Astron. Astrophys. 342, 101-123 (1999)

4. Discussion

Extrapolating from our knowledge of X-ray point-like sources in the Galaxy, its neighboring galaxies in the Local Group and further nearby galaxies, the sources detected in NGC 253 should be from the following classes: (1) X-ray binaries. Here, a compact object (a white dwarf (WD), a neutron star (NS), or a black hole (BH)) accretes material from an accompanying object. Depending on the nature of the donor star, X-ray binaries are classified as high mass X-ray binaries (massive O or B stars as companions) or low mass X-ray binaries (late type stars as companions). The maximum X-ray luminosity that can be achieved in such a system, assuming steady spherical accretion, was calculated by Eddington (1928) as , where M is the mass of the compact object. Only when the compact object is a NS or BH is most of this luminosity emitted in X-rays. As NSs have typical masses of 1 , the commonly assumed maximum luminosity for a NS system is erg s^-1 . If the luminosity of an X-ray source via the Eddington formula indicates a mass - and the source has to be explained as an X-ray binary - only black holes can act as the accreting objects, as theory does not allow such masses for WDs and NSs. (2) Supernovae. In the ROSAT band, X-ray emission from supernovae (SNe) is expected from the interaction between the SN ejecta and the circumstellar matter, heated by the outgoing wave. To date, eight detections of X-ray emitting SNe are reported, with peak X-ray luminosities in the range - erg s^-1, followed by an exponential rate of decline (cf. Schlegel 1995 for a review of X-ray observations of SNe to the year 1995 and Fabian & Terlevich 1996; Lewin et al. 1996; Immler et al. 1998a and Immler et al. 1998b for detections thereafter). Based on the estimated SN rate of per century per () for galaxies of type Sc (van den Bergh 1993) and a total blue luminosity of NGC 253 of (Tully 1988), a SN rate of  1-2 per century is expected within NGC 253, these sources, by now, possibly having evolved into X-ray point sources. It is clear that this simple estimate will only roughly describe the SN rate in the disk of the galaxy and not the one connected to the nuclear starburst activity. Sources close to the nucleus will however be too highly absorbed in any event and - if ever visible at all - unresolvable with the HRI. (3) Supernova remnants. SN remnants in the Milky-Way and our neighboring galaxies attain X-ray luminosities of up to several 10³⁶ erg s^-1. A very X-ray bright remnant ( erg s^-1) is reported for NGC 4449 (e.g. Blair et al. 1983, Vogler & Pietsch 1997). (4) Super-bubbles. In regions of enhanced star formation, correlated winds of massive stars ( yr) and SN explosions ( yr) can heat a part of the interstellar medium to  K. Super-shells surrounding these bubbles might complicate the detection of the hot interior. Additional X-ray emission connected to super-bubbles is expected from X-ray binaries, SNe or SN remnants contained in the star forming regions. The brightest known super-bubbles in the LMC (the NGC 44 super-bubble and shell 5 in 30 Dor) have luminosities of  erg s^-1, and in M 101, five bright H II regions have luminosities from  erg s^-1 to  erg s^-1 (Chu & Kennicutt 1994, Williams & Chu 1995), while most of the known super-bubbles have lower luminosities.

In the Milky-Way, SN remnants and super-bubbles can be spatially resolved with the ROSAT instruments. At the distance of NGC 253, one expects these source classes to be point-like at the resolution of ROSAT, as long as the spatial extent is  pc. This extent can only be exceeded in the case of giant super-bubbles (cf., e.g. the HRI detection of a super-bubble in NGC 3079, Pietsch et al. 1998b).

4.1. Comparison of the ROSAT point source catalog with Einstein

To investigate the long term time variability of the ROSAT detected NGC 253 point sources, the ROSAT results were compared with the Einstein HRI data, collected in July 1979. The Einstein source list contains eight point-like sources (sources E1 - E8, Fabbiano & Trinchieri 1984, cf. our Fig. 2 and the figure caption for the position of the sources). To calculate upper limits (2) to the Einstein luminosity at the positions of (Einstein -undetected) ROSAT sources, Einstein data have been retrieved from the High Energy Astrophysics Science Archive Research Center (HEASARC), operated by the Goddard Space Flight Center (GSFC). Table 9 compares luminosities and upper limits for the Einstein and ROSAT PSPC and HRI detected sources.

