component.

Models incorporating two Raymond & Smith hot plasmas either failed to converge or, if they did, the results obtained were unphysical or had uncomfortably large error regions. It is worth noting that a model incorporating two thermal bremsstrahlung models, both with absorbing columns frozen at the Galactic value, is able to fit the SE spectrum. A very low and a very high temperature plasma is required.

As we have seen therefore, both a hard and a soft component to the unresolved emission appear to exist around IC 4329A, the hard component lying, rather smoothly and symmetrically distributed, around the IC 4329A/IC 4329 pair, the soft component lying almost entirely to the south-east of IC 4329A. As we have regions of emission where only the hard emission appears to exist (to the north-west), it is easier to deal with this aspect of the emission first. Only after this, can we move on to discuss the soft emission component.

4.3. Discussion - The hard component

One possibility as to the origin of the hard residual component, as alluded to (though not explicitly stated) in the previous sections, is that it could be due to the `wing' emission from the very bright central source. This idea is supported by the fact that the spectrum of the hard residual emission (both the NW emission and the hard component of the SE emission) appears very consistent with the IC 4329A spectrum.

To investigate this question further, a radial surface brightness profile (over the channel range 11 to 235) of the central emission was formed. This is shown in Fig. 14 (with the region between and magnified in the inset), and contains many features, as follows: Firstly, the data points show the radial distribution (centered on IC 4329A) of the total X-ray emission, split into radial bins, with the region inside split into radial bins. The long dashed line represents the PSF of a point source with a spectrum best fitted by a power law of photon index 1.28, absorbed by a column of 2.27 cm^-2, i.e. it represents the radial distribution of the emission from IC 4329A, assuming it to be a point source. This was formed by summing together model PSFs (for photons of varying energies), each normalized according to the spectrum of IC 4329A (i.e. normalized according to Fig. 7). The dotted line indicates the radial distribution of the true (i.e. non-diffuse) background. This was formed by firstly eliminating from an image all of the sources to very large radii (5 the FWHM of the PSF). This resulted in an image where approximately none of the unresolved, extended emission, visible in Fig. 9, was included. A polynomial fit was then used to interpolate across the `holes', and the image was then heavily smoothed (with a Gaussian of FWHM ). A radial profile of this image (again centered on IC 4329A) gives the indicated dotted line. The dash-dotted line indicates the level of the `diffuse' emission, the unresolved emission visible in Fig. 9. This was formed by removing sources from an image to radii of twice the FWHM of the PSF. A mask image was created in the same way with zero values at the positions of these holes, and values of unity everywhere else. A radial profile (of binsize ) was formed from the source subtracted image, and this was normalized for the area lost in the `holes', by dividing it by an equivalent profile of the mask image.

Fig. 14. Surface brightness profile of the (0.1-2.4 keV) flux about the centre of IC 4329A (the region between and is magnified in the inset). Data points show the total X-ray emission profile, the long dashed line represents the PSF of a point source with spectral properties identical to that of IC 4329A, the dash-dotted line indicates the level of the `diffuse' emission, all point source emission having been removed, and the dashed line indicates the level of the `true' background, all point source and `diffuse' emission having been removed (see text).

Several features are evident within this figure. The central bright point source is very obvious, but also visible are the two other bright sources, the IC 4329 source (H5-P6), at , and H1-P3, at . The emission is seen to fall to the background level at large radii (), but there is a very notable deviation from this level closer in. Between and (and possibly up to ), the emission is seen to be significantly enhanced with respect to the background, and there appears to be two components to this enhancement. One is the presence of some underlying, unresolved, perhaps truly diffuse emission, as indicated by the dash-dotted line. The second is the presence of point sources, notably the bright sources P6, P9 and P16, visible in the radial profile figure, between and , as data points lying above the residual emission profile. Part (or perhaps all) of the enhancement at is due to sources P3 and P4.

