The ATLAS project - XIX. The hot-gas content of early-type galaxies: fast versus slow rotators
For early-type galaxies, the ability to sustain a corona of hot, X-ray emitting gas could have played a key role in quenching their star-formation history. A halo of hot gas may act as an effective shield against the acquisition of cold gas and can quickly absorb stellar-mass loss material. Yet, since the discovery by the Einstein observatory of such X-ray halos around early-type galaxies, the precise amount of hot gas around these galaxies still remains a matter of debate. By combining homogeneously-derived photometric and spectroscopic measurements for the early-type galaxies observed as part of the ATLAS integral-field survey with measurements of their X-ray luminosity based on X-ray data of both low and high spatial resolution (for 47 and 19 objects, respectively) we conclude that the hot-gas content of early-type galaxies can depend on their dynamical structure. Specifically, whereas slow rotators generally have X-ray halos with luminosity and temperature values that are well in line with what is expected if the hot-gas emission is sustained by the thermalisaton of the kinetic energy carried by the stellar-mass loss material, fast rotators tend to display values that fall consistently below the prediction of this model, with similar values that do not scale with the stellar kinetic energy (traced by the stellar velocity dispersion) as observed in the case of slow rotators. Such a discrepancy between the hot-gas content of slow and fast rotators would appear to reduce, or even disappear, for large values of the dynamical mass (above ), with younger fast rotators displaying also somewhat larger values possibly owing to the additional energy input from recent supernovae explosions. Considering that fast rotators are likely to be intrinsically flatter than slow rotators, and that the few -deficient slow rotators also happen to be relatively flat, the observed deficiency in these objects would support the hypothesis whereby flatter galaxies have a harder time in retaining their hot gas, although we suggest that the degree of rotational support could further hamper the efficiency with which the kinetic energy of the stellar-mass loss material is thermalised in the hot gas. We discuss the implications that a different hot-gas content could have on the fate of both acquired and internally-produced gaseous material, considering in particular how the deficiency of fast rotators would make them more capable to recycle the stellar-mass loss material into new stars than slow rotators. This would be consistent with the finding that molecular gas and young stellar populations are detected only in fast rotators across the entire ATLAS sample, and that fast rotators tend to have a larger specific dust mass content than slow rotators.
keywords:galaxies : formation – galaxies : evolution – galaxies : elliptical and lenticular – galaxies : ISM – X-ray : galaxies – Xray : binaries
Early-type galaxies used to be regarded as purely stellar systems over the largest part of the last century, even though the presence of little or no interstellar medium in these old stellar systems was soon recognised as a problem by Faber & Gallagher (1976). Indeed, such a lack of gas contrasts with the notion that old stars lose a considerable fraction of their initial mass during their evolution. In this context, the finding by the Einstein observatory that early-type galaxies are surrounded by massive haloes of hot (K) X-ray emitting gas (e.g. Forman et al., 1979) both changed our view of early-type galaxies and provided a possible solution to the problem posed by Faber & Gallagher, if a mechanism to heat the stellar ejecta to X-ray temperatures could be found. Around the same time, the advent of new and homogeneous sets of optical data for large numbers of early-type galaxies prompted the finding of a correlation between the X-ray luminosity of such haloes and the optical luminosity of the galaxies they contain (e.g. Forman, Jones, & Tucker, 1985; Trinchieri & Fabbiano, 1985), which when interpreted led to a relatively simple model linking the hot gas reservoirs to the stellar-mass loss material of early-type galaxies (e.g. Canizares, Fabbiano, & Trinchieri, 1987). According to this scenario, still part of the current consensus, the gas ejected from evolved stars and planetary nebulae collides and shocks with other ejecta or a pre-existing ambient gas until it is heated to the kinetic temperature that corresponds to the stellar velocity dispersion of a galaxy. Supernovae explosions (in particular of type Ia, e.g. Mathews & Loewenstein, 1986; Ciotti et al., 1991) contribute to further heat this gas, but may lead also to its escape. This would have generally been the case when galaxies were young and SNe explosions more frequent, whereas today SNe-driven global outflows should be more restricted to the less massive systems. On the other hand, bigger galaxies can hold better to their hot gas, and in amounts that scale with their optical luminosity given the stellar origin of such a material.
Yet, despite its tremendous success in explaining the broad trend of the relation, this idea alone does not account for the wide range of X-ray luminosities, of about two orders of magnitude, that is observed in galaxies of similar optical luminosity. A number of physical phenomena have been examined to account for the scatter of the relation (see, e.g., the reviews of Mathews & Brighenti, 2003 and Pellegrini, 2012). Earlier explanations invoked different degrees of ram-pressure stripping, since galaxies in dense galactic environment were found to be X-ray faint (White & Sarazin, 1991), or different stages of the hot-gas evolution caused by the steady decrease over time of the energy provided by SNe or the thermalisation of the stellar motions, which by eventually falling short of the power required to extract the gas from a galaxy drives galaxies from an X-ray faint wind and outflow phase to an X-ray bright inflow phase (and at different speed for galaxies with different structural parameters such as central density, effective radius or kind of radial profile; Ciotti et al., 1991). Subsequently, the finding by Eskridge, Fabbiano, & Kim (1995a, b) that S0s and flat Es show lower X-ray luminosities than rounder elliptical galaxies of the same optical luminosity prompted further theoretical studies concerning the role of intrinsic flattening (which reduces the binding energy of the hot gas and makes it harder to retain it; Ciotti & Pellegrini, 1996) or rotation (D’Ercole & Ciotti, 1998; Brighenti & Mathews, 1996). More recent studies considered also the impact of secondary gas accretion events (Pipino et al., 2005) and showed the importance of contamination from the radiation of the intracluster medium (ICM, which originates most likely from cosmological flows; Mathews & Brighenti, 2003) in the measurements obtained from relatively shallow surveys (Matsushita, 2001).
The launch of the Chandra and XMM space telescopes, with their higher spatial resolution, spectral range and sensitivity, has now made it possible to isolate the X-ray emission of galactic haloes from the signal of unresolved low-mass X-ray binaries (LMXBs), active nuclei (AGNs) or the ICM that otherwise enters the coarser Einstein or ROSAT measurements of . These measurements are very expensive in terms of telescope time, however, and as a result the number of objects with pure values for hot-gas haloes is still quite limited. This is particular true considering that good quality and homogeneous optical data, as well as robust distance estimates, are also highly desirable in order to understand what drives the scatter in the X-ray properties of galaxies. On the other hand, since also photometric and spectroscopic optical measurements for nearby galaxies are witnessing a quantum leap both in quality and sky coverage, it may be possible to make progress in this domain while still using measurements based on ROSAT or Einstein data, which cover a large number of objects. This was for instance the case of the study of Ellis & O’Sullivan (2006), who used the near-infrared measurements from the Two-Micron All Sky Survey (2MASS; Skrutskie et al., 2006) to revisit the stellar to X-ray luminosity relation while still using nearly all of the 401 early-type galaxies in the catalog of O’Sullivan, Forbes, & Ponman (2001), still as of today the largest compilation of measurements obtained with ROSAT and Einstein.
By drawing on new optical data from the complete ATLAS survey (Cappellari et al., 2011a, hereafter Paper I), in this paper we will use both such approaches to reconsider the origin of the scatter in the X-ray luminosity of early-type galaxies, starting from ROSAT or Einstein data for the total X-ray luminosity of 47 galaxies from the ATLAS survey to then analyse a second ATLAS sub-sample of 19 objects with consistent Chandra measurements of the X-ray luminosity of their hot gas. In particular, we will revisit the possible role of galactic rotation and flattening in driving the hot-gas content of galaxies, building on the first observational tests of Pellegrini, Held & Ciotti (1997) and in light of the emerging distinction among early-type galaxies between a dominant population of fast-rotating objects (with regular velocity fields; Krajnović et al., 2011, hereafter Paper II) and a less common class of slowly-rotating galaxies (often hosting kinematically distinct cores; Emsellem et al., 2011, hereafter Paper III). Fast and slow rotators are indeed not only likely to have formed in different ways (Cappellari et al., 2007; Emsellem et al., 2007) but also to have distinct distributions for their intrinsic shapes, with fast rotators being nearly as flat as spiral galaxies and slow rotators being much rounder and possibly triaxial (Weijmans et al. in preparation, but see already Cappellari et al., 2011b, hereafter Paper VII).
