Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~kim/pap/2015_Scaling.pdf Äàòà èçìåíåíèÿ: Thu Oct 15 18:58:27 2015 Äàòà èíäåêñèðîâàíèÿ: Sun Apr 10 06:47:49 2016 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: dust disk

The Astrophysical Journal, 812:127 (14pp), 2015 October 20
© 2015. The American Astronomical Society. All rights reserved.

doi:10.1088/0004-637X/812/2/127

X-RAY SCALING RELATIONS OF "CORE" AND "CORELESS" E AND S0 GALAXIES
Dong-Woo Kim and Giuseppina Fabbiano
Smithsonian Astrophysical Observatory, 60 Garden Street, Cambridge, MA 02138, USA Received 2015 April 3; accepted 2015 August 26; published 2015 October 15

ABSTRACT We have re-examined the two X-ray scaling relations of early-type galaxies (ETGs), LX,GAS­LK and LX,GAS­TGAS, using 61 ATLAS3D E and S0 galaxies observed with Chandra (including ROSAT results for a few X-ray bright galaxies with extended hot gas). With this sample, which doubles the number of ETGs available for study we confirm the strong, steep correlations reported by Boroson et al. Moreover, the larger sample allows us to investigate the effect of structural and dynamical properties of ETGs in these relations. Using the sub-sample of 11 "genuine" E galaxies with central surface brightness cores, slow stellar rotations and old stellar populations, we find that the scatter of the correlations is strongly reduced, yielding an extremely tight relation of the form LX,GAS TGAS4.5 0.3. The rms deviation is only 0.13 dex. For the gas-rich galaxies in this sample (LX,GAS > 1040 erg s-1), this relation is consistent with recent simulations for velocity dispersion supported E galaxies. However, the tight LX,GAS­TGAS relation of genuine E galaxies extends down into the LX 1038 erg s-1 range, where simulations predict the gas to be in outflow/ wind state, with resulting LX much lower than observed in our sample. The observed correlation may suggest the presence of small bound hot halos even in this low luminosity range. At the high luminosity end, the LX,GAS­TGAS correlation of core elliptical galaxies is similar to that found in samples of cD galaxies and groups, but shifted down toward relatively lower LX,GAS for a given TGAS. In particular cDs, central dominant galaxies sitting at the bottom of the potential well imposed by the group dark matter, have an order of magnitude higher LX,GAS than our sample core galaxies for the same LK and TGAS. We suggest that enhanced cooling in cDs, which have higher hot gas densities and lower entropies, could lower TGAS to the range observed in giant Es; this conclusion is supported by the presence of extended cold gas in several cDs. Instead, in the sub-sample of coreless ETGs--these galaxies also tend to show stellar rotation, a flattened galaxy figure and rejuvenation of the stellar population--LX,GAS and TGAS are not correlated. The LX-TGAS distribution of coreless ETGs is a scatter diagram clustered at LX,GAS < 1040 erg s-1, similar to that reported for the hot interstellar medium (ISM) of spiral galaxies, suggesting that both the energy input from star formation and the effect of galactic rotation and flattening may disrupt the hot ISM. Key words: galaxies: elliptical and lenticular, cD ­ X-rays: galaxies 1. INTRODUCTION Two families of early-type galaxies (ETGs) have been identified, based on the central surface brightness distribution: (1) core elliptical galaxies, which typically have large stellar masses (of which LK is a proxy), high stellar velocity dispersion (), round isophotes, uniformly old stellar populations with no sign of recent star formation (SF), and slowly rotating stellar kinematics; and (2) power-law (also called coreless) galaxies, which instead tend to be smaller, disky, rotating, and with some recent SF. Core -supported genuine Es tend to have a larger amount of hot X-ray emitting gas (see Pellegrini 2005; Kormendy et al. 2009; Lauer 2012; Sarzi et al. 2013). Kormendy et al. (2009) suggested that core Es are primarily formed by dry mergers. The hot interstellar medium (ISM) of core Es may provide the working surface necessary for active galactic nucleus (AGN) feedback by storing and smoothing episodic energy input, and by shielding against the accretion of fresh gas, thus impeding SF (Binney 2004; Nipoti & Binney 2007; Kormendy et al. 2009; Gabor & Dave' 2014). Coreless galaxies, instead, may be the product of gas-rich wet mergers, with ensuing SF (Kormendy et al. 2009). The X-ray properties of ETGs are important for constraining these scenarios. X-ray scaling relations have been widely used to investigate the origin and evolution of the hot ISM of ETGs (e.g., Fabbiano 1989; Mathews & Brighenti 2003; BKF; Kim & Fabbiano 2013; Civano et al. 2014). Recent numerical simulation studies (e.g., Choi et al. 2015; Negri et al. 2014) have attempted to reproduce the observed scaling relations, to 1 constrain the physical mechanisms that shape the hot ISM. While the correlation between optical and X-ray luminosities has been extensively investigated, most pre-Chandra studies have used the total X-ray luminosity (LX,Total) which is close to the X-ray emission of the hot ISM (LX,GAS) only for gas-rich X-ray luminous ETGs. Chandra observations have given us the ability to separate the different X-ray emission components of ETGs, extracting accurate measurements of LX, GAS (Boroson et al. 2011, hereafter BKF). This work has shown that the wellknown factor of 100 spread (e.g., Fabbiano 1989) in the previous ETG scaling relation, LX­LOptical, increases to a factor of 1000 when LX,GAS is used (BKF; Kim & Fabbiano 2013). With Chandra we have also been able to measure the temperatures of these gaseous components, and derive the LX, GAS­TGAS scaling relation for ETGs. This relation is tighter than the LX,GAS­LK (BKF), at least for gas-rich galaxies (LX, 40 -1 GAS > 10 erg s ), and may indicate virialization of the hot gas in the dark matter halos (Kim & Fabbiano 2013). The LX, GAS­TGAS relation of normal ETGs, however differs in both normalization and slope from that observed in more massive systems: cluster, groups and cD galaxies (BKF; O'Sullivan et al. 2003; Pratt et al. 2009). Because the relative importance of baryonic physics over the gravitational pull of dark matter may be the cause of these differences, comparing scaling relations may constrain the processes effective in different mass ranges. For gas-poor ETGs (LX,GAS < 1040 erg s-1), simulations suggest that the hot ISM is in the outflow/wind state, leading to larger TGAS and lower LX,GAS. Moreover, galaxy