With regard to the unresolved emission therefore, it is the dash-dotted, unresolved emission profile, and the possible contribution to this from the dashed IC 4329A point source that are of interest. It can be seen from Fig. 14 that the idealised point source model of IC 4329A does contribute to the unresolved emission within the inner annulus (NW1 & SE1: ), though beyond this, in the outlying regions, the expected contribution drops rather quickly. The fact that the X-ray source associated with IC 4329A may not be an idealised point source however, as is suggested by Fig. 14, where some deviation from the (spectrally corrected) point source PSF model can be seen), may boost the true wing contribution in the outlying regions from this source, to values perhaps in line with that of the residual emission.

Perhaps therefore, the hard residual emission is due to the wings of IC 4329A. The hard residual emission is well fit by the IC 4329A spectrum, and the number of counts expected from the IC 4329A wings may be close to being in agreement with what is observed (though this is unlikely). Furthermore, the hard residual emission appears approximately circular, and both smooth and regular, as one would expect, if this emission were just due to the wings of a bright point source. Unfortunately, there is one aspect of the hard residual X-ray emission that is very difficult to reconcile with it being due to the IC 4329A wings - the emission is not centred on IC 4329A. Instead, it appears centred on a point midway between IC 4329A and its bright elliptical companion, IC 4329, a point coincident with the `bridge' of X-ray emission visible in the HRI data (Fig. 11). This fact, that the emission is not centred on IC 4329A, makes it very unlikely that the hard residual emission is due entirely to the IC 4329A wings.

Instead, the above points are very suggestive of the emission being due partly to the `wings' of IC 4329A, and partly to hot gas surrounding the IC 4329A/IC 4329 pair. A significant fraction of the hard residual emission will be due to the IC 4329A wings, but this only makes a marked contribution at smaller radii, dropping to a level below half that of the total hard residual emission beyond . The fact that the spectral properties of the hard residual emission appear indistinguishable from those of IC 4329A may at first seem strangely coincidental. In fact, this is not too surprising as the spectral capabilities of the ROSAT PSPC are not too good for harder spectra, and, to the PSPC, an IC 4329A-type spectrum and a hot (5-10 keV) plasma spectrum appear almost identical, as is apparent in the spectral fitting results (Table 5).

Nevertheless, we have attempted to further analyse the north-western unresolved emission spectra, in an attempt to extract any information regarding the hot gas component that may exist. A two-component model, comprising of a power-law component, representing the emission from the IC 4329A `wings', and a Raymond and Smith hot plasma component, representing the diffuse hot gas component, was fitted to the full NW spectrum and to the three sub-spectra NW1, NW2 and NW3. Here, the power-law component parameters were frozen at the IC 4329A values ( cm^-2, ), and the Raymond & Smith component absorption column was frozen at the Galactic value ( cm^-2). Examination of the resultant Raymond & Smith component normalization-temperature parameter space reveals a similar effect in all but one of the cases; In the fits to the full NW spectrum, and to the inner NW1 and the outer NW3 spectrum, the temperature of the hot gas component is seen to be very ill-defined, with large, and highly irregularly-shaped confidence contour levels. Furthermore, no significant improvements in the fit quality are seen, reduced s (and number of degrees of freedom) for these three fits being: NW 0.86(9), NW1 0.63(5) NW3 0.30(3) (compare these values with the one-component Raymond and Smith values given in Table 5).

In the case of the NW2 spectrum however, a significant improvement is seen in the fit quality (reduced (n.d.o.f) - compare with Table 5). Also the hot gas component normalization-temperature parameter space is seen to be quite well-defined, with (at least at the 68% confidence level) reasonably small and regularly-shaped contour levels (see Fig. 15). The best fit to the NW2 spectrum (indicated by the dot in Fig. 15) contains a IC 4329A-like component and a hot gas component with a temperature of 1.53 keV.

Fig. 15. Two-component model spectrum fit to the NW2 (middle) spectrum. Gaussian contour levels of 1, 2 and 3 in the hot gas (Raymond & Smith) component temperature-normalization plane for the (IC 4329A-like) power-law plus Raymond & Smith plasma fit to the NW2 spectrum (see text).

We believe that we are only able to see this hot gas component, even reasonably, within the NW2 fit, because it is only at this radial distance from the central bright source that the contamination from the IC 4329A `wings' drops sufficiently enough to allow the hot-gas component (which is itself dropping with radius) to become visible. In the NW1 spectrum, although the hot gas component is large, the IC 4329A wing component is huge, and the contamination is too large to allow a reasonable determination of the hot gas component parameters. In the NW3 spectrum, even though the IC 4329A wing component has become extremely low, the hot gas component is now very low itself, and the statistics involved are insufficient for a reasonable spectral determination.