This paper is organised as follows. In § 2 we define our two samples and the optical parameters from the ATLAS survey that will be needed for our investigation. We derive our results in § 3, considering first in § 3.1 an ATLAS sub-sample with X-ray measurements based on ROSAT and Einstein data. We then expand our analysis in § 3.2 to ATLAS galaxies with Chandra data, and further check in § 3.3 the results from both § 3.1 and § 3.2 using an additional but less homogeneous set of Chandra X-ray measurements from the literature. We discuss our findings in § 4, and wrap up our conclusions in § 5.
2 X-ray Data and Samples Properties
The two sets of galaxies that will be used in this work simply consist of: a) all the objects in the ATLAS sample in Paper I for which O’Sullivan, Forbes, & Ponman (2001) provides a ROSAT or Einstein measurement for their (considering only detections), and b) of all the ATLAS sample galaxies for which Boroson, Kim, & Fabbiano (2011, hereafter also Boroson et al.) could use Chandra data to separate the contribution from the otherwise unresolved population of LMXBs and from the hot, X-ray emitting gas. These two ATLAS sub-samples include 47 and 19 galaxies, respectively, and since the main difference between them is the spatial resolution of the X-ray data from the literature that were used to compile them, hereafter we will refer to these two sets of objects as the low and high X-ray resolution samples. Furthermore, we note that whereas O’Sullivan et al. include all sort of early-type galaxies, Boroson et al. restricted themselves to well studied X-ray objects and explicitly excluded the dominant galaxies of groups and clusters that are associated to extended hot gas haloes confined by the group or cluster gravitational potential.
For the galaxies in these two ATLAS sub-samples Paper I provides their currently best distance estimates, consistently derived values for their K-band luminosity starting from 2MASS apparent magnitude, and effective radius . Homogeneously derived values for the apparent ellipticity were obtained in Paper III through the analysis of the images at our disposal (mostly from the Sloan Digital Sky Survey), whereas Cappellari et al. (2012b, hereafter Paper XV) provide us with accurate dynamical mass measurement, , based on the detailed Jeans anisotropic dynamical modelling (Cappellari, 2008) of the stellar kinematics obtained from our SAURON data.111 More specifically, the values used here are those based on the self-consistent model (B) of Paper XV, where they are indicated as . Such values provide an approximation for twice the mass contained within a sphere enclosing half of the galaxy light. Given that the stars dominate the mass budget inside that sphere, also provides a good estimate for the total stellar mass of a galaxy, while also including possible variations of the initial stellar mass function (see, e.g., Cappellari et al., 2012a).
The SAURON data also allow us to probe the depth of the gravitational potential well using values for the stellar velocity dispersion extracted within the same physical aperture and to assess the degree of rotational support using the parameter introduced by Emsellem et al. (2007). Since we are interested in the global shape and degree of rotational support of our sample galaxies, we adopt here the values for and that in Paper III were computed within the galaxy effective radius . Similarly, as we are interested in tracing most of the potential well, we will use the stellar velocity dispersion values (also provided by Paper XV) that are extracted within an elliptical aperture of equivalent circular radius . The use of in conjunction with will help us in separating face-on but intrinsically flat and fast-rotating objects from much rounder and slowly-rotating galaxies, as well as in identifying galaxies that may be intrinsically quite flat also in this latter class of objects (most likely as they are triaxial in shape).
All the quantites described in this section are listed in Tabs. 1 and 2 for the galaxies in the low and high X-ray resolution subsamples, respectively. These tables also include rescaled X-ray luminosity values from the literature assuming the ATLAS distance estimates, which, unless otherwised stated, will be used in our analysis. Note: These tables will appear on the present arXiv version of the paper as soon as Paper XV will also be posted on it.
3.1 Low X-ray Resolution Sample
We start this section by showing in Fig. 1 how the galaxies in the low X-ray resolution sample fare in the classic diagram when using the values for the B-band and X-ray luminosity taken from the catalogue of O’Sullivan, Forbes, & Ponman (2001), without any distance rescaling. The dashed and solid lines in Fig. 1 trace the expected contribution to the observed from the unresolved emission of low-mass X-ray binaries, as first estimated by O’Sullivan et al. and subsequently by Kim & Fabbiano (2004). The dotted lines also show the uncertainties associated with the Kim & Fabbiano calibration, which was based on the luminosity function of LMXBs resolved with Chandra images in a sample of 14 E and S0s. The other early-type galaxies with detections in the O’Sullivan et al. catalogue are also plotted in Fig. 1, to illustrate better how traditionally a would mark the onset of a trend due to the presence of a diffuse component. X-ray bright objects at low values in this diagram are most likely outliers that own their X-ray flux either to a central AGN or the surrounding ICM, which should be isolated when considering the nature of the scatter in the values from the hot gas. In fact, the three labelled sources in Fig. 1 were already identified by O’Sullivan et al. as powerful AGNs. Fig. 1 also shows that due to the volume-limited nature of the ATLAS sample our objects fall short from covering the entire range of optical luminosity spanned by the sample of O’Sullivan et al., although only by a factor two.
We now move on from this first diagram and start making use of our ATLAS database by considering in Fig. 2 the near-infrared K-band luminosity of our sample galaxies, which corresponds much more closely to the total stellar mass than does the B-band optical luminosity, in particular in the occasional presence in early-type galaxies of a younger stellar sub-component. Furthermore, in Fig. 2 we now adopt the currently best estimates for the distance of our sample galaxies, not only to compute the luminosity from the 2MASS apparent K-band magnitudes but also to rescale the values given in O’Sullivan et al.. Compared to Fig. 1, in the diagram the bright galaxies with the largest X-ray luminosities appear to distance themselves more from other objects with exceeding the value expected from discrete sources. All such galaxies are either deeply embedded in the intracluster medium of the Virgo cluster (NGC 4486, NGC4406) or are the central member of their own group of galaxies and show a rather extended X-ray halo (NGC 4636 and NGC 5846; Matsushita, 2001; Mulchaey et al., 2003; Santos, Mendes de Oliveira, & Sodré, 2007). This suggests that the ICM contributes largely to the values of these objects, which should then be flagged in the remainder of our analysis as in the case of the AGNs identified above. We also note that in the diagram is a more accurate predictor for the contribution of LMXBs to the total values, although whether most or just a few of our sample objects could be considered to possess an X-ray halo still depends, like in Fig. 1, on the adopted calibration for the from LMXBs. Yet, the Kim & Fabbiano (2004) calibration appears problematic since according to it there would appear to be a considerable fraction of galaxies in the diagram whose values are nearly half what would be expected from LMXBs alone. The more recent calibration of Boroson, Kim, & Fabbiano (2011) does not suffer from this bias, which is why we will adopt it throughout the rest of our analysis.
To further understand the nature of the emission in our sample galaxies, in Fig. 3 we have placed them in the vs. diagram. This is a near-infrared and integral-field version of the vs. diagram introduced by Canizares et al. (1987), where the combination of and allows to trace the X-ray luminosity that is expected if the energy lost by the hot gas through radiation is compensated by the thermalisation of the kinetic energy that the stellar ejecta inherit from their parent stars. Such an input rate of kinetic energy would indeed correspond to the stellar-mass loss rate (traced by ) times the specific kinetic energy associated with the stellar random motions (traced by ). More specifically, we take . To show the prediction of such a model in Fig. 3 we computed the average ratio between the K- and B-band luminosity of our sample galaxies () and assumed a mass-loss rate of (as given by Ciotti et al., 1991) with an average stellar age (in units of 15 Gyr) of 12 Gyr. By adding to this model the expected contribution from the unresolved LMXB population according to the calibration of Boroson et al., which we plotted in the diagram using the best fitting near-infrared Faber-Jackson correlation (Faber & Jackson, 1976) between the and values for our low X-ray resolution sample (), we finally obtain the total values that ought to be compared with the data shown in Fig. 3.
