The ROSAT PSPC observations of the edge-on NGC 3079 have shown a complex emission in and outside of the plane of the galaxy. The global X-ray luminosity in the ROSAT band is erg s-1 (d=17.3 Mpc), higher ( factor 10) than what is observed in other edge-on late-type galaxies of similar optical luminosity (for example in NGC 4631, Vogler and Pietsch 1996; NGC 4565, NGC 5907, Vogler et al. 1995). The emission has been separated into three components of similar brightness - disk, central region, and halo - that will be discussed separately in the following subsections.
Two explanations for the enhanced X-ray luminosity of NGC 3079 can be put forward that are triggered by anomalies of the galaxy measured in other wavelength regimes:
While it is not clear which of these types of activity (or even both?) is responsible for the enhanced X-ray emission, both may have been triggered by galaxy galaxy encounters in the NGC 3079 galaxy group that also contains NGC 3073 and MCG 9-17-9 (see Sect. 4.5).
The low X-ray flux associated with NGC 3079 has not allowed us to properly measure the spectral characteristics of the emission components, therefore a physical interpretation of the results is not possible. As shown in Sect. 3.3.2, a power law model with relatively steep spectral index or a soft thermal bremsstrahlung could fit the data equally well. However, the power law index is in all cases very steep () and also the temperature of the thermal bremsstrahlung spectra is rather low if we compare them for instance with spectral fits to individual X-ray binaries or the integral bulge spectrum (primarily X-ray binaries) found with the ROSAT PSPC for M31 ( or temperatures above keV, Supper et al. 1997). It is likely that the simple models that we have assumed here for lack of statistics are inadequate to represent the more complex characteristics of the X-ray emission in these objects (for example a population of individual sources plus a multitemperature or non-equilibrium interstellar medium). Dahlem et al. (1998) have shown that a correct interpretation of the spectra of galaxies requires data in a larger energy range than provided by the PSPC data alone, even though there are still residual ambiguities in the correct model to be used. In particular they find in several instances that two gas phases are needed in addition to a power law component at high energies. The lack of spatial resolution however does not allow a proper investigation of the location of these individual components, that cannot therefore be unambiguously identified with the separate sources of emission in galaxies. In this respect, only future X-ray missions, that combine high sensitivity with good spectral and spatial resolution over a large energy range will allow a significant step forward in our understanding of the X-ray properties of galaxies. From the present data however we could be tempted to assume that indeed a multiphase interstellar medium is present and interpret the (very crude) indication of the hardness ratios in terms of temperature variations in the disk versus central region, and assign a higher temperature to the emission coming from the central source coincident with the super-bubble (see Sect. 4.2). This would be analogous to the results from the more detailed analysis of NGC 253, for which the central starbursting region is harder than the surrounding disk (Dahlem et al. 1998).
In either case however a large low energy absorption is suggested, well above the line-of-sight H I column density. This result is not surprising since the disk is embedded in an extended H I disk (Irwin et al. 1987). Only at large distances from the plane could local obscuration be neglected, and indeed the (though extremely uncertain) results for the halo region do not require large absorbing columns.
4.1. X-ray emission from the disk of NGC 3079
Several bright sources are detected in the galaxy plane of NGC 3079 (Table 5), with luminosities higher than 1038 erg s-1 , i.e. well above the Eddington luminosity for a 1 accreting object. As already discussed, these could be real point-like sources or could be local enhancements in a more diffuse emission. However, the optical image of the galaxy shows a very patchy appearance, and there are a lot of giant H II regions in the plane (Veilleux et al. 1995). Over-luminous H II regions, relative to those observed in our galaxy and in other normal spirals like M101, for which erg s-1 (Williams & Chu 1995), have already been observed in relation with the presence of enhanced star formation ( in the Antennae, Read et al. 1995, Fabbiano et al. 1997). As discussed above, the integral spectrum of the disk region is rather soft compared to bright X-ray binary spectra and can be taken to favor the over-luminous H II region origin of the emission. Deep observations with improved spatial and spectral resolution are needed to solve the origin of the disk emission.