The Astrophysical Journal, 812:127 (14pp), 2015 October 20

Kim & Fabbiano

rotation, flattened potential, embedded disks and recent SF may affect the evolution of this ISM, and change its physical properties by altering the potential field and energy budget (e.g., Pellegrini 2011; Negri et al. 2014). Recent detailed studies of the structural and dynamical properties of increasing larger ETG samples provide the baseline information to compare with the properties of the hot ISM. In this paper, we use the joint information of the ATLAS3D sample (Cappellari et al. 2011) and of Chandra, to take a new look at the X-ray scaling properties of 61 ETG. This is the largest sample that can be investigated to date with a uniform set of well-characterized properties. In Section 2, we describe our sample selection; in Section 3 we describe the Chandra observations and data reduction techniques; in Section 4, we present two X-ray scaling relations of ETGs, LX,GAS­LK and LX,GAS­TGAS, respectively. In Section 5, we discuss the implications of our results and in Section 6, we summarize our results. 2. GALAXY SAMPLE We use the ATLAS3D sample (Cappellari et al. 2011) for our study. ATLAS3D provides extensive multi-wavelength data and uniformly derived galaxy properties, including: stellar rotation (Emsellem et al. 2011), central radial profile (Krajnovic et al. 2013), cold gas mass (CO, HI), and dust presence (references in Table 1). We select for our study the 61 ATLAS3D ETGs, which were observed with Chandra for an exposure time longer than 10 ks. This joint ATLAS3D-- Chandra selection doubles the size of the ETG sample previously studied in X-rays (30 galaxies, BKF), while providing a better characterization of the galaxy properties. The new sample includes 5 cD galaxies; cD galaxies were not included in the BKF sample. In Table 1, we list the sample with information taken from a series of papers published by the ATLAS3D team (see the footnotes in Table 1). Comparing the ATLAS3D properties with results published previously in the literature, we find generally good agreement for morphological type, central stellar velocity dispersion, central radial profile characteristics (core versus coreless), and ellipticity. There are some discrepant measurements for a few galaxies, which we discuss below. These discrepancies do not change our results noticeably. Since stellar ages are not included in the ATLAS3D catalog, we used a range of measurements available in the literature (see Table 1). These stellar ages are primarily luminosity-weighted averages, using single stellar population models. Therefore, they may be inaccurate in some galaxies where a small population of recently formed stars could mimic an overall young stellar population when weighted by luminosity (e.g., Thomas et al. 2005). Additionally, we use the masses of cold atomic and molecular gas, and the presence of dust (from the ATLAS3D publications--see Table 1), as an indirect indication of recent SF. Of the ATLAS3D galaxy properties, the central structure (core versus coreless) is useful to separate two classes of ETGs (Kormendy et al. 2009; Lauer 2012). The morphological type could be misclassified due to hidden disks or dust obscuration. Classifying ETGs by other quantities (, age, ellipticity, rotation) is somewhat arbitrary because of no unambiguous dividing value. We take the classifications of core and coreless galaxies from Krajnovic et al. (2013) who fit the Nuker law (Lauer et al. 1995). As extensively discussed in Krajnovic et al. 2

(2013), different methods (e.g., core-SÈrsic model) may result in different classifications (see below for a few examples, see also Dullo & Graham 2013 for related issues). Although the SÈrsic core and Nuker core may not be structurally the same, there is no significant difference if one is only interested to separate cores from the rest of the profiles (see Appendix A in Krajnovic et al. 2013). We also apply stricter criteria to identify genuine ellipticals as core, passively evolving (an old stellar system with no sign of recent SF), -supported (slow or no rotation), morphologically elliptical galaxies (in Sections 4.2 and 5.1). Using 2D integral field spectroscopy, Emsellem et al. (2011) separated ETGs into two classes, fast and slow rotating galaxies. Instead of the conventional measure of stellar rotation (V/ ), they applied a new rotation parameter, R = ÀRV ß [R (V 2 + s 2 )0.5] and classified as fast and slow rotators with R greater or smaller than 0.310.5, respectively. Comparing the central structure and the stellar kinematics, Lauer (2012) suggested that the slow and fast rotating ETGs are essentially the same as the core and coreless galaxies, respectively, if a slightly different boundary (lR = 0.25) between slow and fast rotators is applied. In contrast, Krajnovic et al. (2013) suggested the presence of genuine populations of slowly rotating coreless galaxies and fast rotating core galaxies. In Figure 1, we plot our sample in the R­ plane. We identify core and coreless galaxies (as determined by Kranjnovic et al. 2013) with red filled squares and blue filled circles, respectively; galaxies with intermediate profile, with large blue open circles; galaxies with no measurement of the central radial profile with small blue open circles. The solid green curve is R = 0.31 0.5 (Emsellem et al. 2011; Krajnovic et al. 2013) and the dashed line is R = 0.25 (Lauer 2012). Whatever the classification, on average core galaxies are slower rotators than coreless galaxies. In particular, fast rotating core galaxies (red squares above the green curve in Figure 1), are the slowest among fast rotators. If we apply a cut at R = 0.2 (slightly lower than that used by Lauer) to our sample, we would identify 17 out of 19 core galaxies as slow rotators with only 2 exceptions and 27 out of 29 coreless galaxies (or 22 out of 23 without intermediate ones) as fast rotators again with only 2 exceptions (or 1 exception without intermediate ones). Given the location in the lR ­e plane, the 13 galaxies with unknown central structure (small open circles in Figure 1) are likely to be fast rotating coreless galaxies. The classification of core and coreless galaxies by Kranjnovic et al. (2013) is in most cases consistent with earlier results by Lauer et al. (2007), Hopkins et al. (2009) and Kormendy et al. (2009). However, there are a few notable exceptions, which are marked by galaxy name followed by the different classification in Figure 1. The misclassification, if confirmed, may explain some extreme outliers. NGC 524, the core galaxy with the highest R in our sample, was identified as intermediate by Lauer et al. (2007). NGC 4473 was identified as a power-law (coreless) galaxy by Kormendy et al. (2009-- see a rather long discussion on the definition of cores in their Section 9.2; see also Dullo & Graham 2013). NGC 5576, a slow rotating galaxy, was classified as intermediate by Kranjnovic et al. (2013), but as a core galaxy by both Lauer et al. (2007) and Hopkins et al. (2009). NGC 3607, with the lowest R among those not listed in Kranjnovic et al. (2013), is identified as a core by Lauer et al. (2007) and Hopkins et al. (2009).