A great range in temperature is able to fit the hot gas component of the full NW spectrum. We assume here though, that the temperature of the entire hot gas component is equal to the temperature found for the NW2 spectrum (1.53 keV). Fitting of the full NW spectrum, assuming this temperature, does lead to a very good fit indeed, with a reduced of 0.52 (with 10 degrees of freedom).

Mean physical properties for this north-western hot gas can be inferred from the above results if we make some assumptions about the geometry of the emission. Here we have assumed the simple geometry of the north-western hot gaseous emission being hemispherical with a radius of (in actuality, only a rough approximation to the gas properties can be calculated here and assumption of a slightly different radius gives rise to very similar results).

Using the volume derived for this hemispherical `bubble' model, the fitted emission measure (where is the `filling factor' - the fraction of the total volume V which is occupied by the emitting gas) can be used to infer the mean electron density, , and hence the total mass , thermal energy and cooling time of the gas.

Performing these calculations, after first accounting for the extra emission lost in the `holes' left after the source-subtraction procedure, one arrives at approximate values to the physical properties of the north-western hot gaseous emission as follows; X-ray luminosity (0.1-2.4 keV; intrinsic) 2.6 erg s<sup>-1</sup>; , 1.8 cm<sup>-3</sup>; , ; , 8.1 erg; , 99 Gyr. In the later discussion, we will see that no spectral information regarding the hot gaseous emission to the south-east can be obtained. This is primarily because of this emission being severely contaminated, not only, as it is to the north-west, by the IC 4329A `wings', but also by the very soft emission discussed earlier. If one makes the simple assumption that the properties of the hot gas to the south-east are identical to those to the north-west, a quite valid assumption, given the physical appearance of the emission and the level of accuracy we can attain, then the values of the physical properties for the total hot gas surrounding the IC 4329A/IC 4329 system become double those quoted above (except for the mean electron density , and the cooling time , which remain unchanged).

Remembering that, as discussed in the introduction, the two central galaxies lie close to cluster A3574, and are part of a loosish group of seven galaxies, it is useful to compare here the properties of this hot gaseous emission component with the general X-ray properties of groups and clusters.

As regards the comparison with clusters, though the temperature of the hot gas may be comparable, the luminosity ( erg s^-1) is extremely low. Clusters typically have X-ray luminosities greater than erg s^-1 (Edge & Stewart 1991; Yamashita 1992; White 1996), i.e. two orders of magnitude greater than is observed around IC 4329A/IC 4329, and it is therefore very unlikely that we are seeing emission associated with the A3574 cluster. In the following discussion, we will primarily deal with the comparison of the present results with the properties of galaxy groups.

The general X-ray properties of galaxy groups have recently been published, both of Hickson's (1982) compact groups (HCGs) (Ponman et al. 1996), and of other, poor groups (Mulchaey et al. 1995). The group containing IC 4329A and IC 4329 appears rather loose and contains perhaps seven members (a rather large number). A good optical image of the group can be found in Fricke & Kollatschny (1996).

The X-ray luminosity of the hot gaseous emission is low, but in no way uncomfortably low when compared to the results of Ponman et al. (1996) and Mulchaey et al. (1995). Many other galaxy groups at the same distance are seen with very similar hot gas X-ray luminosities (i.e. the emission from the member galaxies having been removed, as has been done here). Furthermore, the ratio of the diffuse hot gas X-ray luminosity to the total blue luminosity of the member galaxies () is seen to be very typical of groups. As regards how the X-ray luminosity relates to the spiral fraction (the fraction of the group member galaxies that are spiral, as opposed to elliptical galaxies), the hot gas around IC 4329A/IC4329 is again, in no way unusual. The spiral fraction of the IC 4329A group is actually very high, IC 4329 being the only non-spiral member, and though Ponman et al. (1996) find only a very weak correlation between spiral fraction and , Mulchaey et al. (1995) find a somewhat stronger correlation, all of the groups they detect having an extended X-ray emitting intragroup medium having a high percentage of early-type galaxies. The emission within the IC 4329A group (a high spiral fraction group) is of a low luminosity, in agreement with the work of Mulchaey et al. (1995) (and of Ponman et al. (1996) though only a far weaker correlation is seen here).