4.2. X-ray emission from nuclear super-bubble of NGC 3079
NGC 3079 hosts the most powerful example of a wind-blown bubble to the east of its nucleus. Recent spectroscopic studies in the near infrared and optical however argue against a starburst origin for the powering of the wind, and suggest an active nucleus with somewhat exceptional properties (Hawarden et al. 1995). In this case, one would expect a wind from an active galactic nucleus (AGN, see Heckman et al. 1986) - how can X-ray observations help understanding the nature of this source?
Emission in the center of NGC 3079, coincident with the H loop (Ford et al. 1986) and the kiloparsec scale radio lobes (de Bruyn 1977, Duric & Seaquist 1988) is clearly detected in the HRI data. The source appears complex (see Fig. 11). A possible unresolved component, apparently shifted by about to the east from the position of the optical nucleus (marked by the cross in Fig. 11) could be - within the systematic position errors - coincident with the nucleus. This source, if identified with the hard spectral component measured by ASCA (see discussion in Sect. 3.4) that has been associated with the emission of an highly absorbed AGN, is brighter by a factor of 3 - 4 than the extrapolation of the ASCA spectrum into the ROSAT band would suggest. If the positional displacement however is real, it could represent the brightest and hottest part of the out-flowing material that is not shielded by the absorbing material in the central disk. Such a displacement to the east is expected in such a scenario due to the high absorption in the inner disk of NGC 3079 and the disk orientation. The inclination of and the orientation of the galaxy (the western side of the disk is closer to us than the eastern side) lead to reduced absorption in the direction to the nuclear area in the east and high absorption in the west ( in the inner disk is cm-2 according to Irwin & Seaquist 1991). The high absorption may also cause that no X-ray or H emission coinciding with the western component of the radio jet is detected.
A more extended component elongated to the north is possibly connected to a feature seen at larger scale (Fig. 11), and may indicate a preferred channel to fuel the emission in the halo.
The HRI attitude solution carefully derived in Sect. 2.1 clearly contradicts the solution put forward by Dahlem et al. (1998) which is shifted to the north by and - if correct - would exclude a connection of the central X-ray emission with either the nucleus or the nuclear super-bubble of NGC 3079.
The small extension of the bright central X-ray emission region compared to the PSPC spatial resolution coupled with it's low flux prevents us from reliably determining spectral properties. In the innermost 30" radius, that correspond to the minimum cell size for count extraction that is reasonable for PSPC data, the HRI resolution indicates that at least three different components are present, all contributing to the overall emission: the disk with two non nuclear point-like sources, the nuclear bubble and the unresolved point source. Neither of these dominates, with the possible exception of the disk emission, which can in part be subtracted by choosing a local background (see Table 6), and therefore a spectral analysis should take all of them into account. This is clearly unrealistic given the small number of photons that can be collected in the area and also given the limited spectral capabilities of the PSPC, that indeed indicate a single model as acceptable. Whether one can however use the results of any such fit and interpret them as giving a physical interpretation of the observations remains questionable. Observations with better statistics and higher spectral and spatial resolution and an extension to higher energies are needed badly. Only then will we be able to answer the question if the - with ROSAT HRI - unresolved component to the east of the optical nucleus is coincident with the optical nucleus and the hard ASCA source (see above) or - our prefered explanation - the brightest and hottest part of the out-flowing material.
We can compare the emission from the nuclear super-bubble to the X-ray emission from the plume detected in NGC 253 (see Pietsch et al. 1998), the anomalous arms of NGC 4258 (see Pietsch et al. 1994), and the extended emission resolved with the HRI in NGC 2146 (Armus et al. 1995), X-ray features of similar appearance and proposed origin. While in the prototypical starburst galaxy NGC 253 the plume has an extent perpendicular to the major axis of pc and a luminosity of erg s-1 , in the LINER galaxy NGC 4258 the extent is 1.5 kpc and the luminosity is erg s-1 . For the starburst galaxy NGC 2146 (after correcting for an inclination of , Tully (1988)) the HRI extent is 2.2 kpc and the unresolved luminosity in the range of erg s-1 . Due to the low inclination of NGC 2146, however, it is not possible to determine the scale height of the extended emission perpendicular to the galaxy disk and separate point-like and extended components. In comparison the NGC 3079 central component has an extent of 1.7 kpc and a luminosity of erg s-1 . NGC 253 and NGC 2146 show coinciding optical line emission, that is explained by an out-flowing wind driven by the nuclear starburst. Close to the nucleus of NGC 253 embedded in the plume emission a slightly extended source represents the hottest part of the out-flowing material and may reflect the point-like nuclear component in NGC 3079. The extended nuclear features in NGC 3079 and NGC 4258 are more X-ray luminous than in the pure starburst galaxies NGC 253 and NGC 2146 and are not only traced by optical emission lines but in addition by jet-like continuum radio emission indicating the importance of magnetic fields funneling the outflow and the existence of a driving AGN.