Incidentally, except for two objects, all the slowly-rotating galaxies in our low X-ray resolution sample that are massive enough to heat up the stellar-loss material at X-ray temperatures and lead to detectable X-ray luminosities against the LMXBs background (in Fig. 3 we considered a generous lower limit for of , corresponding to ) align themselves along the lines of such a simple model, falling either directly on top or just above the predicted values. We note that a situation where the radiation losses of the hot atmosphere are balanced by the thermalisation of the kinetic energy of the stellar ejecta would implicitely require that the stellar-mass loss material is steadily removed from the galaxy, in order the prevent the accumulation of gas and catastrophic cooling. For a typical old and massive early-type galaxy such a quasi-static situation could be maintained by SNe type Ia explosions, since at their present-day rate the SNe energy output is comparable to the energy that is necessary to extract the stellar-mass material (see, e.g., the spherical testcase of Pellegrini, 2012, with a that would correspond to ).
On the other hand, most of the fast-rotating galaxies in this sample fall short of the X-ray luminosity expected from stellar ejecta that through shocks and collisions contribute to heat the hot gas. If we consider the early finding of Eskridge, Fabbiano, & Kim (1995a, b) that S0s and flat Es have a lower X-ray luminosity compared to rounder elliptical of similar optical luminosity, and that when early-type galaxies are split by their degree of rotational support 94% of S0s and 66% of Es fall in the fast-rotating category (Paper III), then it seems natural to ask whether the apparent X-ray deficiency observed in Fig. 3 for this class of objects is actually driven by the parameter or by the apparent flattening .
For the objects in our low X-ray resolution ATLAS sub-sample, Fig. 4 shows how the ratio of their total X-ray luminosity to the expected contribution from unresolved discrete sources only or to the predicted X-ray luminosity when including also the emission from stellar-mass loss material heated to the stellar kinetic temperature correlates with the parameter, apparent flattening and dynamical mass . A Spearman rank analysis reveals that both the and ratios correlate with a modest degree of significance with both and (see Fig. 4 for the probability of the null hypothesis for each correlation). The excess would appear to be also somehow correlated to the dynamical mass , although no such trend is observed in the case of the deficit. In other words, pretty much independently of their dynamical mass, increasingly rounder and more slowly-rotating galaxies in our low X-ray resolution ATLAS sub-sample appear to show also larger values that progressively exceed what is expected from the LMXB population and eventually become consistent with the total predicted X-ray luminosity from LMXBs and shock-heated stellar ejecta. Conversely, the flatter and more rotationally-supported a galaxy, the larger its deficit from the expectation of this simple model.
The most notable outliers from the trends observed in Fig. 4 not only support this interpretation of the data, but also appear to suggest that the intrinsic flattening of a galaxy, rather than its degree of rotational support, is what really drives the deficiency of fast rotators. When considering the correlations betwenn and either or , which are the most significant in Fig. 4, we note that three of the roundest galaxies (NGC 3193, NGC 3640 and NGC 4753), are likely to be intrinsically flat objects that are viewed face-on, which would move to the right of the central panels of Fig. 4 if seen from a less improbable angle. This is supported by their large values for , which indicates that they are in fact rotationally-supported systems just like all the other fast rotators with which they share the trends with that is observed in the left panels of Fig. 4. Conversely, in these same panels, the only slow rotators with values consistent with just emission from LMXBs or that fall short of the predicted total from LMXBs and hot gas sustained by the thermalisation of the kinetic energy of stellar ejecta, NGC 4365 and NGC 5322, are incidentally the flattest objects among the class of slow rotators. In fact, that these (fairly massive) objects appear to align themselves with the rest of the fast rotators in the anti-correlations between the apparent flattening and either or further suggests that it is their intrinsic shape, rather than their dynamical state, which determines whether they are capable to retain a halo of hot gas. When such flat slow rotators are excluded from the Spearman rank analysis the correlations involving in Fig. 4 become more significant, and the same holds when excluding the face-on fast rotators from the correlations involving . Finally, we note that even a modest contamination from a central AGN or the the intra-group or -cluster medium to the total X-ray in the least massive objects could significantly impact on the total values for these object. In turn, this would generally weaken the significance of any correlation between or and the or ratios, since low-mass galaxies generally tend to be fast rotators (Paper III). The possible impact of AGNs or the intra-group or -cluster medium is most noticeable in the correlation between and the deficit (lower left panel of Fig. 4), where the majority of the most rotationally-supported galaxies with values near unity have values below of . At such a low regime the typical AGN X-ray emission of early-type galaxies could be comparable to the predicted values for and (Fig. 3). For instance, the median value for the ATLAS sample galaxies that appear in the study of Liu (2011, 42 objects) is .
To summarize the results of this subsection, in Fig. 5 we redraw the same and diagrams presented in Figs. 2 and 3, but now with elliptical symbols coded by colour and flattening according to the values of and the of the galaxies they are representing. This is done in order to convey more visually our preliminary conclusion, based on our low X-ray resolution sample, that only round slow rotators can sustain a halo of hot gas from the thermalisation of the kinetic energy carried by their stellar-mass loss material whereas fast rotators (and apparently also flat slow rotators) have progressively fainter X-ray halos the flatter and more rotationally-supported they appear.
3.2 High X-ray Resolution Sample
The previous conclusions on the more limited ability of fast-rotating, or even more generally just flat, early-type galaxies to retain their halos of hot gas have to be taken with care given that they are based on X-ray data of rather coarse spatial resolution where the contribution from a central AGN, the intra-group or -cluster medium or LMXBs to the total values can only be estimated, and which in the case of LMXBs is known to be the subject of considerable scatter.222 If we consider that globular clusters (GCs) may be the birthplace of the LMXB population in galaxies (e.g. Sarazin, Irwin, & Bregman, 2000; White, Sarazin, & Kulkarni, 2002; Zhang, Gilfanov, & Bogdan, 2012) and that in turn lenticular galaxies may have a lower specific frequency of GCs than elliptical galaxies (Kundu & Whitmore, 2001a, b), it may be even possible that a systematically smaller number of LMXBs in fast rotators contributes to their lower X-ray luminosity. Nearly all (94%) S0 galaxies and most (66%) Es belong in fact to the class of fast-rotating galaxies (Paper III). It is therefore important to ask whether our results hold when such factors can be properly isolated through the use of Chandra or XMM.
Recently, Boroson, Kim, & Fabbiano (2011) published one of the largest samples of early-type galaxies with consistently derived measurements, which overlaps well with the ATLAS sample and excludes the dominant members of groups and clusters of galaxies. For the objects that we have in common with the sample of Boroson et al., Fig. 6 shows how their and values compare to each other, thus allowing to directly check whether the hot-gas luminosity corresponds to the input rate of kinetic energy from stellar-mass loss. The use of the same color coding and squashing of Fig. 5 for our data points makes it easy to appreciate in Fig. 6 that, except for one, all the slow rotators in this ATLAS sub-sample agree still remarkably well with the same simple prediction for (Fig. 3 and right panel in Fig. 5). Fast rotators, on the other hand, display values that fall systematically short of such expected values, as does also NGC 4365, which is the flattest slow rotator in the high X-ray resolution sample and was also a low-lier compared to the rest of the slow rotators in Fig. 5 (for a more detailed picture of this galaxy, see also Davies et al., 2001, Statler et al., 2004 and van den Bosch et al., 2008).
If Fig. 6 already seem to supports the picture that was drawn in §3.1 using ROSAT or Einstein data, Fig. 7 further confirms that the X-ray deficiency of fast rotators, as quantified by the ratio, correlates indeed with their degree of rotational support traced by the parameter and apparent flattening . The connection with in particular is remarkably significant given the modest number of objects at our disposal, and the general trend in the left panel of Fig. 7 can be recognised also in the lower-left panel of Fig. 4 that showed the correlation between and the deficit for our low X-ray resolution sample, if one were to disregard (with the benefit of hindsight) the least massive fast rotators. The behaviour of the outliers to the present correlations with and is also similar to what is observed in Fig. 4, with the flat slow rotator NGC 4365 falling below the main trend between and but aligning itself with the majority of the fast rotators in the vs. diagram and with a few round fast rotators (NGC 3384, NGC 4278 and NGC 3379) acting in the opposite way, which adds to the idea that the ability of a galaxy to retain a halo of hot gas relates primarily to its intrinsic shape (Ciotti & Pellegrini, 1996; D’Ercole & Ciotti, 1998). In fact, that the correlation between the ratio and is more significant than in the case of the correlation with suggests that may be a good indicator for the intrinsic flattening of fast rotators, if this is indeed the main parameter driving the X-ray deficit.