As regards the temperature of this hot gas however the emission may be unusual. As discussed earlier, it has proved very difficult to separate this emission from the wings of the bright central source associated with IC 4329A, and a good deal of caution should be taken here. Indeed, the present case is very similar to the case of HGC 4, as, while Saracco & Ciliegi (1995) suggest the existence of an extended component, Pildis et al. (1995) find that it is impossible to detect any extended diffuse component on account of the emission from the central active galaxy in the group being so strong.

Our tentative value though, for the temperature of the hot gas ( keV) appears to be high when compared to the values of Ponman et al. (1996), which range, in the majority of cases, from 0.6-1 keV, and also when compared to the values of Mulchaey et al. (1995). Though a large error on the temperature does exist (see Fig. 15), it does appear that the temperature is constrained to be greater than 1 keV. The hot gaseous emission seen surrounding the IC 4329A/IC 4329 pair appears therefore, to be quite hot when compared to typical groups, and as such, it sits rather uncomfortably on the temperature relationship of Ponman et al. (1996), being of too low a luminosity for its temperature (or too hot for its ).

For rich clusters of galaxies (Edge & Stewart 1991), there appears to be a good correlation between and the optical velocity dispersion . Extending the Edge & Stewart (1991) cluster relationship down, one would expect, in the case of groups, the higher velocity dispersion systems to have higher X-ray luminosities, and indeed, this does appear to be the case (Ponman et al. 1996; Mulchaey et al. 1995); higher- groups are brighter in X-rays. The IC 4329A group of galaxies has a rather large value of , approximately 390 km s^-1, calculated using the Galaxy-corrected recession velocities given in the Third Reference Catalogue of Bright Galaxies (RC3; de Vaucouleurs et al. 1991). This value is greater (though within the range of) Hickson et al.'s (1992) study of 100 HCGs, which have a median value of 200 km s^-1. It is also larger than typical values for loose groups (208 km s^-1 Geller & Huchra 1983; 183 km s^-1 Maia et al. 1989), though much less than for rich clusters, where, for example, Zabludoff et al. (1990) established a median for 65 Abell clusters, of 744 km s^-1. Interestingly, there is good indication of a correlation between and the group emission temperature (Ponman et al. 1996), a correlation which appears to extend from the poorest of groups to the richest of clusters. IC 4329A, on account of its high temperature and its high velocity dispersion, lies directly on Ponman et al.'s (1996) regression line, between the groups and the clusters. The hot gaseous emission within the IC 4329A group however, is not as X-ray luminous as the group's velocity dispersion would imply, and it lies below the regression line fitted to the data of Ponman et al. (1996). It should be stressed however, that it does lie well within the scatter of the data, and cannot be called in any way, unusual.

In terms of the amount of gas present, the IC 4329A group appears again, quite normal. Its estimated gas mass of is very typical of groups (Mulchaey et al. 1995), lying on the low side of average, but well within the scatter.

In summary then, part of the emission detected surrounding IC 4329A and IC 4329, and extending out to two other galaxies in the group, IC 4327 and NGC 5298, may well be hot gaseous emission associated with the galaxy group. If it is, then the estimated temperature of the hot gas ( keV, quite hot for groups) agrees well with the (high) velocity dispersion of the group galaxies. Its X-ray luminosity however, appears to be lower than average (for its velocity dispersion and/or temperature). Similarly, the mass of gas involved is perhaps lower than average. In no way though, do any of the physical properties of the emission indicate that this IC 4329A group emission is very unusual.

Interestingly, the single galaxy group in Ponman et al.'s (1996) survey most similar to the IC 4329A group is HCG 48. This group has a high velocity dispersion as well, and a correspondingly high temperature ( keV) is found for the hot gas. It's X-ray luminosity however, is low, and it is seen to lie significantly below the -temperature relationship of Ponman et al. (1996), as do the present IC 4329A group results. Importantly, like the IC 4329A group, HCG 48 itself lies within a larger cluster, A1060, and it appears to be falling into this cluster. It may well be the case that we are seeing in both cases, the HCG 48 case and the IC 4329A case, the effects of gas stripping from these groups, as they fall through their respective clusters centres. This reduction in hot group gas would have the effect of reducing the observed (and estimated gas mass), as is observed.