Table 9. Comparison of Einstein and ROSAT detected point sources in NGC 253 ^a Luminosities measured with the Einstein HRI, the ROSAT HRI and the ROSAT PSPC. Given in 10³⁷ erg s^-1 (0.1-2.4 keV). Assumed spectral model: 5 keV thermal bremstrahlung, corrected for Galactic foreground absorption. For non-detected sources we give upper limits. ^b Maximum luminosity (cf. Sect. 3.3 and Table 6)

27 sources are detected within the ellipse NGC 253 by ROSAT, 9 of which show time variability within the ROSAT observations (cf. the results in Sect. 3.3). For four of the time variable sources (X12, X14, X28 and X40) the Einstein upper limits lie below the ROSAT measurements, and this further strengthens the idea that these sources are time variable. Two additional time-variable ROSAT sources (X17, X21) were also detected with Einstein , and for these sources, the different Einstein and ROSAT luminosities support the picture of time variability. Two ROSAT sources with no known time variability were detected with Einstein , namely the bright point source close to the nucleus (X33) and X36. ROSAT and Einstein luminosities for these sources agree within the errors, and no time variability can be inferred. Due to the reduced sensitivity of the Einstein observation however (detection limit  erg s^-1, compared to  erg s^-1 for ROSAT), we cannot make any long term time variability arguments for the remaining fainter ROSAT sources. Conversely, four sources within or close to the ellipse of NGC 253 were seen with Einstein , but not with ROSAT: E3, E4, E6 and E7. These non-detections with ROSAT (the ROSAT upper limits being significantly below the Einstein luminosities) argue for the detection of transients in the case of these sources.

4.2. The nature of the NGC 253 point sources

The nature of the sources associated with the NGC 253 disk is discussed in this section. As will be shown in Sect. 4.4, only a negligible number (of order 1) of foreground or background X-ray sources is expected within the ellipse of NGC 253.

Combining the ROSAT and Einstein observations, a total of 31 point sources in NGC 253 are seen. Two Einstein sources, E3 and E4, located close to the D₂₅ ellipse of NGC 253, are included in this number. This seems justified though, as HI observations of NGC 253 (Puche et al. 1991) and deep optical observations (Beck et al. 1982) indicate an extent of NGC 253 far beyond the D₂₅ ellipse. Also, their transient nature (see Sect. 4.1) strengthens the idea that E3 and E4 are members of NGC 253, as had already been proposed by Fabbiano & Trinchieri (1984).

Time variability is detected in 13 of the 31 NGC 253 sources, and nearly one half of these (X12, X14, E3, E4, E6 and E7) show transient behavior. X12 and X14 reach their highest luminosities during the ROSAT observation blocks 4 and 6, respectively. Compared to the lowest upper limits in the case of non-detections (block 7 for X12 and block 4 for X14), the peak luminosity is higher by 4.0 and 5.6 ( represents the measurement error of the peak luminosity) for X12 and X14, respectively. While X14 is not detected with the PSPC, X12 is detected during block 3. The luminosity is smaller than the one measured with the HRI however, due to the smaller error of the PSPC measurement, the PSPC peak luminosity is 19.8 above the upper limit for block 7. In the case of the Einstein detected transients E3, E4, E6, and E7, the significances of the luminosity above the ROSAT upper limit are 4.4, 4.3, 3.4, and 3.2, respectively.

The interpretation of the bright time-variable sources is relatively straightforward. Taking into account the fact that - with the exception of SN 1940E - no SNe have been reported in NGC 253, the time-variable sources have to be classified as X-ray binaries. With the exception of X33, the derived maximum luminosity of the variable sources ( erg s^-1) and the variability could be well explained if one assumes X-ray binary systems containing accreting objects of mass ( neutron stars) radiating close to the Eddington limit. However, if additional absorption, intrinsic to the sources or from the H I disk of NGC 253 is included, higher intrinsic luminosities of the sources are implied. An additional of cm^-2 or cm^-2 (cf. the H I map presented in Puche et al. 1991), for example, would imply source luminosities higher by factors of 1.6 and 4, respectively. Under these assumptions, the maximum X-ray luminosities of all the mentioned binary candidates would be close to, or even exceed the Eddington limit for a neutron star binary. If one assumes that X33 is an X-ray binary radiating at the Eddington limit, a mass for the compact object of can be calculated, that clearly puts the object in the mass range expected for a black hole.