It is interesting to observe in the right panel of Fig. 7 how the direct measurements derived from the Chandra data now reveal that the dynamical mass must also play a role in determining how much hot gas a galaxy can retain. This is not unexpected, since the ratio of the energy that could be injected by SNIa in the interstellar medium and the energy required to steadily extract the mass lost by stars in itself depends on the galaxy mass, so that the less massive galaxies would host SN-driven outflows and thus possess systematically less massive and X-ray fainter hot-gas atmospheres. In principle, it may be even possible that the correlation with between and is partly driven by the correlation with , since low rotators tend to be more massive then fast rotators (Paper III). Yet, a large appears to be only a necessary, but not sufficient condition for having a bright X-ray halo. Almost all fast rotators in our high X-ray resolution sample are nearly as massive, within a factor two, as the slow rotator NGC 4552 and yet they all display considerably smaller values. Conversely, the flat slow rotator NGC 4365, shares with the flat rotators of our sample a relatively faint hot gas halo despite being more massive, by up to an order of magnitude, than nearly all of them.
We should also stress that the deficiency does not correlate with any of the indicators for the environmental galactic density that were adopted in Paper VII, which also confirms that the impact the intra-cluster or group medium in our high-resolution X-ray sample has been properly avoided by following the selection of Boroson et al.
Boroson, Kim, & Fabbiano (2011) also provide estimates for the hot gas temperature , which we can use to further test whether stellar motions are the main heat source for the X-ray emitting gas (see, e.g., Pellegrini, 2011). If this is the case, the gas temperature should relate to the stellar velocity dispersion since this traces the stellar kinetic energy. Traditionally has been compared to where is the mean particle mass of the gas ( is the proton mass and for Solar abundance) and was taken as the central stellar velocity dispersion, although should relate more strictly to the stellar-density weighted average of the intrinsic, three-dimensional velocity dispersion across the entire galaxy. Rather than venturing directly in the construction of the stellar dynamical models needed to properly evaluate this quantity, in Fig. 8 we plot the values from Boroson et al. against our integral-field measurements for the stellar velocity dispersion within one effective radius , which should already fare better than the central in tracing the global stellar kinetic energy.333 Pellegrini (2011) shows that the kinetic temperature associated to the stellar motions can be expressed more generally as , where relates to the dark-matter fraction and to the shape of both the stellar and dark-matter density radial profiles. is always and typically range between and (see Fig. 2 of Pellegrini, 2011). This would include the value that the ratio would take if the central is measured within one eighth of the effective radius (that is, 0.76, according to eq. (1) of Cappellari et al., 2006), as is often the case in the literature. is therefore a good approximation for , just as it can be shown to be for in the case of . In fact, by adopting the dividing line between fast and slow rotators introduced in Paper III, we note in Fig. 8 that the hot-gas temperature of the slow rotators in our high X-ray resolution sub-sample correlates well with , with values that are on average above by 50%. On the other hand this does not seem to be the case for fast rotators, which, with the exception of the particularly cold and under-luminous X-ray halo of NGC 821, show similar values between 0.25 and 0.35 keV across a range of values and where seems to constitute just a lower limit for the temperature of the hot gas.
As already pointed out by Pellegrini (2011), the excess of slow rotators could be interpreted as a need for an extra source of heating, possibly due to inflowing and adiabaticaly compressed gas in the central regions of these objects, which would be consistent with this excess being itself proportional to . Yet, adding an extra heat source for the gas may not seem strictly necessary given that the thermalisation of the stellar kinetic energy can already account for the observed X-ray luminosity (see Fig.6 and the right panel of Fig. 8). In fact, the agreement between the observed values and those traced by could be easily maintained even if we were to scale up the kinetic energy traced by by simply considering a smaller value for the stellar-mass rate , as indicated for instance by the observations of Athey et al. (2002) who give rather than the rate of that we currently adopt from Ciotti et al. (1991).
Why fast rotators do not follow the same relation of slow rotators and instead scatter in the 0.25 – 0.35 keV interval is more difficult to understand. One possibility is that in fast rotators may not trace particularly well the intrinsic luminosity-weighted velocity dispersion that would relate to the relative velocity involved in those shocks between stellar ejecta that are supposed to thermalised the kinetic energy inherited the stellar-mass loss material, in particular since measures the total kinematic broadening within one and thus factors in also the impact of net rotation. On the other hand, the hot gas of fast rotators may be in an altogether different dynamical state compared to slow rotators, being most likely outflowing due to their flattening (Ciotti & Pellegrini, 1996) or even driven out in a supersonic wind powered by an additional number of supernovae associated with a recently formed population of stars. If gravitational heating associated to inflows is the reason why slow rotators lie above the line, then the possible absence of inflows in fast rotators may explain why a good fraction of them appear to fall close to this limit. On the other hand, the color-coding of the symbols in Fig. 8 according to their global age (within one , from McDermid et al. in preparation) points to an extra input of energy from supernovae as the reason why another good fraction of fast rotators display values well above what is expected from their (by a factor two), since most of these objects show the presence of young populations. The right panel of Fig. 8 shows that such younger objects seem also to show brighter X-ray halos compared to older fast rotators, although more low-mass systems would certainly be needed to confirm this impression.
Finally, we should also contemplate the possibility that the kinetic energy inherited by the stellar mass-loss material is not thermalised with the same efficiency in slow and fast rotators, in particular if we consider that this process is supposed to happen through shocks as the ejecta collide with each other or interact with an already existing hot medium. For instance, compared to the case of a dynamically supported system, the stellar ejecta may find it harder to find each other in a galaxy with net rotation, and eventually collide only at small angles and relative velocities. Furthermore, if we consider that the hot medium of a rotating galaxy may have inherited part of the angular momentum of the stars from which it originated444 This is in fact quite possible given that the hot and warm phases are most likely in pressure equilibrium (as the inferred values for the ionised-gas density suggest, see Mathews & Brighenti, 2003, and references therein) and considering that fast rotators can display extended ionised-gas disks that rotate in the same sense as the stars. , then at a given orbital velocity the newly shed ejecta of a rotating galaxy may shock with the hot gas on average at a smaller velocity than in the case of the ejecta of a non-rotating galaxy, which would have imprinted no net rotation on its hot-gas halo. According to these qualitative arguments we should expect that the kinetic energy of stellar ejecta is thermalised less efficiently in fast rotators. Ciotti & Pellegrini (1996) quantified this effect in the limit where none of the stellar rotational motion is thermalised, finding that the hot-gas temperature could decrease by up to 30%. Incidentally, in Fig. 8 this seems to be fairly close to the limit by how much colder the hot gas of fast rotators can get compared to the hot gas of slow rotators.
To conclude we note that in Fig. 8 the flat and X-ray underluminous slow rotator NGC 4365 displays a value in line with the behaviour of the other members of its class. This further indicates that if on one hand flattening could hamper the ability of a galaxy to retain its hot gas, on the other hand the degree of rotational support of a galaxy may also reduce the efficiency with which the kinetic energy inherited by the stellar ejecta is thermalised in the hot gas, which may be the case of fast rotators but not of a dynamically hot system such as NGC 4365.
3.3 Additional High X-ray Resolution Data
If the analysis presented in the previous two sections suggests a different specific hot-gas content of fast and slow rotators, we need to keep in mind that these two kinds of early-type galaxies are rather relegated to low and high values for their dynamical mass , respectively (Paper III). In fact, this is in particular the case of the objects in the high X-ray resolution ATLAS sub-sample that revealed also a possible role of in driving the X-ray deficit of early-type galaxies (Fig. 7, right panel), with the most massive fast rotators showing values that appear nearly as consistent with the input rate of kinetic energy from stellar-mass loss as in the case of slow rotators (Fig.6 and the right panel of Fig. Fig. 8, where is fairly traced by ). It would be therefore useful to observe the behaviour of other relatively massive fast rotators with estimates, as well as of slow rotators with lower values than presently found in our high X-ray resolution sample, even if this will entail a loss of consistency in the X-ray or optical measurements.