4.4. Discussion - The soft component

The soft residual component reveals itself in several different ways. It can be seen in the soft band (Ch. 8-41) image of Fig. 3, where the emission to the south-east of the central bright IC 4329A source is seen to be significantly enhanced with respect to the emission to the north-west. The most striking depiction of the residual soft component however, is in Fig. 10, where the adaptively filtered soft band image shows what appears to be a roughly semi-circular distribution of diffuse emission, `centred' on IC 4329A, and extending out, though almost entirely to the south-east, to a radius of perhaps , past the two outlying sources H2/4-P4 and H16-P12. As seen, the spectral properties of this emission indicate that it is very likely to be due to a very low temperature ( keV) plasma, absorbed by a column consistent with that out of own Galaxy.

What is this soft emission then? One question that needs to be addressed first is, is this feature actually associated with IC 4329A? Perhaps it is a foreground feature. Perhaps it is very local, within our own Galaxy, or even within our local bubble. The spectral information present in the PSPC data is very useful here, as in Fig. 13, where the 1, 2 and 3 Gaussian contour levels in the soft-component temperature-absorption column parameter space are shown, it can be seen that this component's spectrum appears to be best fit with a plasma spectrum with an absorption column equal to, or greater than, the column out of the Galaxy, i.e. the feature is very likely to be extragalactic, if no intrinsic absorption is present, a likely fact, given the amorphous, diffuse nature of the feature. The feature therefore, appears likely to be extragalactic. Is it associated with IC 4329A though? A couple of points regarding the structure of the feature are very suggestive, we believe, of this being the case. Firstly, a good correlation is seen between this soft feature and the disc of IC 4329A. Secondly, within this feature, a large plume of emission is visible lying directly perpendicular to the IC 4329A disc, to the south-east (see Fig. 10), i.e. along the galaxy's minor axis. This plume is in the same direction as the 2.3 GHz radio feature discovered by Blank & Norris (1994). Furthermore, the HRI image (Fig. 11) shows good evidence for an extension in this south-eastern direction also. Lastly, the soft feature, although lying almost exclusively to the south-east of IC 4329A, is very symmetric with respect to the minor axis of IC 4329A, i.e. the level of emission seen to the north-east of the plume discussed above appears very similar to that to the south-west.

Assuming therefore, that the emission is connected in some way to the IC 4329A/IC 4329 system, then the fact that it is soft and intrinsically, almost completely unabsorbed, is very reminiscent of the galactic winds seen within the halos of many different systems (notably starburst systems) (Watson et al. 1984; Fabbiano 1988; Pietsch 1992; Strickland et al. 1997; Read et al. 1997). Indeed, many multi-wavelength studies of this system indicate that a wind is possibly present (see Sect. 1). Both Colbert et al. (1996) and Mulchaey et al. (1996) detect [NII ] features extending along the minor axis, to (3 kpc) on both sides of the nucleus, and both sets of authors believe that these features represent an outflow from the nucleus similar to the superwinds commonly seen in edge-on infrared-luminous galaxies (e.g. McCarthey et al. 1987; Armus et al. 1990).

There are however, many differences between this feature and the `classic' winds seen in such nearby starburst systems as M82 and NGC 253. Firstly, the feature is very large, extending to a radius of perhaps , which at the assumed distance of 64 Mpc, corresponds to a size of almost 200 kpc. This is over an order of magnitude greater than the classic features mentioned above. Secondly, it appears almost entirely on one side only of the system, to the south-east, and thirdly, it is rather luminous when compared to the classic starburst winds.

As for the hot gaseous emission in the previous section, mean physical properties for this cool, soft-component SE gas can be inferred, if we make some assumptions about the geometry of the emission. We could consider the emission to be in the form of a hemisphere, purely in the south-eastern direction. This is physically unreasonable however, and upon closer inspection, the best simple estimate for the geometry of the soft residual emission appears to be a sphere, centred to the south of IC 4329A, with a radius of (150 kpc at the assumed distance of 64 Mpc). Again, only a rough approximation to the gas properties can be calculated here and assumption of either model gives rise to very similar results.

