Astron. Astrophys. 364, 53-69 (2000)

9. What is the origin of the X-ray emission in early-type galaxies?

The presence of at least 2 components in the X-ray spectra of early type galaxies is now well established for all galaxies spanning the whole range of L ratio, although the origin of these component might be different, as will be discussed later.

Even though the quality of the present data does not allow us to derive very precise spectral parameters for the galaxies studied here, we can discuss their properties in light of these results together with current discussions in the literature.

When the full spectroscopic range of keV is taken into account, the need of at least two components to the model fit becomes quite stringent. As already noticed often in the literature (cf. Buote 1999 and references therein), the present data stress the importance of including the contribution of the harder spectral component at soft energies, even though it might not be possible to measure its strength or detailed characteristics, to avoid a faulty interpretation of the data.

A thin plasma spectrum is most likely the best model to fit the low temperature component in these intermediate and low L objects. For NGC 1407 and NGC 4125 a simple thermal bremsstrahlung fails to reproduce the data at energies below 1 keV, indicating either the need of an additional component or of line emission. Lines are also required to model the low energy component of X-ray bright objects, interpreted as due to the emission of hot gas, although the detailed characteristics (mostly abundances), of the lines have not been unambiguously determined at the present time. In the lower L galaxies, the presence of lines cannot be used to firmly identify the soft component with emission from hot interstellar gas, since line emission below keV has also been found in the spectra of Galactic low mass binaries (Her X-1 and 4U1626-67, Oosterbroek et al. 1997; Owens et al. 1997).

In fact the presence of hot gas at the low end of the X-ray luminosity distribution has been recently challenged by Irwin & Sarazin (1998) who, based on the similarity between the X-ray colors (in the ROSAT band) of low X-ray luminosity galaxies and the bulge of M31, NGC 1291 and galactic low Mass X-ray Binaries (LMXB), have suggested that the LMXB sources could account for the whole spectrum of low galaxies. BeppoSAX and ASCA data on the bulge of M31 (Trinchieri et al. 1999), together with a new analysis of ROSAT PSPC data (Irwin & Bregman 1999), have confirmed the need of a soft component to model the 0.1-10 keV spectrum. The spatial evidence from high resolution X-ray images indicates that a major fraction of the emission is resolved into individual sources, while a smaller percentage could be accounted for by either a population of lower luminosity sources or by a truly diffuse component (the fraction attributed to each component however is different in the Einstein and ROSAT data, see Trinchieri & Fabbiano 1991 and Primini et al. 1993). This argues against a significant contribution from an ISM in the bulge of M31, and suggests that the soft emission detected could indeed be attributed to the low mass binary population. However, whether LMXBs could account for all of this component, and whether the same applies to low luminosity early type galaxies as well is still a very open question: 1) the relative contributions of the soft and hard components in M31 do not appear to be the same as those measured in early type galaxies, at least based on the evidence given by BeppoSAX and ASCA (Trinchieri et al. 1999; however, see Irwin & Bregman 1999); 2) Kim et al. (1998) discuss the positive detection of gas in NGC 1316, one of the low L early type galaxies, which further reinforces the need of more than just binaries in these objects; 3) Borozdin & Priedhorsky (2000) argue that the spectra of the diffuse emission and of the binary sources in M31 differ, suggesting that the soft component is to be attributed all to the unresolved component and not to the LMXB population.

Fig. 10 shows the comparison between the soft emission observed in the bulge of M31 with BeppoSAX , ASCA and ROSAT and that observed in the lowest L early type galaxies. In all objects where the soft component has been measured there is excess emission over even the higher estimate of the expected contribution from binary sources, suggesting that an additional component should be present or that the binaries contribute a higher proportion in the lowest L galaxies. Better quality data however, such as will be provided by Chandra and XMM-Newton observations, are needed to understand the nature of the soft emission in early type galaxies.