For this reason, in Fig. 9 we added to the – diagram also the other galaxies in the Boroson et al. sample, using their adopted central values for as well as their and values, together with four more objects in the ATLAS sample with measurements from other relatively large studies based on Chandra data, which have a few objects in common with the Boroson et al. that allow for some control on the consistency between estimates. These ATLAS objects, for which we can rescale the values using our adopted distance estimates and have access to the rest of our ATLAS imaging and integral-field spectroscopic measurements, are the slow rotator NGC 6703 from the field sample of Mulchaey & Jeltema (2010) and the fast rotators NGC 4459, NGC 4494 and NGC 5845 from Athey (2007). Other compilations based on Chandra data, such as those of Diehl & Statler (2007) or Sun et al. (2007) either did not include other ATLAS objects than those already considered by Boroson et al., or only featured ATLAS objects that are deeply embedded in their intra-cluster or group medium. Based on the deviations observed for the Boroson et al. objects that are in the ATLAS sample (and have thus been observed with SAURON) we expect that for their remaining objects the and measurements would be within 0.05 and 0.12 dex, respectively, of the values that they would have if they had been selected as an ATLAS target, with no significant systematic deviations. Mulchaey & Jeltema have only two objects in common with the Boroson et al. sample, with values that are consistent with theirs (assuming the same distance). Athey (2007) has 12 objects in common with Boroson et al., but in this case the reported values are systematically lower by 0.18 dex, albeit with a similar scatter. Although these deviations are modest, for the sake of consistency, we did not consider objects that are not in the ATLAS sample with other than those of Boroson et al.
To assign the Boroson et al. galaxies that are not part of the ATLAS sample to the slow- or fast-rotator families, we still adopted the dividing line of Paper III but now resort to an indirect estimate for and adopt the 2MASS K-band measurement of the minor- to major-axis ratio k_ba as a measure of the global flattening . To estimate we use the relation between and the ratio given in Paper III (eq. B1) and in turn derive the ratio by applying the correction of Cappellari et al. (2007, their eq. 23) to our best estimate of the ratio from long-slit data in the literature. The only exception to this procedure is represented by NGC 720, which is one of the “special” objects observed in the course of the SAURON survey (de Zeeuw et al., 2002) and for which Cappellari et al. already reported values for , and . Based solely on its value, NGC 720 was considered a fast-rotator by Emsellem et al. (2007), albeit already as an exceptional one by Cappellari et al., and falls now in the slow-rotator family when is compared to .
In Fig. 9 the objects with ATLAS data (from Mulchaey & Jeltema, 2010 and Athey, 2007) are colour-coded as in Fig. 8 according to our global age estimates (from McDermid et al. in preparation) whereas for the remaining galaxies of the Boroson et al. sample we limited ourselves to identify only objects with clear signs of recent star-formation episodes. These are NGC 1316, for which Silva, Kuntschner, & Lyubenova (2008) used near-IR data to establish a mean stellar age of Gyr, and NGC 1052, for which Fernández-Ontiveros et al. (2011) report a global value for the NUV-r colour of 4.9, typical for early-type galaxies that experienced a recent star-formation episode (e.g. Kaviraj et al., 2007). For the remaining sample galaxies of Boroson et al. it is difficult to draw a consistent picture of their stellar age, given the varying quality of the absorption-line strength measurements found in the literature (not always corrected for nebular emission) and the different choices for the models used to derive the stellar population properties.
The inclusion of these additional objects in our analysis confirms that fast rotators tend to have fainter X-ray halos than slow rotators with similar expectation values (from the thermalisation of the kinetic energy brought by the stellar-mass loss material), although Fig. 9 further indicates that such a discrepancy may reduce or even disappear as the dynamical mass of galaxies increases. As in the previous exceptional cases of NGC 4365 and NGC 5322, the slow rotator with the relatively faintest hot-gas halo among the additional objects, NGC 1427, also happens to be a relatively flat galaxy (k_ba=0.7; see D’Onofrio et al., 1995 for its long-slit kinematics). On the other hand, Fig. 9 introduces in our picture two flat and well-studied slow rotators, NGC 720 and NGC 3923 (with a k_ba value of 0.55 and 0.64, respectively; for the kinematics of NGC 3923 see Pellegrini, Held & Ciotti, 1997), for which appears well in line with the prediction. Thus, it would appear that flattening can, but not necessarily does, drive a deficit in hot-gas content of slow rotators. Finally, younger stellar populations are found in three out of the four fast rotators that in Fig. 10 sit the closest to the prediction, consistent with a possible extra energy input from SNe in younger objects.
By combining photometric and integral-field spectroscopic measurements with X-ray data of both low and high spatial resolution we have found that flat and rotationally-supported early-type galaxies (i.e., fast rotators) tend to fall systematically short of the X-ray luminosity that is expected if the hot gas radiation is sustained by the thermalisation of the kinetic energy that the stellar-mass loss material inherits from their parent stars. On the other hand, non- or slowly-rotating rotating galaxies (i.e., slow rotators), which are likely to be also intrinsically fairly round (Weijmans et al., in preparation), generally display values in line with this model, and in fact also show hot-gas temperatures that correspond well to the stellar kinetic energy, as traced by the global stellar velocity dispersions within the optical regions of these galaxies . This is not the case for fast rotators, which show nearly the same hot-gas temperature across a range of values except for a few objects with younger stellar population that would appear to have hotter and also brighter gas halos, possibly owing to the additional energy input from more recent supernovae explosions. Such a discrepancy in the hot-gas content of fast and slow rotators appears to reduce, or even disappear, for large values of the dynamical mass (beyond , corresponding in the ATLAS sample to ), whereas the few slow rotators with values considerably below the bar set by the thermalisation of the kinetic energy brought by stellar-mass loss material always appear to be relatively flat. This last result could suggest that it is the intrinsic shape of galaxies, rather than their degree of rotational support, that reduces their ability to retain the hot gas (Ciotti & Pellegrini, 1996), although care is needed as there exist also flat slow rotators (albeit very massive) with perfectly in line with the expectation values.
The situation observed for slow-rotators, whereby , is likely due to the fact that currently in these galaxies, so that most the SNe type Ia energy is presumably used to steadily remove the stellar-mass loss material and thus maintain a quasi-static hot-gas atmosphere.555 In fact, as discussed in Pellegrini & Ciotti (1998) and Pellegrini (2012), even in such a quasi-static situation the gas flow is likely to be decoupled into an outflowing outskirt and a central inflow where radiative cooling outstrips the SNe energy feedback. On the other hand, by virtue of their flattening and possibly also of their lower mass, exceeds in fast rotators so that these systems are actually degassing and thus display more tenuous and X-ray fainter hot-gas halos.
A systematic offset in the specific hot-gas content of fast and slow rotators may be relevant to understand also the disparity between the incidence of molecular gas (Young et al., 2011) and associated episodes of recent star formation (Kuntschner et al., 2010) in these two classes of early-type galaxies (with slow rotators being devoid of both molecular gas and young stars), which in turn may link to an altogether different origin and fate for their gas (Sarzi et al., 2007). Indeed, much as the intra-cluster or group medium may suppress the cold-gas reservoir of galaxies falling in crowded galactic enviroments and drive the cosmic rise of lenticular galaxies (e.g. Dressler et al., 1997), the presence of a halo of hot gas around a galaxy may inhibit the star formation triggered either by the acquisition of gas-rich satellites or by the accumulation of stellar-mass loss material. This is because small amounts of fresh gas would quickly heat up and evaporate in such an hot medium, by thermalising its kinetic energy as it moves against the hot gas and inevitably shocks with it (e.g. Parriott & Bregman, 2008, for stellar winds), and possibly also through thermal conduction (e.g. Nipoti & Binney, 2007, for accreted gas). Furthermore, if hot-gas halos inherit part of the angular momentum of the galaxies that they enshroud (through the interaction between the hot gas and the stellar-mass loss material, as advocated by Marinacci et al., 2011 in the case of disk galaxies), the accretion of gas from satellites could proceed differently depending on whether such material travels with or against the hot gas. In the latter case, for instance, ram-pressure stripping would be more efficient, since its strength depends on the square of the velocity relative to the hot medium.