Using the volume derived for this spherical `bubble' model, the fitted emission measure can again be used to infer the mean electron density, , the total mass , the thermal energy and the cooling time of the gas, as follows; X-ray luminosity (0.1-2.4 keV; intrinsic) 8.7 erg s^-1; , 3.4 cm^-3; , ; , 7.0 erg; , 2.6 Gyr.

Comparison of these values with values obtained for the soft extended features seen in several nearby spiral systems (including the `classic' winds in NGC 253, M82, NGC 3079 and NGC 3628 (Read et al. 1997)), leads to the conclusion that the soft emission seen here, to the south-east of NGC 4329A, is even more unusual than at first thought. Not only, as mentioned above, is it much larger and brighter (both by at least an order of magnitude) and observed on only one side of the system, but it appears to be far less dense (again by over an order of magnitude), and extremely massive (perhaps a hundred times moreso than for the NGC 253/M82 features). One must be careful here though, as a factor of is still involved in the above results, and this could have quite a large effect on the results.

The soft feature seen here therefore, appears to be completely unlike the classic starburst winds (except in terms of the temperature, which appears to be consistent with the classic wind temperatures (Read et al. 1997). What this feature bears much more of a resemblance to, are the features seen associated with the ultraluminous far-infrared galaxies (FIRGs) Arp 220, NGC 2623 (Read & Ponman 1997) and NGC 6240 (Fricke & Papaderos 1996). Here, large, soft, outlying features are seen to one side, and one side only, of these high-luminosity systems, systems believed to be at the most energetic stage in the evolution of a merger between two galaxies, i.e. at the point where the nuclei of the two galaxies merge. However, the features seen in these systems, though more similar to the IC 4329A feature than the `classic' wind features, are, in some respects, still rather different. Though they are spectrally soft ( keV), and large ( kpc), much larger than the `classic' wind features, they are still significantly smaller than the IC 4329A feature. Furthermore, though they are far brighter than the `classic' wind features, they still have luminosities perhaps an order of magnitude smaller than the IC 4329A feature. It must be borne in mind here though that, as mentioned previously, IC 4329A is a superluminous Seyfert galaxy, with an X-ray luminosity some two orders of magnitude greater than that of Arp 220 (Read & Ponman 1997). In this respect, the IC 4329A `wind' feature makes up a far smaller fraction of the total X-ray luminosity than in any of the FIRGs or starbursts.

The most striking similarity between the IC 4329A soft feature, and the FIRG soft features however, is perhaps the strangest facet of their emission - their one-sided nature. Why does the structure seen in IC 4329A appear only on one side of the system? As regards the FIRGs, and as is discussed in Read & Ponman (1997), the answer may lie in the fact that these systems are rapidly evolving. MacLow & McCray (1988) numerically modelled the growth of superbubbles: large thin shells of cold gas surrounding a hot pressurized interior, - essentially a galactic wind, or at least, the progenitor of, in various stratified atmospheres. They discovered that superbubbles blow out of the H I layer (i.e. they depart from the `snowplow' phase and evolve into the `blowout' phase), and are then able to move out into the inter-galactic medium at velocities of several thousand km s^-1 (see Heckman et al. 1993). This, MacLow & McCray (1988) find, occurs when the superbubbles have a radius of between one and two scaleheights. What they also find however, is that these bubbles will blow out on one side only of a disc galaxy if the bubble centres are more than 50-60 pc from the centre of the disc of the galaxy. Now, in a (relatively) non-evolving system, such as M82 or NGC 253, the starburst is seen to be very symmetrically positioned with respect to the galactic disc, and bipolar structures are seen in the X-ray (Watson et al. 1984; Fabbiano 1988; Pietsch 1992; Strickland et al. 1997; Read et al. 1997). As discussed in Read & Ponman (1997) though, in the rapidly-evolving ultraluminous merging systems Arp220 and NGC 2623, the central burst is highly unlikely to be so centrally positioned with respect to the quickly-moving and highly distorted gaseous components, and the direction of steepest pressure gradient (along which the bubble will most rapidly expand) will now no longer be along both directions of the disc's minor axis (as is the case in M82-type systems), but will be in just one direction (as is predicted by MacLow & McCray 1988), hence the observed one-sided blowout.