It is more difficult to determine the nature of the remaining, less bright X-ray sources in NGC 253. Firstly, the non-detection of time variability with ROSAT might be due to the low photon statistics or the time windows of the observation blocks, and, in general, one cannot rule out the idea that the luminosity of these sources is variable. Secondly, none of the X-ray positions coincide with point sources visible in radio maps that could be due to SNe, SNRs or H II regions. This may not be too astonishing, bearing in mind that the brightest supernova remnant in the LMC (N158A) has  erg s^-1 (Chu & Kennicutt 1994), well below the detection limit of the ROSAT observations for NGC 253. Therefore, one might again argue in favor of an X-ray binary identification. Another explanation one could put forward is that unresolved emission from SNe and SN remnant and X-ray binaries, embedded in the hot interstellar medium of star forming regions, would suppress time variability. Such a scenario might also explain the bright source X36 (no time variability established) with a luminosity ( erg s^-1) comparable to that of X-ray bright super-bubbles reported in M 101 (Williams & Chu 1995).

In contrast to face-on galaxies like M 101, the edge-on orientation of NGC 253 complicates the detection of H II regions in H observations, and only for X42 ( erg s^-1) could a positional coincidence be established (reported by Waller et al. 1988).

4.3. Non-detection of SN 1940E?

The type I supernova SN 1940E (cf., e.g. Barbon et al. 1989) is the only historical SN reported in NGC 253. The SN is located 71" west and 17" south of the nucleus, and its position is marked with a cross () in Fig. 2. Our X-ray catalog contains no source close to the position of the SN. We searched for faint emission at the position of SN 1940E, by extracting HRI counts from a ring of radius 10". The background was taken from an annulus of 10" to 15" radius. This procedure resulted in a residual count rate of (2.31.3)10^-4 cts s^-1 (converting to  erg s^-1). It is not clear, however, whether this 1.8 excess is really emission from the SN or is just produced by the patchiness of the diffuse X-ray emission in the inner spiral arms of NGC 253.

To date, only one possible X-ray detection of a SN of type I soon after the outburst has been reported (SN 1994I in M 51,  erg s^-1, Immler et al. 1998b). Comparing with this paper, and taking the luminosity upper limit above and with a cooling function of erg cm³ s^{- 1} (Raymond et al. 1976), one can estimate a mean density of (for a shell expansion velocity in units of 10 000 km s^-1) and a total mass of X-ray luminous gas of inside a sphere of radius cm. While this density limit is typical for the interstellar medium in the disk of galaxies, it is four orders of magnitude lower than that expected for the gas deposited by type I SNe due to non-conservative mass transfer to a companion or due to stellar wind prior to the outburst (cf. Immler et al. 1998b). There are several ways to explain this discrepancy: SN 1940E may be embedded deep in the NGC 253 disk, and the count to luminosity conversion may underestimate this effect, the assumptions for the cooling may be wrong, leading to a too high luminosity after 55 years, or the emission from SN 1994I may not be typical for type I SN.

4.4. Possible contributions from foreground or background sources

We have attributed all X-ray sources found within the optical extent of NGC 253 to the disk of NGC 253 (with the exception of X58, a background object, cf. Appendix A). We demonstrate that this assumption is justified, in the following, by proving that only a negligible number of foreground and background sources is expected in this area.

Firstly, one can estimate the number of X-ray sources due to foreground objects by extrapolating the local density of X-ray detected stars in the field. To do so, only HRI sources were taken into account. In this way we avoided PSPC sources that might partly represent diffuse emission in the halo of NGC 253. From the 34 HRI sources outside the ellipse of NGC 253, only the sources X31 and X61 are identified with foreground objects (stars, see Appendix A). Keeping this in mind and comparing the area covered by the disk of NGC 253 (120 arcmin²) to that of the remaining HRI field of view ( arcmin²), one expects 0.25 foreground source within the disk of NGC 253, making the detection of such a source rather unlikely.