The Astrophysical Journal, 812:127 (14pp), 2015 October 20 Table 1 Sample Early Type Galaxies and Their Properties Name (1) N0474 N0524 N0821 N1023 N1266 N2768 N2778 N3245 N3377 N3379 N3384 N3412 N3414 N3599 N3607 N3608 N3665 N3945 N3998 N4026 N4036 N4111 N4203 N4233 N4251 N4261 N4278 N4342 N4365 N4374 N4382 N4406 N4459 N4472 N4473 N4477 N4486 N4494 N4526 N4552 N4564 N4596 N4621 N4636 N4649 N4697 N4710 N5198 N5322 N5422 N5576 N5582 N5638 N5813 N5838 N5845 N5846 T (2) -2.0 -1.2 -4.8 -2.7 -2.1 -4.4 -4.8 -2.1 -4.8 -4.8 -2.7 -2.0 -2.0 -2.0 -3.1 -4.8 -2.1 -1.2 -2.1 -1.8 -2.6 -1.4 -2.7 -2.0 -1.9 -4.8 -4.8 -3.4 -4.8 -4.3 -1.3 -4.8 -1.4 -4.8 -4.7 -1.9 -4.3 -4.8 -1.9 -4.6 -4.8 -0.9 -4.8 -4.8 -4.6 -4.4 -0.9 -4.8 -4.8 -1.5 -4.8 -4.9 -4.8 -4.8 -2.6 -4.9 -4.7 d (Mpc) (3) 30.9 23.3 23.4 11.1 29.9 21.8 22.3 20.3 10.9 10.3 11.3 11.0 24.5 19.8 22.2 22.3 33.1 23.2 13.7 13.2 24.6 14.6 14.7 33.9 19.1 30.8 15.6 16.5 23.3 18.5 17.9 16.8 16.1 17.1 15.3 16.5 17.2 16.6 16.4 15.8 15.8 16.5 14.9 14.3 17.3 11.4 16.5 39.6 30.3 30.8 24.8 27.7 25.6 31.3 21.8 25.2 24.2 MK (mag) (4) -23.91 -24.71 -23.99 -24.01 -22.93 -24.71 -22.23 -23.69 -22.76 -23.80 -23.52 -22.55 -23.98 -22.22 -24.74 -23.65 -24.92 -24.31 -23.33 -23.03 -24.40 -23.27 -23.44 -23.88 -23.68 -25.18 -23.80 -22.07 -25.21 -25.12 -25.13 -25.04 -23.89 -25.78 -23.77 -23.75 -25.38 -24.11 -24.62 -24.29 -23.08 -23.63 -24.14 -24.36 -25.46 -23.93 -23.53 -24.10 -25.26 -23.69 -24.15 -23.28 -23.80 -25.09 -24.13 -22.92 -25.01
R

Kim & Fabbiano

(6) 0.192 0.034 0.392 0.363 0.193 0.472 0.224 0.442 0.503 0.104 0.065 0.441 0.194 0.080 0.185 0.190 0.216 0.090 0.170 0.556 0.555 0.582 0.154 0.277 0.508 0.222 0.103 0.442 0.254 0.147 0.202 0.211 0.148 0.172 0.421 0.135 0.037 0.173 0.361 0.047 0.560 0.254 0.365 0.094 0.156 0.447 0.699 0.146 0.307 0.604 0.306 0.320 0.091 0.170 0.361 0.264 0.062

Rot (7) F F F F F F F F F F F F S F F S F F F F F F F F F S F F S S F S F S F F S F F S F F F S F F F S S F S F F S F F S

(5) 0.213 0.325 0.273 0.391 0.638 0.253 0.572 0.626 0.522 0.157 0.407 0.406 0.073 0.282 0.209 0.043 0.410 0.524 0.454 0.548 0.685 0.619 0.305 0.564 0.584 0.085 0.178 0.528 0.088 0.024 0.163 0.052 0.438 0.077 0.229 0.221 0.021 0.212 0.453 0.049 0.619 0.297 0.291 0.036 0.127 0.322 0.652 0.061 0.073 0.600 0.102 0.564 0.265 0.071 0.521 0.404 0.032

o (km s-1) (8) 167.1 243.2 199.5 207.0 78.9 202.3 164.1 209.9 145.5 213.3 161.4 105.2 231.7 74.8 229.1 193.2 224.9 182.8 278.6 190.1 188.8 158.9 159.2 217.3 137.1 294.4 248.3 252.9 255.9 288.4 183.7 216.8 178.6 288.4 193.6 171.4 314.1 154.2 235.0 261.8 174.2 151.0 223.9 199.5 314.8 180.7 105.0 201.4 248.3 174.6 185.8 151.4 159.2 225.9 276.1 269.8 231.7

Core (9) 2 1 3 3 9 2 3 3 3 1 3 3 3 3 9 1 9 3 2 3 9 9 3 9 9 1 1 3 1 1 1 1 3 1 1 2 1 3 9 1 3 3 3 1 1 3 9 1 1 2 2 9 9 1 3 3 1

log(MH2) (M) (10) <7.68 7.97 <7.52 <6.79 9.28 7.64 <7.48 7.27 <6.96 <6.72 <7.11 <6.96 <7.19 7.36 8.42 <7.58 8.91 <7.50 <7.06 <6.99 8.13 7.22 7.39 <7.89 <7.11 <7.68 <7.45 <7.24 <7.62 <7.23 <7.39 <7.40 8.24 <7.25 <7.07 7.54 <7.17 <7.25 8.59 <7.28 <7.25 7.31 <7.13 <6.87 <7.44 <6.86 8.72 <7.89 <7.76 <7.78 <7.60 <7.67 <7.60 <7.69 <7.56 <7.50 <7.78

log(MHI) (M) (11) L L <6.91 9.29 L 7.81 <7.06 <7.00 <6.52 <6.49 7.25 <6.55 8.28 <7.03 <6.92 7.16 <7.43 8.85 8.45 8.50 8.41 8.81 9.15 L <6.97 L 8.80 L L <7.26 <6.97 8.00 <6.91 L <6.86 <6.95 L <6.84 L <6.87 <6.91 <7.13 <6.86 L <7.18 L 6.84 8.49 <7.34 7.87 9.65 L L L L L

log(MHIc) (M) (12) L L <6.53 6.84 L <6.61 <6.68 <6.61 <6.14 <6.11 <6.19 <6.17 <7.70 <6.64 <6.53 <6.53 <7.05 <6.73 7.42 <7.14 <6.80 6.94 7.03 L <6.58 L 6.06 L L <6.88 <6.59 <6.40 <6.53 L <6.47 <6.56 L <6.46 L <6.48 <6.53 <7.13 <6.48 L <7.19 L 6.71 <6.98 <6.96 7.43 L <6.88 L L L L L