4.3. Extended X-ray halo of NGC 3079
The PSPC contour plots (Figs. 3 and 10) and the radial distribution of the PSPC and HRI detected photons (Fig. 6) clearly indicate that the emission from NGC 3079 extends to a radius of 13.5 kpc. The distributions along the major and minor axis (Fig. 7) demonstrate that this emission is not only originating within the NGC 3079 disk but is filling the halo to at least the inclination corrected D25 diameter of the galaxy. The spectral investigations clarified that the halo emission contributes about one third to the total X-ray luminosity (erg s-1 ) of the galaxy in the 0.1 - 2.4 keV ROSAT band; emission from the center and the disk complete the luminosity in about equal parts.
The extent of the halo of NGC 3079 is comparable or slightly bigger than the one of the starburst galaxy NGC 253 (Pietsch et al. 1996, 1998) and bigger by a factor of than that of the active galaxy NGC 4258. On the other hand, the luminosities (and temperatures) for the halos are similar in the two galaxies with known active nuclei,while they are higher than in the starburst galaxy. The halo emission has a X-shaped structure (c.f. Figs. 8 and 10) similar to NGC 253 (however not as clearly resolved due to the greater distance of NGC 3079). X-shaped filaments of the diffuse ionized medium within a radius of 5 kpc, emerging from the inner disk and rising more than 4 kpc above the disk plane of NGC 3079, have been reported from optical [N II ] measurements (Heckman et al. 1990, Veilleux et al. 1995). According to Veilleux et al., the morphology, kinematics, and excitation of the filaments suggests that they form a biconic interface between the undisturbed disk gas, and gas entrained in a wide-angle outflow. In H I observations of NGC 3079 numerous arcs and filaments are present extending away from the plane of the edge-on galaxy which seem to be unrelated to the nuclear activity (Irwin & Seaquist 1990). They may indicate gas flow between disk and halo in addition to violent outflow from the active nucleus (Heckman et al. 1990, Filippenko & Sargent 1992).
If we assume that the halo X-ray emission is due to hot gas with a temperature of K (cooling coefficient according to Raymond et al. (1976) of erg cm3 s-1), a luminosity of erg s-1 (see Table 7) distributed in a sphere of 13.5 kpc radius with filling factor we can calculate parameters of the plasma using the model of thermal cooling and ionization equilibrium of Nulsen et al. (1984). The electron densities, masses, and cooling times of the X-ray emitting gas in the halo are: cm-3, , yr.
These gas parameters differ significantly from the parameters derived by Read et al. (1997) for NGC 3079 which only use a strongly simplified geometrical model attributing the X-ray emission to a nuclear point source and just one diffuse extended emission component. They find a luminosity that is higher by a factor of and nearly twice the temperature compared to the halo parameters as given in Table 7; i.e. their values mix up halo and disk, leading to higher electron densities and gas masses, and slightly longer cooling times. Read et al. analyzed several nearby spiral galaxies using similar simplified models. The resulting parameters for the diffuse gas always overestimate a possible halo component preventing a sensible comparison to the NGC 3079 halo results.
As mentioned above to derive the gas parameters we assumed a spherical distribution with constant temperature. This approximation however, only describes the situation to first order. At least two further details indicated by the ROSAT data should be addressed:
To better understand and characterize these effects spatially resolved spectra are strongly needed which, due to the low counting statistics, can not be tackled with the ROSAT data at present. Temperature profiles of the diffuse emission will help to solve the question of whether the E - W asymmetry can be attributed to inhomogeneities in the galaxy halo or instead it reflects a hot medium connected to the potential of the NGC 3079 group (or both components could be present). This question is also relevant when interpreting H I data of the companion galaxy NGC 3073 which exhibit an elongated tail aligned with the nucleus of NGC 3079; an explanation of the tail due to ram pressure stripping by out-flowing gas from NGC 3079 has been put forward in favor of stripping by movement in an intergroup medium (Irwin et al. 1987). A galactic wind perpendicular to the disk in the general direction of the minor axis can explain the X-shaped halo structure of the X-ray and optical [N II ] data (see above).