Secondly, one can estimate the number of background sources shining through the disk of NGC 253 in a similar way. We can assume conservatively that all objects outside the NGC 253 disk with the exception of X31 and X61 are background objects. In addition, we have to keep in mind that the source flux for objects behind the NGC 253 disk will be reduced due to the additional absorption of X-rays that these objects will suffer from the interstellar medium within the disk of NGC 253. While the detection limit for field sources in the HRI field was  erg s^{- 1} cm^-2, we only expect to detect sources from behind the NGC 253 disk if they have intrinsic fluxes of at least  erg s^-1 cm^{- 2} (correcting for a typical column density cm^-2 (Puche et al. 1991)). Only 18 background sources were detected above this limit outside the NGC 253 disk. From this and the ratio of the NGC 253 disk/outside areas, one can predict 2.3 background sources within an area covered by the NGC 253 disk. With the help of optical spectroscopy (see appendix), we have already identified one background QSO, X58, on the border of the ellipse of NGC 253.

Another approach to determine the contamination of NGC 253 disk sources with background objects uses the deep field luminosity functions derived in the Lockman hole (Hasinger et al. 1991, Hasinger et al. 1993). If we again correct for an average of cm^-2 within the the disk of NGC 253, 1.6 background sources are predicted that should be detectable shining through the NGC 253 disk. This number is - within the statistics - consistent with the one derived from the field objects.

4.5. Comparison to results previously published on ROSAT and ASCA observations

A sample of ROSAT PSPC observed spiral galaxies - including the 22.9 ks PSPC observation of NGC 253 - has been homogeneously analyzed by Read et al. (1997) to search for point source and diffuse emission components. In the NGC 253 field, they detect 15 point sources (R1 - R15), seven of which are located within the disk of NGC 253 (i.e. R4 corresponding to the source cataloged above as X12, R6/X15, R7/X17+nearby source X18, R8/X21, R11/X34+nearby source X33, R12/X36 and R13/X40+nearby source X41). Due to their small separations, the sources X17/18, X33/34 and X40/42 are not resolved as individual sources within the PSPC data. The count rates given by Read et al. (1997) for the sources R4 - R13 are slightly higher () than our rates. This can be understood if one keeps in mind that we used a multi source fit technique to calculate the count rates, and excluded contributions from nearby sources, which otherwise might increase the count rates. In addition, due to a more sensitive source search, our catalog contains more sources.

Ptak et al. (1997) report on ASCA observations of NGC 253. Because of the large PSF of ASCA (FWHM ), point sources cannot be spatially separated from diffuse emission components. However, because of the good spectral resolution of the detectors, they were able to fit multi component spectral models to the integral emission of NGC 253. When comparing with ASCA, one has to keep in mind that due to their geometry the ASCA detectors do not cover the entire NGC 253 disk, and therefore ASCA count rates have been only been extracted from a circle of 6´ radius around the center of NGC 253. Also, the ASCA energy coverage (0.5-10 keV) differs from that of ROSAT. In the overlapping 0.5-2.0 keV band, Ptak et al. report a flux of  erg s^-1 for NGC 253. Extracting from the same area one obtains a count rate of 0.33 cts s^-1 in the corresponding ROSAT hard band. To be independent of errors introduced by different spectral models, the same spectral parameters as used for the ASCA data are used to convert the ROSAT count rate to a luminosity. Ptak et al. (1997) fit the entire emission of NGC 253 in the 0.5-10 keV ASCA band as a combination of a thin thermal plasma and a power law component. They obtain cm^-2 and keV for the thin thermal plasma component, and cm^-2 and a power law index of 2.0 for the higher absorbed power law component with unabsorbed luminosities (0.5-2.0 keV) of  erg s^-1 for the thin thermal plasma component and  erg s^-1 for the power law component. From this, the calculated luminosities corrected for Galactic absorption are  erg s^-1 and  erg s^-1 for the thermal plasma and power law component, respectively. For the same spectral model, the ROSAT H band count rate translates to  erg s^-1, a luminosity in very good agreement with the one measured by ASCA. If one still wants to explain the slightly higher ROSAT luminosity (by  erg s^-1), one can argue in terms of the existence of time-variable sources which might have been picked up with ROSAT during the six observation blocks distributed over several years but not with ASCA (only one observation block). In this way, the transient X12 contributes an average luminosity of to the luminosity determined in the ROSAT PSPC hard band.