Dust (13) N N N N F N N N N N N N N N D N D FBR N N F N N F N N N N N N N N D N N N N N D N N N N N N N D N N D N N N N N N N

Age (Gyr) (14) 7.1 12.2 8.9 4.7 L 10.0 6.4 5.8 3.6 10.0 3.2 1.9 3.6 1.2 3.6 8.0 2.0 L L L L L L L 1.9 16.3 12.0 L 5.9 12.8 1.6 L 1.9 9.6 4.0 11.7 19.6 6.7 1.6 12.4 5.9 L 15.8 13.5 14.1 8.3 L 15.1 2.4 L 2.5 17.0 9.5 16.6 11.2 8.9 14.2

Reference (15) 6 6 1 5 L 3 1 8 1 1 5 2 5 8 2 1 8 L L L L L L L 2 1 1 3 1 2 L 5 1 3 6 3 8 4 1 2 L 3 7 1 1 L 5 8 L 3 2 1 1 6 5 1

3


The Astrophysical Journal, 812:127 (14pp), 2015 October 20 Table 1 (Continued) Name (1) N5866 N6017 N6278 N7457 T (2) -1.3 -5.0 -1.9 -2.6 d (Mpc) (3) 14.9 29.0 42.9 12.9 MK (mag) (4) -24.00 -22.52 -24.19 -22.38
R

Kim & Fabbiano

(6) 0.566 0.455 0.409 0.470

Rot (7) F F F F

(5) 0.319 0.429 0.576 0.519

o (km s-1) (8) 160.3 109.9 217.8 74.8

Core (9) 9 9 3 3

log(MH2) (M) (10) 8.47 <7.73 <7.98 <6.96

log(MHI) (M) (11) 6.96 L <7.67 <6.61

log(MHIc) (M) (12) 6.67 L <7.28 <6.22

Dust (13) D D N N

Age (Gyr) (14) 1.8 L L 3.4

Reference (15) 2 L L 6

Notes. 1. Galaxy NGC name. 2. Morphological type from Cappellari et al. (2011). 3. Distance mainly from Tonry et al. (2001) and from Cappellari et al. (2011) if not listed in the former. 4. Absolute K-band mag from 2MASS. 5. R = ÀRV ß [R (V 2 + s 2 )0.5] measured at Re from Emsellem et al. (2011). 6. Ellipticity measured at Re from Emsellem et al. (2011). 7. F (fast) or S (slow) rotator from Emsellem, E., et al. (2011). 8. Central velocity dispersion in r < 1/8 Re from Cappellari, M. et al. (2013). 9. Central radial profile 1 (core), 2 (intermediate), 3 (cusp), 9 (unknown) from Krajnovic et al. (2013). 10. Mass of H2 molecular gas from Young, L. M. et al. (2011). 11. Mass of HI gas from Serra, P. et al. (2012). 12. Mass of central HI gas from Young, L. M. et al. (2014). 13. Dust features: D (dusty disk), F (dusty filament), B (blue nucleus) and BR (blue ring) from Krajnovi, D. et al. (2011). 14. Ages and references. References. (1) Thomas et al. (2005, ApJ, 621, 673). (2) Terlevich & Forbes (2002, MNRAS, 330, 547). (3) Howell (2005, AJ, 130, 2065). (4) Gallagher et al. (2008, ApJ, 685, 752). (5) McDermid et al. (2006, MNRAS, 373, 906). (6) Kuntschner et al. (2010, MNRAS, 408, 97). (7) Annibali et al. (2010, AA, 519, A40). (8) Denicolo et al. (2005, MNRAS, 356, 1440).

If we accept possible misclassifications, the core galaxies in our sample have lower R than coreless galaxies, N3607 having the highest R = 0.209 among core galaxies. The only exception would be NGC 3414, which both Lauer et al. (2007) and Krajnovic et al. (2013) classified as coreless. Given its round shape ( = 0.19), NGC 3414 may be affected by projection effects, as pointed out by Lauer (2012). 3. X-RAY DATA ANALYSIS X-ray temperatures and luminosities are listed in Table 2. We followed BKF and Kim & Fabbiano (2013), to measure LX, GAS and TGAS. We generated light curves to check for background flares and excluded these events (see BKF for more details). The background spectra were then extracted from source-free region within the same CCD. We also use sky background data (http://cxc.cfa.harvard.edu/ciao/threads/ acisbackground/), after rescaling them by matching the count rates at 9­12 keV (see http://cxc.harvard.edu/contrib/maxim/ acisbg/). The two methods give consistent results. In the least X-ray luminous galaxies, the X-ray luminosity of the hot gas may be lower than the integrated contribution of discrete sources, including low-mass X-ray binaries (LMXB) and even fainter X-ray sources, such as active binaries (AB) and cataclysmic variables (CV), see BKF. Therefore, careful determination and subtraction of all stellar contributions is critical to measure LX,GAS accurately. After subtracting the bright LMXBs detected in Chandra observations, we fitted the remaining 0.3­5 keV unresolved emission with a 4-component model, consisting of (1) thermal plasma APEC to model the hot

Figure 1. Galaxy rotation parameter R is plotted against ellipticity () as measured by Emsellem et al. (2011). The core and coreless galaxies (as determined by Krajnovic et al. 2013) are separately marked by red squares and blue circles, respectively. The galaxies with an intermediate profile are marked by large blue open circles. Those with no measurement of the central radial profile structure are marked by small blue open circles. The large red circles indicate 5 cDs. The solid green curve is R = 0.31 0.5 used by Emsellem et al. (2011) and Krajnovic et al. (2013) and the dashed line is R = 0.25 used by Lauer (2012).