Fig. 10. Correlation between the optical (B-band) and soft X-ray luminosity (0.1-2 keV band) in the lowest L early type galaxies (i.e., those belonging to group 1, as defined in Table 1). The solid lines represent the expected contribution from LMXBs estimated by Irwin & Bregman (1999) from an analysis of the bulge of M31; the lines shown include the extreme contributions allowed by different spectral models that are considered equally good by the authors. The dashed line is normalized to the estimate of the emission of a very soft component in the bulge of M31 from BeppoSAX data (Trinchieri et al. 1999). Data are from this paper for NGC4125, Fabbiano et al. (1994) for NGC 4382 and NGC 4365, Kim et al. (1996) for NGC 1316, Pellegrini (1994) for NGC 5866, and Matsushita et al. (2000) for NGC 4697. The open symbols indicate the estimates of the soft component in NGC 3115 and NGC 3379 (see text). Errors on the flux are given for NGC 4125 and M31 and are estimated from the 90% uncertainties on the normalization of the soft spectral component.

At the opposite end of the spectrum, the origin of the emission that dominates at higher energies remains unclear. The evolved stellar population in early type galaxies should give rise to hard emission proportional to the stellar content, similar to what is observed for late type galaxies. In addition, there could be emission from a nuclear source, although in most cases a further requirement that the source is heavily obscured is necessary when the high resolution data at soft energies do not show evidence of an unresolved nuclear source. If the evolved stellar population is responsible, we expect the spectral characteristics of the hard component to be similar to those of the LMXB in our Galaxy, in M31 and to the integrated emission in late type galaxies, where this component is thought to dominate. We also expect its luminosity to be almost linearly correlated to the optical luminosity, as in the spiral galaxy sample (Fabbiano et al. 1992), indicating that a constant fraction of stars has evolved in the compact binary systems. Moreover, the emission should be clearly extended, and follow the radial distribution of the optical light. Signatures of nuclear activity at other wavelengths combined with no spatial extension and appropriate spectral characteristics would instead point to a nuclear source in the galaxy.

The existence of the hard component was first suggested on the basis of very crude spectral data from the Einstein IPC (Kim et al. 1992) but was measured for the first time with ASCA data in an increasingly large number of galaxies (Matsushita et al. 1994; Matsumoto et al. 1997). However, there is to date no firm determination of its spectral characteristics: in most cases (and the BeppoSAX and ASCA data presented here are consistent with this), either a power law or a thermal model can be used to parameterize its spectrum equally well (see also Allen et al. 2000, Table 3).

The temperature measured for the high energy component (kT in the range keV, see Table 3) is consistent with the spectral characteristics of the LMXBs in our Galaxy (van Paradijs 1998) and of the bulge and several sources of M31 measured with BeppoSAX (Trinchieri et al. 1999). For power law spectra, the indexes reported in Table 3 appear to be consistent with the canonical slope of Seyfert galaxies (see also Matsumoto et al. 1997; Buote & Fabian 1998). This is not consistent with the results for a sample of high X-ray luminosity galaxies studied by Allen et al. (2000), whose hard component appears to have a flatter spectrum. However, the flat component in the very high luminosity objects is in general a small fraction of the emission in the hard band and appears to dominate only at very high energies (above 5 keV in the spectrum of M87, see Fig. 2 in Allen et al. 2000).

The spatial analysis of the BeppoSAX data indicates that the high energy emission of NGC 1553 is extended and that a point source and an extended component account for the emission in NGC 3115 (Fig. 8). This argues against a dominant contribution from the nucleus in these two objects (although in NGC 3115 the two could contribute equally). The BeppoSAX data on a third object, NGC 3923 (Pellegrini 1999), also argue against nuclear emission. For the other two galaxies observed with BeppoSAX , NGC 3379 and NGC 4125, the spatial evidence is not as conclusive: there is an indication of excess emission over the PSF, but mostly due to other, possibly unrelated, components. However, a nuclear contribution is likely to be important in NGC 3379, since Roberts & Warwick (2000) report the detection of a single nuclear source in this galaxy from HRI data.

We therefore can exclude a nuclear origin of the hard emission only for NGC 1553 (and NGC 3923), for which a stellar origin of the emission is most likely. From the spatial analysis of NGC 3115 we expect a mixture of nuclear plus stellar component in this galaxy. A large contribution is expected from the nucleus of NGC 3379.