Based on the distribution of the misalignment between the angular momentum of stars and ionised gas, Sarzi et al. (2006) already recognised (based on the smaller SAURON representative sample) a possible difference in the relative role of external gas accretion and stellar-mass loss in fast and slow rotators, where in the latter class the warm gas would appear purely external in origin. Using the ATLAS sample, Davis et al. (2011) not only confirmed this difference but also highlighted the impact of both a dense galactic environment (that is, inside Virgo) and a large mass (for or ) in suppressing the acquisition of external gas by fast rotators. A bias towards finding corotating gas and stars in the most massive fast rotators is also consistent with the conclusion of Shapiro et al. (2010) that there exists two modes of star-formation in fast rotators, such that in low-mass systems star-formation is found throughout the galaxy and is triggered by minor mergers whereas in galaxies more massive that (which in the ATLAS sample corresponds to ) such an activity occurs only in central or circumnuclear regions and the ionised-gas rotates in the same sense as the stars. Interestingly, slow rotators would appear to accrete their ionised-gas despite tending to be more massive and to reside in denser galactic environments than fast rotators (Davis et al., 2011). Perhaps, the present tendency of slow rotators to experience minor mergers can be regarded as a snapshot of their overall assembly history, since numerical simulations indicate that such a constant “bombardment” is required in order to reduce the specific stellar angular momentum to the values observed in this class of objects (Bois et al., 2011; Khochfar et al., 2011, hereafter Paper VI and VIII, respectively).
In any event, the gas acquired in slow rotators does not appear to condense to a molecular phase and form new stars. Instead it must join the hot interstellar medium, most likely through thermal conduction as first suggested by Sparks, Macchetto, & Golombek (1989) and later quantified by Nipoti & Binney (2007).666 As a reference, Nipoti & Binney find that gas clouds of mass below and (at most, they are more likely to be 100 times smaller) would eventually evaporate as they fall in the hot medium of galaxies of baryonic mass of and , respectively. In fact, the lack of slow rotators with corotating gas and stars means that a similar fate also awaits any stellar-mass loss material injected in such hot-gas rich systems, whereas the systematic deficiency of fast rotators may allow for some fraction of this material to cool down amidst their more tenuous hot medium.
A good way of testing whether fast rotators, as opposed to slow rotators, can manage to cool down a fraction of their stellar mass-loss material - thus possibly being more efficient at recycling such gas into new stars - would be to compare the specific content of the diffuse dust in these two classes of objects. Observed at infrared wavelengths, such diffuse dust can indeed trace more directly the material lost by stars than done for instance by the ionised-gas component of the interstellar medium, in particular since the nebular flux may reflect also a varying UV ionising flux from a number of different sources (Sarzi et al., 2010) whereas the dust of quiescent galaxies is mainly heated by the optical stellar radiation (e.g. Sauvage & Thuan, 1992). Furthermore, the launch of the Herschel Space Observatory has now made it possible to probe the colder dust component of early-type galaxies and thus estimate more accurately their total dust mass than previously allowed by less sensitive instruments covering a more limited wavelength range, such as IRAS, ISO or Spitzer (see, e.g., Goudfrooij & de Jong, 1995; Ferrari et al., 2002; Athey et al., 2002; Temi, Brighenti, & Mathews, 2007).
Recently Smith et al. (2012) published estimates based on Herschel data for a sample of 62 nearby early-type galaxies, which include 57 objects from the ATLAS sample. Unfortunately, within this ATLAS sub-sample there are only five galaxies with both secure detections and measurements, making a direct comparison between the specific dust content and the X-ray properties in our sample galaxies not very meaningful. On the other hand, since the ATLAS objects with estimates have a range of and values that matches well that of the high X-ray resolution ATLAS sub-sample and of the additional objects that we have analysed in § 3.2 and § 3.3, respectively, we can use our results to gauge the deficiency of the Smith et al. galaxies in our sample, and correlate that with their specific dust content.
For this reason, in Fig. 10 we show how the ratio correlates with the specific dust content of the ATLAS objects analysed by Smith et al. This is similar to Fig. 12 of that work, except that here we use as a measure of the total stellar mass and regard as an optical tracer of the deficiency compared to what is expected from the thermalisation of the kinetic energy of the stellar-mass loss material (§ 3.2). Furthermore, in Fig. 10 we also show objects with only upper-limits on and color-code those with secure detections according to whether their ionised-gas component shows an angular momentum that is considerably misaligned compared to that of the stars (i.e. by more than 30).
Fig. 10 shows that fast rotators generally extend to larger values for the specific dust content than slow rotators, with very few fast rotators showing values reaching down to the values observed in the more dust-poor slow rotators. A similar message is conveyed also by the upper-limits on , in that they tend to constrain the specific dust content to consistently lower values as decreases. In fact, all the slow rotators with secure detections in Fig. 10 show largely decoupled nebular and stellar motions consistent with an external origin for their gas, which suggests that the specific content of internally generated dust in these systems is probably much lower than currently inferred from their total far-infrared emission. On the other hand, most of the fast rotators with a specific dust content in excess of what is observed in slow rotators (that is for ) display well-aligned gas and stellar angular momenta (in nine out of ten cases, corresponding at most to a 20% fraction of systems where the gas was accreted), suggesting that such a larger dust content originates from stellar-mass loss material. This finding does not appear to be a result of sample selection. For instance, taken together all the fast rotators with detections are consistent with a even split between objects with internal and external origin for their gas just like for the entire ATLAS population of fast rotators (Davis et al., 2011). Furthermore, the fraction of objects in high-density environments (i.e., within the Virgo cluster) or with relatively large mass (i.e., ) does not change when considering all the fast rotators with detections or just those with a higher specific dust mass content. This suggests that these otherwise important factors in determining the origin of the gas cannot entirely explain why the dustier fast rotators in Fig. 10 also happen to have co-rotating gas and stars.
Thus, once complemented with our kinematic information, the Herschel data of Smith et al. would appear to confirm that fast rotators have a larger specific dust content which relates to the fact that some fraction of the stellar-mass loss material is allowed to cool in the less hostile hot-gas medium of these objects, compared to the case of slow rotators. Such an ability of preserving some of their stellar-mass loss material from joining the hot medium could also contribute (together with our previous conjecture on the possible role of a rotating X-ray halo) to the finding of Davis et al. (2011) that the most massive fast rotators do not seem to further acquire external gas, even when found in relatively sparse galactic environments. In fact, it may be even possible that in time such fast rotators build up a gaseous disk that is sufficiently massive to absorb counter-rotating gas from smaller satellites, which, on the other hand, would perturb more significantly the structure of less massive fast rotators. Yet, it is important to keep in mind here that not all fast rotators display dust or ionised-gas, pretty much independent of galaxy mass and only mildly dependent on galactic environment (see Tab. 2 of Davis et al., 2011). Even low-mass systems in the field, where both external accretion and internal recycling could occur, can be devoid of any gas. Apparently other mechanisms, beyond the scope of this paper, must operate to also remove gas from early-type galaxies.
A different specific hot-gas content of slow and fast rotators and the likely different fate of their stellar-mass loss material could also contribute to understand the fraction of rejuvinated early-type galaxies that has been estimated in much larger spectroscopic surveys of more distant galaxies. For instance, based on a consistent treatment of Sloan Digital Sky Survey spectra for over 3,000 morphologically selected early-type galaxies, Thomas et al. (2010) revised the conclusion of Thomas et al. (2005, based on a more heterogeneous sample) and found that the impact of environment on the fraction of rejuvinated galaxies decreases with the mass of galaxies. In practice, for values of above the fraction of galaxies with evidence of recent star formation steadily decreases till reaching values consistent with zero pretty much independent of galactic environment (see Fig 8 of Thomas et al., 2010). Given that our results are based on galaxies exceeding similar values of (Fig. 7) we suggest that the internal processes regulating the star-formation history at the massive end of the early-type galaxy population could be driven by the hot-gas content of these objects and their ability to recycle the stellar-mass loss material into new stars, which in turn could relates to the relative number of slow and fast rotators as well as the different degrees of deficiency displayed by the latter class of objects. For instance, where the fraction of rejuvenated objects in Thomas et al. starts to be negligible, around , is incidentally also the point from which the fraction of slow rotators in the ATLAS sample shoots-up rapidly from values around 5–15% to a fraction of 60% for (Fig. 11 of Paper III, but see also Fig. 10 of Cappellari et al. 2012c, hereafter Paper XX), by which mass the hot-gas content of fast rotators may not longer differ too much from that of slow rotators.
To conclude, we note that gravitational heating of infalling gas during minor mergers (in particular around the most massive galaxies, Dekel & Birnboim, 2006; Khochfar & Ostriker, 2008; Johansson et al., 2009) is expected to deposit significant amount of energy in the hot gas. The relative contribution of this process compared to the thermalization of the kinetic energy inherited by the stellar-mass loss material remains to be quantified, but unless to consider a very small efficiency for the latter mechanism, our results would suggest that - at least when considering the hot gas around galaxies that is not bounded by a group or cluster potential - gravitational heating presently does not play an dominant role.