4


The Astrophysical Journal, 812:127 (14pp), 2015 October 20 Table 2 X-Ray Properties of Sample Early Type Galaxies Name N0474 N0524 N0821 N1023 N1266 N2768 N2778 N3245 N3377 N3379 N3384 N3412 N3414 N3599 N3607 N3608 N3665 N3945 N3998 N4026 N4036 N4111 N4203 N4233 N4251 N4261 N4278 N4342 N4365 N4374 N4382 N4406 N4459 N4472 N4473 N4477 N4486 N4494 N4526 N4552 N4564 N4596 N4621 N4636 N4649 N4697 N4710 N5198 N5322 N5422 N5576 N5582 N5638 N5813 N5838 N5845 N5846 N5866 N6017 N6278 N7457 Obsid 07144 06778 multi multi 11578 09528 11777 02926 02934 multi 04692 04693 06779 09556 02073 02073 03222 06780 06781 06782 06783 01578 10535 06784 04695 09569 multi 04687 multi 00803 02016 00318 02927 00321 04688 09527 multi 02079 03925 02072 04008 02928 02068 00323 multi multi 09512 06786 06787 multi 11781 11361 11313 05907 06788 04009 00788 02879 11363 06789 04697 r () 60 60 30 60 30 60 30 90 30 90 60 30 30 60 240 90 60 30 30 30 30 30 60 30 60 60 90 60 120 L 90 L 60 L 60 90 L 120 30 90 30 60 90 L L 120 60 30 30 30 30 30 30 120 60 30 L 90 30 60 60 TGas(keV) 0.19 0.50 0.09 0.30 0.81 0.31 0.54 0.30 0.19 0.25 0.26 0.30 0.57 0.16 0.59 0.40 0.34 0.37 0.26 0.28 0.46 0.47 0.25 0.20 0.28 0.76 0.30 0.59 0.46 0.73 0.39 0.82 0.40 0.95 0.31 0.33 1.50 0.34 0.31 0.59 0.27 0.21 0.27 0.73 0.86 0.31 0.33 0.36 0.33 0.24 0.52 0.30 0.30 0.70 0.42 0.39 0.72 0.32 0.30 0.52 0.30 (0.00­9.00) (0.43­0.57) (0.00­9.00) (0.29­0.32) (0.71­1.02) (0.30­0.32) (0.19­0.99) (0.24­0.36) (0.00­0.26) (0.23­0.27) (0.00­0.58) (0.00­9.00) (0.37­0.74) (0.00­0.25) (0.52­0.66) (0.34­0.49) (0.31­0.41) (0.29­0.53) (0.22­0.32) (0.17­1.06) (0.32­0.61) (0.41­0.51) (0.18­0.33) fixed (0.05­0.83) (0.75­0.77) (0.29­0.31) (0.55­0.64) (0.44­0.49) (0.73­0.74) (0.37­0.40) (0.81­0.83) (0.34­0.55) (0.94­0.96) (0.28­0.35) (0.32­0.34) (1.50­1.50) (0.25­0.86) (0.29­0.32) (0.59­0.60) (0.00­0.72) (0.19­0.24) (0.21­0.35) (0.72­0.73) (0.86­0.86) (0.31­0.32) (0.30­0.39) (0.26­0.62) (0.30­0.38) (0.02­0.42) (0.30­0.72) fixed fixed (0.69­0.70) (0.35­0.50) (0.25­0.66) (0.72­0.73) (0.30­0.34) fixed (0.30­0.76) fixed LX,Tot 0.831 1.612 0.883 0.615 2.256 2.612 0.296 1.288 0.306 0.864 0.617 0.137 4.477 0.588 2.953 1.147 3.044 1.873 32.555 0.215 2.998 1.108 6.019 4.383 0.554 19.540 3.956 0.427 3.787 9.001 2.806 130.057 1.318 28.180 0.666 1.556 908.124 1.546 1.525 4.825 0.285 0.784 1.274 32.989 21.295 1.160 0.279 1.561 1.576 0.594 0.482 0.298 0.410 74.016 1.526 0.424 52.722 0.891 0.223 4.483 0.114 L
X,GAS

Kim & Fabbiano

(1040 erg s-1)a (0.000­0.332) (1.027­1.325) (0.015­0.037) (0.057­0.066) (0.459­0.632) (0.716­0.790) (0.010­0.049) (0.211­0.349) (0.007­0.013) (0.035­0.048) (0.004­0.033) (0.000­0.056) (0.106­0.207) (0.014­0.068) (1.416­1.784) (0.348­0.480) (1.036­1.482) (0.228­0.371) (0.761­1.236) (0.010­0.039) (0.369­0.513) (0.446­0.515) (0.105­0.210) (0.003­0.107) (0.015­0.090) (6.735­6.952) (0.232­0.258) (0.151­0.178) (0.451­0.490) (5.474­7.826)b (0.944­1.014) (126.631­128.915)b (0.204­0.292) (22.244­26.202)b (0.105­0.147) (0.887­0.964) (904.186­906.812)b (0.053­0.216) (0.262­0.299) (2.067­2.133) (0.024­0.053) (0.223­0.303) (0.036­0.090) (31.121­32.367)b (16.675­19.755)b (0.180­0.199) (0.098­0.133) (0.160­0.396) (0.567­0.761) (0.013­0.070) (0.032­0.080) (0.011­0.068) (0.000­0.111) (70.313­70.980) (0.623­0.844) (0.033­0.084) (49.407­51.618)b (0.231­0.272) (0.000­0.035) (0.151­0.365) (0.002­0.034)

0.166 1.174 0.025 0.061 0.522 0.753 0.025 0.280 0.010 0.041 0.018 0.025 0.157 0.040 1.600 0.414 1.260 0.300 0.999 0.023 0.428 0.480 0.208 0.018 0.048 6.844 0.245 0.164 0.470 6.650 0.979 127.773 0.251 24.223 0.126 0.926 905.499 0.130 0.281 2.154 0.038 0.263 0.063 31.744 18.215 0.189 0.115 0.281 0.665 0.039 0.055 0.029 0.000 70.654 0.735 0.056 50.513 0.251 0.000 0.246 0.012

Notes. a LX,GAS in 0.3­8 keV. b Used ROSAT results from O'Sullivan et al. (2001).

5


The Astrophysical Journal, 812:127 (14pp), 2015 October 20

Kim & Fabbiano

Figure 2. Left--LX,GAS plotted against LK for early type galaxies selected from the ATLAS3D sample. The cyan and green lines indicate the BKF linear relations for LX,LMXB and LX,AB+CV, respectively, and the red line (LX,GAS LK 3) is the best fit relation for the entire sample. The five galaxies at the highest LX marked by a big circle are cD galaxies. Right--LX,GAS plotted against TGAS for early type galaxies selected from the ATLAS3D sample, with cD galaxies marked by a big circle. The 5.4 red dashed line is the best fit for the entire sample (LX,GAS TGAS ). The thick green line is the relation determined among group/cluster dominant galaxies by O'Sullivan et al. (2003).