4.6. Comparison to other wavelengths

We compared our ROSAT point source catalog to images taken at other wavelengths to identify possible supernova remnants, HI holes and HII regions within the NGC 253 disk.

Radio maps from 0.3 GHz-4.7 GHz (Carilli et al. 1992, Beck et al. 1994) show a bright nuclear source surrounded by diffuse radio emission. The diffuse emission covers the entire NGC 253 disk and protrudes from the disk into the halo of the galaxy. On top of the diffuse radio emission covering the bulge, disk and halo of NGC 253, several enhancements in the radio emission are visible. Within the disk of NGC 253, no X-ray sources are found at the position of this enhanced radio emission, with the exception of the nuclear area. VLA observations of NGC 253 (e.g. Ulvestad & Antonucci 1997) resolve the central nuclear region of NGC 253 at a spatial scale of  pc. These radio detected SN remnants and H II regions can be attributed to the starburst nucleus of NGC 253. However, the resolution of the ROSAT data (pc) is insufficient to identify (a part) of the extended central X-ray emission with individual radio point sources. On the other hand, the enhanced radio emission coincides nicely with the bright near-infrared emission of the starburst nucleus (Sams et al. 1994); also the young, luminous, compact stellar clusters detected with the WFPC2 camera on the Hubble Space Telescope (Watson et al. 1996) are located in the same area. The maximum of the central extended X-ray source (X34, cf. Sect. 3.2.1) is found with an offset of to the southeast from the cluster of these bright nuclear radio sources. The fact that this offset is slightly exceeding the systematic position errors, may indicate that the nuclear extended X-ray emission is caused by the hottest part of the gas outflowing from the nuclear region and not by a collection of individual point sources (see further discussions in PEA).

The X-ray point source catalog can be compared with fainter compact radio sources in the disk by making use of the reanalysed 6 and 20 cm VLA data (Ulvestad & Antonucci, in preparation). From a diameter of they report 27 compact sources in the NGC 253 disk (outside the nuclear starburst) and 5 sources close-by. While several sources within this new catalog coincide with source positions that we have already derived from the radio images of Carilli et al., no coincidences with the X-ray catalog are seen. This is slightly surprising as one might have expected that some of these SNRs or H II regions detected in radio would also be bright enough to be detected in X-rays. An explanation for this behavior may be that, due to the edge on view on NGC 253, the X-ray emission of these radio sources (mostly soft for this class of emitters) is heavily absorbed and therefore not detected.

A comparison of the X-ray source catalog to H images (Waller et al. 1988) suggests a coincidence of X42 and a bright H II region. Therefore X42 may represent emission of hot gas connected with the H II region. This would be consistent with the non-detection of X-ray variability and also the X-ray luminosity of the source. The hardness ratio HR2 of 0.2, on the other hand, suggests a hard spectrum, that would not be expected for this class of source.

4.7. Comparison to X-ray point sources in other spiral galaxies

After discussing of the nature of the individual point sources in NGC 253, we may compare the point source population of NGC 253 with results from other spiral galaxies.

The local group spiral galaxies M 31 (Supper et al. 1997) and M 33 (Schulman & Bregman 1995) have been investigated to luminosities below that reached for NGC 253 (cf. Table 10). We compared the integrated point source content (cf. Table 10) as well as the luminosity distributions of the individual point sources of these galaxies (cf. Fig. 9) to NGC 253. The whole optical extent of M 31 has been observed with a ROSAT PSPC raster scan. In the central area of M 31 ( kpc corresponding to 5´) the PSPC source confusion is very high. While this central region was therefore excluded from the luminosity distribution, the integrated point source content of the region was estimated, assuming that 90 of the bulge luminosity is made up of point sources. This assumption seems to be justified by Einstein HRI and ROSAT HRI data (Trinchieri et al. 1988; Primini et al. 1993), which do not show diffuse or plume like emission as in the case of the NGC 253 nuclear region. The ROSAT HRI observation of M 33 was centered on the nucleus of the galaxy (Schulman & Bregman 1995). Due to the large optical extent of M 33 ( ellipse of , Tully 1988), the HRI observation (field of view ) did not cover the entire outer disk of the galaxy. At the center of M 33 is a bright X-ray source ( erg s^-1), its true nature not yet being clear. The source might be, e.g, a mildly active nucleus. To reduce the contributions of active or starburst nuclei to our point source comparison the central sources of M 33 and NGC 253 (the extended source X34) were excluded from the total point source luminosity and the luminosity distribution diagram. However, we are aware that the nuclear area of M 33 could contain a sample of bright X-ray point sources instead of a mildly active nucleus, and that the total point source luminosity as well as the luminosity distribution of M 33 as assumed by us would have to be corrected.