For a galaxy, being embedded in a corona of hot, X-ray emitting gas can be a key element determining its recent star-formation history. A halo of hot gas can indeed act as an effective shield against the acquisition of cold gas and can quickly absorb any stellar-mass loss material, thus preventing its cooling and consequent recycling into new stars. In other words, the ability to sustain a hot-gas halo can contribute to keep a galaxy in the so-called red-sequence of quiescent objects.
Since the discovery by the Einstein observatory of such X-ray halos around early-type galaxies, the origin of the X-ray emission and the precise amount of hot gas around these galaxies have been the matter of long debates, in particular when trying to understand the rather loose correlation between the optical and X-ray luminosity of early-type galaxies. In the past, this situation resulted from the limited ability to isolate in earlier X-ray data the additional contribution from an active nucleus, the unresolved population of X-ray binaries and the X-ray emission from the intra-cluster medium, although the use of heterogeneous optical data may have also contributed to such an impasse. Today, with new X-ray telescopes such as Chandra or XMM and large collections of both photometric and spectroscopic data it is possible to gain more insight on the hot-gas content of early-type galaxies.
In this paper we have combined the homogeneously-derived photometric and spectroscopic measurements for the early-type galaxies of the ATLAS integral-field spectroscopic survey with measurement of their X-ray luminosity from X-ray data of both low and high spatial resolution, finding that the hot-gas content of early-type galaxies can depend crucially on their dynamical structure. Specifically, in the framework of the revised kinematic classification for early-type galaxies advanced in the course of both the SAURON and ATLAS surveys (Emsellem et al., 2007, Paper III), we find that:
Slow rotators have hot-gas halos with X-ray luminosity values that are generally consistent - across a fair range of values - with what is expected if the hot gas radiation is sustained by the thermalisation of the kinetic energy that the stellar-mass loss material inherits from their parent stars, so that closely follows .
Fast rotators display values that tend to fall short of the prediction of such a model, and the more so the larger their degree of rotational support as quantified using the integral-field parameter.
Such a deficiency in fast rotators would appear to reduce, or even disappear, for large values of the dynamical mass (beyond ), whereas the few slow rotators with values considerably below the bar set by the thermalisation of the kinetic energy brought by stellar-mass loss material always appear to be relatively flat, nearly as much as on average fast rotators do.
Still consistent with a stellar origin for the heat needed to sustain their hot-gas emission, slow rotators also display hot-gas temperatures that correspond well to the global stellar kinetic energy, as traced by the stellar velocity dispersions measured within the optical regions .
Fast rotators, on the other hand, show similar values of across a range of , except for a few objects with younger stellar population that would appear to have both hotter and brighter X-ray halos, possibly owing to the additional energy input from more recent supernovae explosions.
These results indicate that the thermalisation of the kinetic energy brought by stellar-mass loss material, presumably through shocks and collisions of the ejecta with each other and the hot interstellar medium, sets a natural upper limit to X-ray emission originating from the hot gas bounded by the gravitational potential of present-day quiescent galaxies. This is observed mostly in the case of slow rotators, where the energy brought by SNe type Ia is similar to that needed to steadily extract most of the mass lost by stars, thus helping maintain a quasi-static hot-gas atmosphere. On the other hand, the fact that fast rotators as a class are most likely to be intrinsically flatter than slow rotators (with fast rotators being nearly as flat as spiral galaxies and slow rotators being much rounder and possibly triaxial, Weijmans et al. in preparation) leads us to interpret the deficit of fast rotators in light of the scenario of Ciotti & Pellegrini (1996) whereby flatter galaxies have a harder time in retaining their hot gas. Flattening indeed reduces the binding energy of the gas, which would cause SNe to drive global outflows in fast rotators, thus reducing their hot-gas density and X-ray luminosity. The finding that the few -deficient slow rotators also happen to be relatively flat would lend further support to this hypothesis, although care is needed as there are also flat slow rotators with perfectly in line with what is observed in the more standard and round slow rotators. We also note that if the intrinsic shape of a galaxy could determine the ability of a galaxy to retain its hot-gas halo, the degree of rotational support could further lower the efficiency with which the kinetic energy carried by the stellar-mass loss material is thermalised in the hot gas. Indeed, this efficiency appears to reach its peak in the dynamically supported slow rotators, where traces well .
Finally, using recent Herschel measurements for the diffuse dust emission of early-type galaxies and the kinematic information on the relative motions of stars and gas provided by our integral-field spectroscopic observations, we have also shown that fast rotators have a larger specific dust content compared to slow rotators, and that this is likely due to the fact that some fraction of the stellar-mass loss material is allowed to cool down in the more tenuous medium of these objects. Such an ability to recycle the stellar-mass loss links well with the fact that among massive early-type galaxies, only fast rotators have detected molecular gas and display signs of recent star-formation, something that cannot occur in slow rotators where both stellar-mass loss and accreted material quickly fizzles in the hot medium.
We wish to thank the anonymore referee, whose comments helped much in improving the paper. MS is indebted to Gary Mamon, Silvia Pellegrini, Nicola Brassington, Martin Hardcastle and Luca Cortese for useful discussion, and to both ESO and the Institut d’Astrophysique de Paris for their hospitality during much of the preparation of this paper. MS acknowledges support from an STFC Advanced Fellowship ST/F009186/1. MC acknowledges support from a Royal Society University Research Fellowship. This work was supported by the rolling grants ‘Astrophysics at Oxford’ PP/E001114/1 and ST/H002456/1 and visitors grants PPA/V/S/2002/00553, PP/E001564/1 and ST/H504862/1 from the UK Research Councils. RLD acknowledges travel and computer grants from Christ Church, Oxford and support from the Royal Society in the form of a Wolfson Merit Award 502011.K502/jd. SK acknowledges support from the Royal Society Joint Projects Grant JP0869822. RMcD is supported by the Gemini Observatory, which is operated by the Association of Universities for Research in Astronomy, Inc., on behalf of the international Gemini partnership of Argentina, Australia, Brazil, Canada, Chile, the United Kingdom, and the United States of America. TN and MBois acknowledge support from the DFG Cluster of Excellence ‘Origin and Structure of the Universe’. NS and TD acknowledge support from an STFC studentship. The authors acknowledge financial support from ESO. We acknowledge use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.