gas with metal abundance in the APEC model for the hot gas fixed at solar (Grevesse & Sauval 1998); (2) 7 keV thermal Bremsstrahlung to model the undetected LMXB population; (3) additional APEC component and (4) power-law to model the AB+CV population (see BKF and Kim 2012). In all cases, we find that the spectral determination of the undetected LMXB component is consistent within errors with the estimate derived from extrapolating the X-ray luminosity function of LMXBs (Kim & Fabbiano 2010; see also Section 3.4.2 in BKF). To model the AB+CV contribution, we used the spectral parameters determined from the Chandra spectra of M31 and M32, where all the LMXBs can be detected and removed (see appendix in BKF). The normalizations were scaled, based on the K-band luminosity within the region of interest. For extremely gas-poor galaxies where LX, GAS is lower than that of the soft (APEC) component of AB+CV, we considered the error of the soft component of the AB+CV emission and added it in quadrature to the statistical error. The error of the hard component (power-law) of AB+CV is negligible. When the gas temperature is not well-constrained, we fixed TGAS at 0.3 keV (a typical value for gas-poor galaxies) to calculate LX, GAS. We only used these galaxies in the LX,GAS­LK correlation study, but not in the LX,GAS­TGAS correlation. For some well-studied gas rich galaxies in our sample, the emission of the hot gaseous halo extends radially over more than 4 arcmin, and therefore falls outside the boundaries of a Chandra ACIS CCD chip. In these cases (marked in Table 2), we used LX,Total from the ROSAT measurements by O'Sullivan et al. (2001) and corrected it for our energy band (0.3­8 keV) and the distances in Table 1. We also subtracted the LMXB contribution by applying the BKF scaling relation LX,LMXB/ LK= 1029 erg s-1 L K -1. Because of the scatter in this relation, LX, LMXB may vary by 50% (BKF). This scatter could cause an error up to 15% in LX,GAS. We added this error in the estimates shown in Table 2. 6

For galaxies with extended hot gas (mostly those for which we use ROSAT data, see Table 2), TGAS may vary as a function of distance from the galaxy center, e.g., TGAS increases from 0.6 to 0.8 at the central region to 1­1.2 keV at the outskirts (e.g., Diehl & Statler 2008; Nagino & Matsushita 2009). In these cases, we use an emission-weighted average temperature, obtained by fitting the spectra extracted in multiple annuli. For galaxies with extended gas beyond the ACIS chips, we compared our measurements with results in the literature. For example, Nagino & Matsushita (2009) recently analyzed Chandra and XMM-Newton observations for a sample of gas rich elliptical galaxies and reported their gas luminosities and temperatures in concentric annuli. We estimated the emissionweighted average temperature and confirmed that our results are consistent with those of Nagino & Matsushita within the uncertainties. For the overlapping ETGs, our results are comparable to those in BKF, although should be considered slightly improved because of the improved atomic data in AtomDB (www.atomdb.org) used in the present work, and of the additional data used, when available in the Chandra archive. 4. THE SCALING RELATIONS Figure 2 (Left) shows the LX, GAS­LK scatter plot for our sample of 61 ETGs. As in BKF, we find a good correlation between LX,GAS and LK. The best fit line, LX, GAS ~ LK 3.0 0.4, is marked by the red dashed line. Figure 2 (Right) shows the LX, GAS­TGAS scatter plot. As in BKF, we find a positive correlation between LX,GAS and TGAS, with best fit relation is LX,GAS TGAS5.4 0.5. The distribution of the five cD-type galaxies in this plot is consistent with the LX,GAS­TGAS of group/cluster dominant galaxies (O'Sullivan et al. 2003; the green line) which is shifted upward toward higher luminosities for a given gas temperature relative to non-cD E galaxies (see Section 5.2 for more discussions).


The Astrophysical Journal, 812:127 (14pp), 2015 October 20

Kim & Fabbiano

Figure 3. Same as Figure 1(a), but with two classes of ETGs marked with different colors, based on (a) their morphological types, (b) central stellar velocity dispersion, (c) central radial profile, (d) mean stellar age, (e) flattening, and (f) stellar rotation.

7


The Astrophysical Journal, 812:127 (14pp), 2015 October 20

Kim & Fabbiano

Figure 4. Same as Figure 2(b), but with two classes of ETGs marked with different colors (same as Figure 3). The red dashed line is the best fit relation among the core galaxies (LX,GAS TGAS4.4 ).

8


The Astrophysical Journal, 812:127 (14pp), 2015 October 20 Table 3 Best Fit LX, GAS­LK Relation Sample ( ( ( ( ( 1) 2) 3) 4) 5) Full Core Core-cDs Noncore Coreless
X,GAS

Kim & Fabbiano

N 59 19 14 40 23

A 2.98 4.27 2.84 1.83 1.60
11

Err 0.36 0.72 0.43 0.22 0.28

B -0.25 -0.41 -0.52 -0.54 -0.64

Err 0.11 0.21 0.17 0.09 0.12

p-values Pearson Spearman 3 â 10-10 0.002 0.0007 0.0006 0.008 9 â 10-11 0.004 0.0004 0.0008 0.01

rms Deviation in log (LX,GAS) 0.79 0.91 0.50 0.55 0.47

log(L

/1040 erg s-1) = A log(LK/10

L

K

)+B

Notes. (1) The entire sample. (2) Core galaxies only (1 in Table 1 column 9). (3) Same as (2) but also excluding 5 cD galaxies. (4) All but core galaxies (2, 3, and 9 in Table 1 column 9). (5) Only coreless galaxies (3 in Table 1 column 9). Table 4 ­TGAS Relation Err 0.43 0.11 0.06 0.05 L L 3 2 2 1 p-values Pearson Spearman â 10-11 â 10-8 â 10-8 â 10-8 0.37 0.94 1 4 3 4 â 10- â 10- â 10- â 10- 0.34 0.99
8 7 7 9

Best Fit L Sample ( ( ( ( ( ( 1) 2) 3) 4) 5) 6) Full Core Core-cD Core-cD-SF Noncore Coreless
X,GAS

X,GAS

N 49 19 14 11 30 16

A 5.39 6.16 4.36 4.57 L L

Err 0.60 0.70 0.30 0.26 L L

B 0.16 0.34 0.14 0.07 L L

rms Deviation in log(LX,GAS) 0.72 0.47 0.20 0.13 L L

log(L

/1040 erg s-1) = A log(kTGAS/0.5 keV) + B

Notes. (1) The entire sample. (2) Core galaxies only (1 in Table 1 column 9). (3) Same as (2) but also excluding 5 cD galaxies. (4) Same as (3) but also excluding 3 possibly star forming galaxies. (5) All but core galaxies (2, 3, and 9 in Table 1 column 9). (6) Only coreless galaxies (3 in Table 1 column 9).