4.4. NGC 3079 as LINER/Seyfert 2 galaxy
The unified model for active nuclei (Antonucci 1993) has been proposed to explain the properties of the different types of AGN. In this scheme the difference between Seyfert 1 and 2 galaxies is determined by the viewing angle to the nucleus that is embedded in a torus of molecular material. Type 1 Seyfert galaxies allow direct observation of the nucleus, whereas for type 2 galaxies nuclear X-ray emission can only be seen via reflection above or below the torus or dust gains, or via transmission through this torus. X-ray spectra of the nuclear emission are directly probing the absorption depth under which an AGN is seen, and therefore are an important test for the unified scheme.
As discussed in Sects. 3.4 and 4.2, only very little direct emission from the LINER/Seyfert 2 type nucleus of NGC 3079 - if any at all - is detected in the ROSAT band. Due to the many different equally bright components that contribute to the overall X-ray spectrum of NGC 3079 and due to limited counting statistics, integral spectra of the galaxy as obtained by ASCA can not fully separate the nuclear emission component and only give an indication of a highly absorbed power law component that could be related to the nucleus. X-ray spectroscopy with high spatial resolution as expected from the next generation of X-ray observatories (AXAF, XMM) will resolve the ambiguity. From mm-wave estimates of the extinction to the nucleus of NGC 3079 (Sofue & Irwin 1992) and by comparison with integral broad band X-ray spectra of other Seyfert 2 galaxies obtained with the ASCA or BeppoSAX satellites (see Maiolino et al. 1998 and references therein) one may infer absorption depths well in excess of 1024 cm-2 also for NGC 3079.
Soft X-ray emission in excess of the high energy spectrum has been discovered in many Seyfert 2 galaxies. In Seyfert 1 galaxies such a component may not be observed because it is covered by the bright low absorbed flat power law spectrum. In the nearby Seyfert 2 galaxies NGC 4258 and NGC 1068 the soft component could be resolved and is attributed to emission from a bipolar jet and from the halo of the galaxies (see Pietsch et al. 1994, Wilson et al. 1992). As reported above NGC 3079 show the same components. Therefore, a luminous X-ray halo gas heated by and ejected via jets from the nucleus may not only explain the spatially unresolved soft X-ray emission components reported in other Seyfert 2 galaxies but also be expected in many Seyfert galaxies and AGN.
4.5. Companion galaxies NGC 3073 and MCG 9-17-9
NGC 3079 (total galaxian mass ) is the dominant galaxy in a group containing NGC 3073 ( ) and MCG 9-17-9 ( )(Irwin et al. 1987). While little is known from observations at other wavelength on MCG 9-17-9 - it's redshift and H I contours may even suggest that it is not physically associated with NGC 3079 - NGC 3073 is discussed as a starburst galaxy situated along the trajectory of the outflow of NGC 3079 with the starburst induced by the super-wind (Filippenko & Sargent 1992).
As shown in Sect. 3.5, MCG 9-17-9 is bright in X-rays with a luminosity of erg s-1 while NGC 3073 is barely detected (erg s-1 ). The luminosity of MCG 9-17-9 is rather high for such a dwarf galaxy (c.f. Markert & Donahue 1985) and may indicate the presence of at least one super-luminous source. Such a source could be a low luminosity active nucleus, a X-ray binary radiating at super-Eddington luminosity, a young supernova or a combination of the above. An alternative explanation could be the existence of several X-ray binary sources radiating close to the Eddington limit for a one star. A similarly high luminosity has been reported for the nearby Magellanic-type star forming galaxy NGC 4449 (Vogler & Pietsch 1997). There the emission could be resolved in seven point-like sources and an additional diffuse component of about equal luminosity. The upper limit to the luminosity of NGC 3073 is within the values expected for a galaxy of this mass. The starburst activity of the galaxy does not reflect in it's X-ray luminosity.
© European Southern Observatory (ESO) 1998
Online publication: November 9, 1998