- Athey et al. (2002) Athey, A., Bregman, J., Temi, P., & Sauvage, M. 2002, ApJ, 571, 272
- Athey (2007) Athey, A. E. 2007, Ph. D. Thesis, Chapter 4, arXiv:0711.0395
- Bois et al. (2011) Bois M., et al., 2011, MNRAS, 416, 1654 (Paper VI)
- Boroson, Kim, & Fabbiano (2011) Boroson B., Kim D.-W., Fabbiano G., 2011, ApJ, 729, 12
- Brighenti & Mathews (1996) Brighenti F., Mathews W. G., 1996, ApJ, 470, 747
- Canizares, Fabbiano, & Trinchieri (1987) Canizares C. R., Fabbiano G., Trinchieri G., 1987, ApJ, 312, 503
- Cappellari et al. (2006) Cappellari, M., Bacon, R., Bureau, M., et al. 2006, MNRAS, 366, 1126
- Cappellari et al. (2007) Cappellari M., et al., 2007, MNRAS, 379, 418
- Cappellari (2008) Cappellari M., 2008, MNRAS, 390, 71
- Cappellari et al. (2011a) Cappellari M., et al., 2011a, MNRAS, 413, 813 (Paper I)
- Cappellari et al. (2011b) Cappellari M., et al., 2011b, MNRAS, 416, 1680 (Paper VII)
- Cappellari et al. (2012a) Cappellari M., et al., 2012a, Nature, 484, 485
- Cappellari et al. (2012b) Cappellari, M., Scott, N., Alatalo, K., et al. 2012b, arXiv:1208.3522 (Paper XV)
- Cappellari et al. (2012c) Cappellari, M., McDermid, R. M., Alatalo, K., et al. 2012b, arXiv:1208.3523 (Paper XX)
- Ciotti et al. (1991) Ciotti L., D’Ercole A., Pellegrini S., Renzini A., 1991, ApJ, 376, 380
- Ciotti & Pellegrini (1996) Ciotti L., Pellegrini S., 1996, MNRAS, 279, 240
- Ciotti & Ostriker (2001) Ciotti L., Ostriker J. P., 2001, ApJ, 551, 131
- Davies et al. (2001) Davies R. L., et al., 2001, ApJ, 548, L33
- Davis et al. (2011) Davis T. A., et al., 2011, MNRAS, 417, 882
- Dekel & Birnboim (2006) Dekel, A., & Birnboim, Y. 2006, MNRAS, 368, 2
- D’Ercole & Ciotti (1998) D’Ercole, A., & Ciotti, L. 1998, ApJ, 494, 535
- de Zeeuw et al. (2002) de Zeeuw P. T., et al., 2002, MNRAS, 329, 513
- Diehl & Statler (2007) Diehl S., Statler T. S., 2007, ApJ, 668, 150
- D’Onofrio et al. (1995) D’Onofrio M., Zaggia S. R., Longo G., Caon N., Capaccioli M., 1995, A&A, 296, 319
- Dressler et al. (1997) Dressler A., et al., 1997, ApJ, 490, 577
- Ellis & O’Sullivan (2006) Ellis S. C., O’Sullivan E., 2006, MNRAS, 367, 627
- Emsellem et al. (2007) Emsellem E., et al., 2007, MNRAS, 379, 401
- Emsellem et al. (2011) Emsellem E., et al., 2011, MNRAS, 414, 888 (Paper III)
- Eskridge, Fabbiano, & Kim (1995a) Eskridge P. B., Fabbiano G., Kim D.-W., 1995, ApJS, 97, 141
- Eskridge, Fabbiano, & Kim (1995b) Eskridge P. B., Fabbiano G., Kim D.-W., 1995, ApJ, 442, 523
- Faber & Jackson (1976) Faber S. M., Jackson R. E., 1976, ApJ, 204, 668
- Faber & Gallagher (1976) Faber S. M., Gallagher J. S., 1976, ApJ, 204, 365
- Fernández-Ontiveros et al. (2011) Fernández-Ontiveros J. A., López-Sanjuan C., Montes M., Prieto M. A., Acosta-Pulido J. A., 2011, MNRAS, 411, L21
- Ferrari et al. (2002) Ferrari F., Pastoriza M. G., Macchetto F. D., Bonatto C., Panagia N., Sparks W. B., 2002, A&A, 389, 355
- Forman et al. (1979) Forman W., Schwarz J., Jones C., Liller W., Fabian A. C., 1979, ApJ, 234, L27
- Forman, Jones, & Tucker (1985) Forman W., Jones C., Tucker W., 1985, ApJ, 293, 102
- Goudfrooij & de Jong (1995) Goudfrooij P., de Jong T., 1995, A&A, 298, 784
- Johansson et al. (2009) Johansson, P. H., Naab, T., & Ostriker, J. P. 2009, ApJ, 697, L38
- Kaviraj et al. (2007) Kaviraj S., et al., 2007, ApJS, 173, 619
- Khochfar & Ostriker (2008) Khochfar, S., & Ostriker, J. P. 2008, ApJ, 680, 54
- Khochfar et al. (2011) Khochfar S., et al., 2011, MNRAS, 417, 845 (Paper VIII)
- Kim & Fabbiano (2004) Kim D.-W., Fabbiano G., 2004, ApJ, 611, 846
- Krajnović et al. (2011) Krajnović D., et al., 2011, MNRAS, 414, 2923 (Paper II)
- Kundu & Whitmore (2001a) Kundu A., Whitmore B. C., 2001, AJ, 121, 2950
- Kundu & Whitmore (2001b) Kundu A., Whitmore B. C., 2001, AJ, 122, 1251
- Kuntschner et al. (2010) Kuntschner H., et al., 2010, MNRAS, 408, 97
- Liu (2011) Liu J., 2011, ApJS, 192, 10
- Marinacci et al. (2011) Marinacci F., Fraternali F., Nipoti C., Binney J., Ciotti L., Londrillo P., 2011, MNRAS, 415, 1534
- Mathews & Loewenstein (1986) Mathews, W. G., & Loewenstein, M. 1986, ApJ, 306, L7
- Mathews & Brighenti (2003) Mathews W. G., Brighenti F., 2003, ARA&A, 41, 191
- Matsushita (2001) Matsushita K., 2001, ApJ, 547, 693
- Mulchaey et al. (2003) Mulchaey J. S., Davis D. S., Mushotzky R. F., Burstein D., 2003, ApJS, 145, 39
- Mulchaey & Jeltema (2010) Mulchaey J. S., Jeltema, T. E., 2010, ApJ, 715, 1L
- Nipoti & Binney (2007) Nipoti C., Binney J., 2007, MNRAS, 382, 1481
- O’Sullivan, Forbes, & Ponman (2001) O’Sullivan E., Forbes D. A., Ponman T. J., 2001, MNRAS, 328, 461
- Parriott & Bregman (2008) Parriott, J. R., & Bregman, J. N. 2008, ApJ, 681, 1215
- Pellegrini, Held & Ciotti (1997) Pellegrini, S., Held, E. V., Ciotti, L., 1997, MNRAS, 288, 1
- Pellegrini & Ciotti (1998) Pellegrini, S., Ciotti, L., 1998, A&A, 333, 433
- Pellegrini (2011) Pellegrini, S. 2011, ApJ, 738, 57
- Pellegrini (2012) Pellegrini, S. 2012, Astrophysics and Space Science Library, 378, 21
- Pipino et al. (2005) Pipino A., Kawata D., Gibson B. K., Matteucci F., 2005, A&A, 434, 553
- Santos, Mendes de Oliveira, & Sodré (2007) Santos W. A., Mendes de Oliveira C., Sodré L., Jr., 2007, AJ, 134, 1551
- Sarazin, Irwin, & Bregman (2000) Sarazin C. L., Irwin J. A., Bregman J. N., 2000, ApJ, 544, L101
- Sarzi et al. (2006) Sarzi M., et al., 2006, MNRAS, 366, 1151
- Sarzi et al. (2007) Sarzi M., et al., 2007, NewAR, 51, 18
- Sarzi et al. (2010) Sarzi M., et al., 2010, MNRAS, 402, 2187
- Sauvage & Thuan (1992) Sauvage, M., & Thuan, Th. X., 1992, ApJ, 396, 69
- Shapiro et al. (2010) Shapiro K. L., et al., 2010, MNRAS, 402, 2140
- Silva, Kuntschner, & Lyubenova (2008) Silva D. R., Kuntschner H., Lyubenova M., 2008, ApJ, 674, 194
- Smith et al. (2012) Smith M. W. L., et al., 2012, ApJ, 748, 123
- Sparks, Macchetto, & Golombek (1989) Sparks W. B., Macchetto F., Golombek D., 1989, ApJ, 345, 153
- Skrutskie et al. (2006) Skrutskie M. F., et al., 2006, AJ, 131, 1163
- Statler et al. (2004) Statler T. S., Emsellem E., Peletier R. F., Bacon R., 2004, MNRAS, 353, 1
- Sun et al. (2007) Sun M., Jones C., Forman W., Vikhlinin A., Donahue M., Voit M., 2007, ApJ, 657, 197
- Temi, Brighenti, & Mathews (2007) Temi P., Brighenti F., Mathews W. G., 2007, ApJ, 660, 1215
- Thomas et al. (2005) Thomas D., Maraston C., Bender R., Mendes de Oliveira C., 2005, ApJ, 621, 673
- Thomas et al. (2010) Thomas D., Maraston C., Schawinski K., Sarzi M., Silk J., 2010, MNRAS, 404, 1775
- Trinchieri & Fabbiano (1985) Trinchieri G., Fabbiano G., 1985, ApJ, 296, 447
- van den Bosch et al. (2008) van den Bosch R. C. E., van de Ven G., Verolme E. K., Cappellari M., de Zeeuw P. T., 2008, MNRAS, 385, 647
- White & Sarazin (1991) White R. E., III, Sarazin C. L., 1991, ApJ, 367, 476
- White, Sarazin, & Kulkarni (2002) White R. E., III, Sarazin C. L., Kulkarni S. R., 2002, ApJ, 571, L23
- Young et al. (2011) Young L. M., et al., 2011, MNRAS, 414, 940
- Zhang, Gilfanov, & Bogdan (2012) Zhang Z., Gilfanov M., Bogdan A., 2012, arXiv, arXiv:1202.2331