Table 10. Comparison of the NGC 253 point source content to other spiral galaxies ^a Tully (1988) ^b Galactic foreground absorption (Dickey & Lockman 1990) ^c H = ROSAT HRI, P = ROSAT PSPC ^d 0.1-2.4 keV band luminosity of point sources (excluding the bulge of M 31 and the sources at the position of the nucleus for all other galaxies), corrected for Galactic foreground ^e Blue luminosities according to Tully (1988), corrected for the given distances and adjusted for reddening ^f No active galactic nucleus or plume like emission in the center region has been established for M 31 from the Einstein HRI observations. Assuming that 90% of the bulge luminosity ( erg s^-1) are caused by point sources, one obtains  erg s^-1 and

Fig. 9. Luminosity distributions of detected point sources (excluding nuclear sources) for the spiral galaxies NGC 253 (ROSAT HRI), M 31 (ROSAT PSPC, Supper et al. 1997) and M 33 (ROSAT HRI, Bregman & Schulman 1995). The error bars (1) for the individual points have been calculated from either a Gaussian error distribution or Poisson statistics when appropriate (cf. Gehrels 1986)

In the case of NGC 253, M 31 and M 33 the ratio between X-ray and optical luminosity, which is independent of distance, has been compared (cf. Table 10), and it differs slightly, being lowest for the nearly edge-on galaxy NGC 253 () and highest for the galaxy with the lowest inclination, M33 (). One can to first order correct for the inclination effect by correcting for an average additional absorption of the NGC 253 sources of cm^-2 (cf. the HI map presented in Puche et al. (1991)), resulting in , lying between the values found for M 31 and M 33. The ratios only differ by a factor of less than 2 in spite of the differences in morphological type and star forming activity of the galaxies. These results fit nicely to the close correlations for X-ray and optical luminosities of spiral galaxies reported by Fabbiano et al. (1992) from an analysis of the Einstein sample of galaxies.

The number of point sources detected in different luminosity ranges for NGC 253, M 31 and M 33 are presented in Fig. 9. The nuclei of NGC 253 and M 31, as well as the inner bulge region of M 33 have been excluded for the reasons mentioned above. As the X-ray sources in M 31 are complete for  erg s^-1 (cf. Supper et al. 1997), the M 31 curve above this limit defines the luminosity function of point sources. Below this threshold it is unclear whether the slower increase in the number of sources per energy bin is due to a change in the point source population or due to incompleteness. Within the errors, the luminosity distributions of the NGC 253 and M 33 point sources match within the overlapping region, to that of M 31. This suggests that source populations for the disks of these galaxies (excluding the nuclear regions) are similar. This result is somewhat surprising, as one might have expected an enhanced number of point sources connected with the star forming activity of NGC 253 that has not only been reported from the nuclear starburst region but also from the boiling galactic disk (Sofue et al. 1994). The present measurements do however, not totally rule out this possibility, as there could be an additional contribution of lower luminosity point sources that are not detected due to the ROSAT sensitivity cut-off.

In a couple of more distant spiral galaxies very bright ( erg s^-1), non-nuclear point-like sources were detected by ROSAT, namely in NGC 891 (Bregman & Pildis 1995), NGC 4559 (Vogler et al. 1997), NGC 4565 (Vogler et al. 1996), NGC 6946 (Schlegel et al. 1994) and M 100 (Immler et al. 1998a). In the case of NGC 891 and NGC 6946, the X-ray emission is due to SNe. In the case of the other galaxies, the nature of the bright sources is still unclear. The sources are point-like at the resolution of the ROSAT HRI (NGC 4559, M 100) and PSPC (NGC 4565, no HRI observation available). The non-detection of recent SN outbursts in these galaxies favors the explanation of X-ray binary systems or SN remnants expanding into high density ( cm^-3) media. In the case of X-ray binaries, these objects would most likely have to be black hole binary candidates, the masses of the central objects exceeding . However, the sources could also be due to a superposition of X-ray binaries and SN remnants, as expected in super-bubbles. Assuming these bright sources are really point-like, one would expect, contrary to the observations, that such sources should also exist in NGC 253, M 31, and M 33. Their absence can either be explained in terms of time variability, where no such source was active during the ROSAT observations, or by a differing source population.