Figures 3 and 4 show the LX,GAS­LK and LX,GAS­TGAS scatter plots, for different classes of ETGs, based on the properties of the ATLAS3D sample listed in Table 1: (a) morphological types (E versus S0), (b) central stellar velocity dispersion (high versus low ), (c) central radial profile (core versus coreless), (d) mean stellar age (old versus young), (e) flattening (round versus flat), and (f) stellar kinematics (slow versus fast rotators). As noticed by Kormendy et al. (2009; see also Lauer 2012), core galaxies (red points in Figures 3(a)­(f)), are characterized by large LK, high stellar velocity dispersion (), round isophotes, no sign of recent SF and slowly rotating stellar kinematics. These galaxies also tend to have larger LX, GAS (see also, Bender et al. 1989; Fabbiano 1989; Eskridge et al. 1995). Coreless galaxies (blue points) are characterized by the opposite properties. While some core (red) galaxies can be hot gas poor, with LX,GAS as low as a few 1038 erg s-1, coreless galaxies are all hot gas poor (as previously reported e.g., Pellegrini 1999). Figure 4 shows that the LX,GAS­TGAS relation is very tight for core galaxies (red points), but there is no correlation for coreless galaxies. To determine the best-fit relations, we have applied a bisector linear regression method and estimated the corresponding error by bootstrap resampling (http://home.strw. leidenuniv.nl/~sifon/pycorner/bces). We also used the 9

Pearson and Spearman correlation tests from the scipy statistics package (http://www.scipy.org) to estimate the p-value for the null hypothesis. The results are summarized in Tables 3 and 4 for the LX,GAS­LK and LX,GAS­TGAS, respectively. In both tables, we list the results for the full sample, and subsamples with different central structure. 4.1. L
X,GAS­LK

For the entire sample, LX,GAS LK 3.0 0.4. The best fit relation becomes steeper for core galaxies, when cDs are included, with LX,GAS LK 4.3 0.7; it is otherwise consistent with that of the full sample. The relation is significantly flatter for coreless galaxies with LX,GAS LK1.6 0.3. The p-values determined by Pearson and Spearman tests for the full sample are very low (10-10), indicating a strong correlation. The correlations are less strong, but still significant, in the subsamples, particularly in the coreless sample (p-value 0.01). The scatter of the correlation is large in all cases, as indicated by the rms deviations reported in Table 3: 0.8 dex in LX,GAS for a given LK, for the entire sample and 0.5 dex for the core (without cDs) and coreless sub-samples. For example, at LK = 1011 LK , which is close to the characteristic luminosity (L*) of the galaxy luminosity function, (e.g., Cole et al. 2001),


The Astrophysical Journal, 812:127 (14pp), 2015 October 20

Kim & Fabbiano

NGC 4621 and NGC 4636 have similar K-band luminosities (1.4 and 1.2 â1011 LK , respectively), but their LX,GAS differ by a factor of 500. At the higher LK end, NGC 5322 and NGC 4486 have similar LK (3 â 1011 LK ) but their LX,GAS differ by a factor of 1400. Note that NGC 4636 and NGC 4486 are cD galaxies. Excluding these cDs, the spread in LX,GAS is somewhat reduced to a factor of 30 in LX,GAS for a given LK in the entire LK range seen in Figure 2(a). 4.2. LX,GAS­TGAS For the entire sample, LX,GAS and TGAS are well correlated with p-value <10-8 (Table 4). The best fit relation is LX, GAS TGAS5.4 0.6, slightly steeper than that of BKF (LX,GAS TGAS4.5). This is mainly because of the cD galaxies included in our sample. The relation is particularly tight (LX,GAS TGAS4.4 0.3) for the normal (non-cD) core galaxies, with a small normalization error and only 0.2 dex rms. This tight relation holds for a range of kTGAS = 0.3­1 keV and LX, 38 41 -1 GAS = a few â10 --several â10 erg s . We further checked the core galaxy sample for signatures of recent SF. NGC 524 is classified as S0 by Lauer et al. (2007) and Kormendy & Ho (2013) as well as by RSA and RC3. Furthermore molecular gas (M(H2) = 108 Me) was detected in this galaxy by the Atlas3D team (see Table 1). NGC 4382 is classified as S0 by Lauer et al. (2007), RSA and RC3, but as E by Hopkins et al. (2009) and Kormendy & Ho (2013). No atomic and molecular gas was detected in this galaxy. However, its average stellar age is quite young, 1.6 Gyr (see Table 1). NGC 5322 is unanimously classified as E, but its stellar age is 2.4 Gyr (see Table 1). Excluding these three galaxies, all the remaining core galaxies are classified as E, their stellar ages are old and no cold gas and dust are detected. With NGC 524 excluded, this sample consists of only slowly rotating galaxies. We can consider them as genuine, passively evolving (no recent SF), -supported (slow or no rotation), morphologically elliptical galaxies. Table 4 shows an even tighter LX,GAS­TGAS correlation for this genuine elliptical sample, with rms deviation of only 0.13 dex. In contrast, no correlation exists for coreless galaxies (pvalue 1). Given that coreless ETGs are generally hot gas poor, their errors (particularly in TGAS) are large. To quantitatively test the possibility that a tight correlation may be hidden because of the large errors, we ran simulations by redistributing points in the LX,GAS­TGAS plane, assuming a Gaussian distribution with equal to the error of each point. In 100 simulations, we found one (or zero) incidence where the pvalue is less than the conventionally used limit of 0.05 (or 0.01), indicating that it is unlikely that the observational errors could hide a real correlation. 5. DISCUSSION Using the well-studied ATLAS3D sample of 61 ETGs, we have rebuilt the scaling relations, LX,GAS­LK and LX,GAS­TGAS, and investigated their behavior in the subsamples of core and coreless galaxies. For core E galaxies, we find that the LX,GAS­ TGAS relation is significantly tighter than the LX,GAS­LK one. In particular, when we eliminate galaxies with possible relatively recent SF, yielding a sample of genuine old core ellipticals, we find the tightest LX,GAS­TGAS correlation. For coreless galaxies, instead, we find a weak LX, GAS­LK correlation, while 10

LX,GAS­TGAS are not correlated. Below we discuss the implications of these results. 5.1. The LX,GAS­TGas Correlation of Genuine Core Elliptical Galaxies As shown in Section 4.2, genuine, passively evolving (no recent SF), -supported (slow or no rotation) core non-cD, morphologically elliptical galaxies present the strongest correlation between gas luminosity and temperature. This is tightest among all the X-ray scaling relations ever reported for ETGs (e.g., Bender et al. 1989; Fabbiano 1989; Eskridge et al. 1995; Mathews & Brighenti 2003; BKF). Both best-fit slope and normalization are well determined, and the rms deviation is only 0.13 dex:
log
10 L X, GAS 40 erg s-1

= (4.6 0.3) ´ log

kTGAS 0.5 keV

+ (0.07 0.05) .