In the present case, it appears unambiguous that some sort of interaction between IC 4329A and IC 4329 is taking place: Both galaxies are close together and are part of a looser group of seven galaxies. IC 4329 appears to be a shell galaxy, a signature of interaction. Photographic co-addition and contrast enhancement of four UKST IIIa-J survey plates indicate the presence of low surface brightness features around IC 4329A (Wolstencroft et al. 1995), again suggestive of an interaction. Evidence for possible interaction-induced activity is also seen, as discussed earlier, in the [NII ] studies of Colbert et al. (1996) and Mulchaey et al. (1996), and in the radio studies of Unger et al. (1987) and Blank & Norris (1994). Good evidence for an interaction is present also in the ROSAT HRI data. As seen previously, Fig. 11 shows what appears to be a `bridge' of emission connecting the two galaxies. This type of feature is believed to have been seen in several different situations, though importantly, always in association with galaxies in the process of interacting. For instance, a near-identical feature is seen between the pair of interacting galaxies, NGC 3395/6 (Read & Ponman 1997), and at the contact region between the two galaxies making up the Antennae system (Read & Ponman 1995). Furthermore, a similar feature is seen within the galaxy group HCG 92 (Stefan's Quintet) (Pietsch et al. 1997). It is thought probable that all of these features, including the present HRI feature between IC 4329A and IC 4329, are due to shocks resulting from strong galaxy interactions. So, though there is very strong evidence of an interaction occurring between IC 4329A and IC 4329, whether, as regards the above one-sided blowout discussion, this interaction is strong enough to displace the disc of IC 4329A with respect to the `central' wind source, is unclear.

There is another possible explanation of this very soft emission related to the harder group emission discussed earlier. Remember that the high velocity dispersion of the IC 4329A group galaxies and the high fitted temperature to the X-ray emitting gas should point to the X-ray emitting gas having a high luminosity. This is observed however, not to be the case, and it was suggested that `stripping' of some of the group gas as the group moves through the surrounding A3574 cluster medium may be taking place. This would result in a reduction in the X-ray luminosity and estimated mass of the group gas, as we see. As discussed earlier, this idea is given some credence by the fact that HCG 48, a group very similar to the IC 4329A group in terms of its high velocity dispersion and high X-ray temperature, combined with a low X-ray luminosity (Ponman et al. 1996), is also observed to be a group within a cluster (A1060), and is thought to be falling through the centre of the cluster.

Could this soft emission, seen to the south-east of the IC 4329A/IC 4329 pair, be this stripped group gas? The addition of the X-ray luminosity and gas mass of this soft component to the equivalent values for the harder `group' component would certainly push the total group gas and to values such that the entire properties of the IC 4329A group, taken as a whole, would appear completely normal and typical. As mentioned in the introduction, this group lies at the easternmost edge of the Hydra-Centaurus supercluster. Though no tangential velocities for these galaxies are obviously available, it could perhaps be safe to assume that the IC 4329A group would be moving tangentially in a general westward direction, towards the centre of this superstructure. Any gas stripped from this group would appear to the east of the group, as is observed in the case of the very soft emission.

Whether the temperature of this soft gas is consistent with this idea, is difficult to say. One might have expected any stripped gas to be harder than its equivalent group emission, though it might be possible that compression has lead to cooling. In fact, the (roughly) estimated cooling time of this gas is fairly short (less than 3 Gyr).

This idea, though interesting, may not be correct however, as perhaps the same morphology should be seen in the case of the hard group emission. Though this hard emission is severly contaminated by the IC 4329A wings, it does appear, on account of the fact that the total hard emission is centred someway between IC 4329A and IC 4329, that the hard group emission lies more predominantly to the opposite side of IC 4329A than the soft emission.

© European Southern Observatory (ESO) 1998

Online publication: July 27, 1998
helpdesk.link@springer.de