We have tried to compare the characteristics of the hard emission of these galaxies to those previously reported in the literature, in an attempt to discriminate between binary and nuclear emission. In Fig. 11 we present a plot similar to Fig. 7 of Buote & Fabian (1998), but we have included only galaxies for which the hard component has a temperature higher than 3 keV, so that we are sure we are not including the contribution from the hot gas, or an hotter phase of the ISM (see Buote & Fabian 1998).

Fig. 11. Correlation between the optical (B-band) luminosity and that of the hard component, with temperature 3 keV, for early type galaxies. The galaxies studied here are labeled with their names. Data for the other galaxies are from Buote & Fabian (1998), Kim et al. (1996), Matsumoto et al. (1997), Matsushita et al. (2000). The nameless dot is NGC3923, also studied with BeppoSAX and also found extended in the hard band (Pellegrini 1999). The solid line represents the expected contribution from the binary component (Canizares et al. 1987); this estimate is coincident with that obtained from ASCA data by Matsumoto et al. (1997). The dashed lines enclose the expected scatter/uncertainty for the binary contribution (Canizares et al. 1987).

As shown by the figure, the scatter in the points is relatively large and the present galaxies are in very good agreement with the other early type galaxies from the literature. All are consistent with what might be expected from emission from the binary population (sketched by the lines in the figure, derived by Canizares et al. (1987), based on globular clusters and sources in the bulge of M31), as already noticed by Buote & Fabian (1998). The extended nature of the emission in NGC 1553 and NGC 3923 guarantees that these estimates are indeed reasonable, as shown by their positions in the plot. NGC 1407 appears to be outside the range of the binary sources. However, since it hosts a compact, flat-nuclear source (Disney & Wall 1977; Dressel & Wilson 1985) and it is located at the center of a small group, we cannot exclude additional emission from either of these components. A similar reasoning can be applied to other galaxies in the same part of the diagram.

Stellar kinematics for two of the present galaxies, NGC 3115 and NGC 3379, testify to the presence of a central mass concentration of M M (Kormendy et al. 1996; Emsellem 1999; Gebhardt et al. 1996, 2000; Magorrian et al. 1998). NGC 3379 has been classified as a LINER 2 in the spectroscopic survey to search for nuclear activity conducted by Ho et al. (1997). A contribution from the nuclei could therefore be expected in these two objects. However, their hard components are within the range expected for the binaries contribution (Fig. 11), and their luminosities are well below what has been observed in other galaxies who host a similar or even lower mass nuclear black holes, NGC 4258 (Fiore et al. 2000), with an X-ray luminosity L erg s^-1 and a black hole mass of M, or the Sombrero galaxy, with L erg s^-1 (Fabbiano & Juda 1997).

If we assume a substantial contribution from the nucleus, then the contribution from binaries alone is lower (from the spatial analysis, a factor of about 2 for NGC 3115, most likely more for NGC 3379), and could fall below the estimates shown in Fig. 11 for the binary contribution. However, given the relatively large range spanned by the points, the poor knowledge of the properties of this component in early type galaxies and the lack of points at the same low optical magnitude of these galaxies, we cannot realistically speculate on whether this would make the high energy emission from these two galaxies peculiar relative to the other objects.

We further notice that the (unknown) intrinsic scatter in the plot could be enhanced both by the limited quality of the data available and by the limited information available on the detailed properties of these sources. Different assumptions on the spectral models, the uncertainties even within a single set of spectral model and the presence of several components (nuclei, binaries, group emission), which cannot be discarded both in different objects and within the same object with the present data, could introduce factors of a few in the estimates of the source luminosities (as already remarked in Sect. 7).

We can therefore only conclude that on average we expect that a large fraction of the hard emission in early type galaxies originates from the evolved stellar population, as seen in later types. Once again, the spatial and spectral capabilities of Chandra and XMM-Newton are required to properly measure the characteristics of this component and therefore better understand its origin.

© European Southern Observatory (ESO) 2000

Online publication: December 15, 2000
helpdesk.link@springer.de