4.8. Point sources outside the disk of NGC 253 possibly correlated to the galaxy

While point sources located inside the ellipse of NGC 253 can fully be attributed to the galaxy (with the exception of one or two background sources, cf. Sect. 4.4), it is more difficult to decide whether point sources located (projected) in the halo of NGC 253 are associated with the galaxy. Such a correlation could be expected for globular clusters, which might contain low mass X-ray binaries, or for sources due to emission from hot gas in the halo.

The X-ray point source list is compared to globular cluster candidate lists based on optical observations (Liller & Alcaino 1983, 63 candidates; Blecha 1986, 32 candidates). No positional coincidence can be established, and this seems to contradict expectations deduced from ROSAT observations of M 31. Of the ROSAT sources in M 31, 43 have luminosities above the detection limit of NGC 253, and more than half (26) of these bright sources have been identified as globular clusters. Assuming a similar ratio for NGC 253, 16 sources would be expected to be globular cluster sources. However, a direct comparison between the two galaxies is rather difficult. of the globular clusters detected in X-rays in M 31 are located within an `inner disk' described by an ellipse half the size of M 31. A list of optical candidates for M 31 globular clusters in this region could be obtained because of the lower inclination ( versus for NGC 253) and the smaller distance (0.69 Mpc versus 2.58 Mpc). Such a list is not available for the inner disk of NGC 253. In addition, X-rays from only one globular cluster candidate outside the area covered by the M 31 disk has been detected at a distance of  kpc from the galactic plane. For NGC 253, however, most of the globular cluster candidates contained in the lists of Liller & Alcaino (1983) and Blecha (1986) are located at large distances from the plane of the galaxy ( kpc for nearly all candidates in the list of Liller & Alcaino 1983).

An example of a source in the NGC 253 halo, caused by diffuse emission, and having a point-like appearance at the resolution of the PSPC (and possibly also at the resolution of the HRI) could be, e.g., a region of (older) dense interstellar medium in the halo that is shock-heated by a superwind (cf. e.g. Suchkov et al. 1994). Seven sources in the halo of NGC 253 are surrounded by diffuse X-ray emission attributed to hot gas, namely the sources X10, X13, X22, X24, X27, X30 and X45. Faint optical objects close to the HRI and PSPC detected sources X13 and X22 (separation ) suggest background AGNs as sources of the X-rays. The faint object at the position of X22 could be spectroscopically identified as QSO (see Appendix A). In the case of X27, only detected with the PSPC, a faint object is found at a distance 12" (less than the position error) on the ROE finding charts. This might suggest an identification with a background AGN. The non-detection of this source with the HRI might indicate time variability. The optical object, however, has not yet been spectroscopically identified. Taking into account the larger separation between the optical and X-ray position of X27, compared to X13 and X22, and that, in addition, only X27 was not detected with the HRI, one alternatively might argue that X27 is a diffuse emission feature.

No optical candidates were found in the cases of X10, X24 (both only detected with the PSPC), X30 (only detected with the HRI) and X45 (detected with the HRI and PSPC). In the case of X30, time variability was established with the help of the HRI observation blocks, indicating that X33 is not a diffuse X-ray emission feature. In the case of X10 and X24, time variability is only suggested via a comparison of the PSPC measured fluxes with the HRI upper limits. While a PSPC true color picture (calculated from the images in the soft, hard1 and hard2 band, Vogler 1997) indicates harder spectral behavior at the position of X10 than for the surrounding diffuse emission, such a signature is not visible for X24. This might suggest that the X-ray emission of X10 is due to a background object, whereas that of X24 might reflect structure in the diffuse halo emission. No time variability could be found for X45, and therefore this source is most likely of a similar origin to X24.

© European Southern Observatory (ESO) 1999

Online publication: December 22, 1998
helpdesk.link@springer.de