The tight correlation indicates that the ability to retain hot gas (LX,GAS), which is a function of the potential depth, and the balance (TGas) between heating and cooling are closely regulated. We can qualitatively understand this correlation as a consequence of virialized gaseous halos in the dark matter potentials of these galaxies (e.g., Mathews et al. 2006; Kim & Fabbiano 2013). Larger galaxies not only retain larger amounts of hot ISM but also add more energy to the ISM from stellar and AGN feedback (e.g., Mathews & Brighenti 2003; Pellegrini 2011; Pellegrini et al. 2012). This relation is tight and steep. At the low end, we have NGC 3399 with kTGAS = 0.3 keV and LX,GAS = 4 â 1038 erg s-1, while at the high end, NGC 4472 and NGC 4649 with LX,GAS = 2 â 1041 erg s-1 and kTGAS = 0.9 keV. In Figure 5 (Left), we show the LX,GAS­TGAS scatter plot for the genuine E sample. In Figure 5 (Right) we show the LX, GAS­ TGAS relation recently determined by Negri et al. (2014) who performed high resolution 2D hydrodynamical simulations separately for fully -supported and rotation-supported galaxies. The similarity between the observation of genuine Es (Figure 5 Left) and the simulation of -supported Es (red in Figure 5 Right) is remarkable for hot-gas rich Es with LX, 40 -1 GAS > 10 erg s . In particular, the slope is well matched. The normalization is offset in that LX,GAS is slightly higher for a given TGAS in the simulations than in the observations, but still within twice the rms deviation. The difference may be caused by the uncertainties in various parameters involved and other effects not considered in the simulations (e.g., AGN feedback, magnetic pressure and turbulence etc.). If AGN feedback is a sporadic event, our result (the tight relation in a wide range of LX,GAS) may suggest that the AGN feedback is not critical among the genuine Es, also shown in the above simulation without AGN feedback. Alternatively, AGN feedback could work continuously in a radio mode by smoothing episodic energy input inside the hot gas rich normal Es (Kormendy et al. 2009) and may also scale with the halo mass as suggested by Booth & Schaye (2010) and Bogdan & Goulding (2015). What is surprising, in comparison with the simulation results, is that the observed tight LX,GAS­TGAS relation holds for the entire sample of genuine Es, extending down to X-ray luminosities in the range of a few 1038 erg s-1. The simulations, instead, predict higher gas temperatures at low LX, both for -supported and rotation-supported galaxies (see


The Astrophysical Journal, 812:127 (14pp), 2015 October 20

Kim & Fabbiano

Figure 5. LX,GAS vs. TGAS for the genuine E sample (left) from observations and (right) from simulations (Negri et al. 2014). The diagonal lines are the best fits (solid line) with rms deviations (dashed lines).

Figure 5(b), where -supported Es are not well represented, but the range of gas temperatures is a robust prediction of Negri et al.). In fact, the ISM of gas-poor Es is likely to be in the outflow/wind state, and then tends to be at a temperature higher than extrapolated from the LX,GAS­TGAS relation valid for inflows (the typical flow phase of gas-rich Es; e.g., Pellegrini 2011; Negri et al. 2014). The presence of a tight correlation extending to these lower X-ray luminosities suggest that the hot ISM may be retained in equilibrium in these galaxies and may require a lower rate of the energy input (e.g., lower SNe rates, reduced AGN cycles). We note that we cannot directly compare the simulations of isotropic rotators with the observations of coreless galaxies (or more specifically fast rotators), because the physical state of their hot ISM may be affected by multiple factors (see Section 5.3). However, the upper reaches of LX, GAS for this subsample are consistent with the distribution of the isotropic rotator models (see Figure 4). 5.2. Giant Ellipticals versus cD Galaxies In our sample, we have five cD-type galaxies, NGC 4406, NGC 4486, NGC 4636, NGC 5813, and NGC 5846. We have found that they have an order of magnitude higher LX,GAS for a given LK or TGAS (Figure 2), compared with normal giant E galaxies, and follow the same LX,GAS­TGAS relation of similar group/cluster dominant galaxies by O'Sullivan et al. (2003). This difference may be related to the associations of cDs with the larger extended group or cluster dark matter halos. NGC 4486 (M87) is in the center of the Virgo cluster. NGC 4636, NGC 5846 and NGC 5813 are dominant galaxies in groups with 10­20 known members in Nearby Galaxies Catalog (Tully 1998) and in the 2MASS group catalog (Crook et al. 2007) NGC 4406 (M86) is not identified as a group dominant galaxy in both catalogs, However, it is the biggest among several nearby galaxies at the similar 3D distance, including NGC 4374, NGC 4388, NGC 4438, and NGC 4458 and also at the center of a clump of dEs with negative velocities (M86 velocity = -250 km s-1), forming a second massive sub11

clumps outside a similar clump centered on M87 (e.g., Binggeli et al. 1993). Therefore, it is likely that M86 is sitting at the bottom of a group potential well. Alternatively, if the X-ray emission from the extended plum on the NW side of the galaxy center (e.g., Randall et al. 2008) is excluded, the X-ray luminosity within the main galaxy body (r < 3) decreases by a factor of 4, while TGAS (0.8 keV) remains similar (e.g., Rangarajan et al. 1995). Then its LX,GAS and TGAS would be close to those of normal giant Es. The gas luminosity (or gas mass), being larger in the central dominant galaxies in groups and clusters, does reflect the total mass. This is consistent with the tight correlation between the total mass and LX, GAS found by Kim & Fabbiano (2013). Qualitatively, the larger amount of dark matter in cDs may work in two ways to increase LX,GAS: (1) by increasing the ability to hold the hot ISM in the deeper potential well, and (2) by adding external pressure from ICM/IGM to confine the hot ISM. The former would also increase the gas temperature (e.g., Negri et al. 2014) if cooling is not enhanced (see below), while the latter could increase LX,GAS without significantly affecting TGAS. To better understand the difference between cDs and giant Es, we increase the sample of high LX galaxies with five additional well-studied hot-gas-rich elliptical galaxies not included in the ATLAS3D sample (see Figure 6). Their properties are summarized in Table 5. The X-ray properties are extracted from the literature (see references in the table). TGAS is estimated as a luminosity-weighted average temperature in the same way as in our data analysis described in Section 3. Three (NGC 507, NGC 5044, NGC 7619) of these galaxies are cD-type galaxies in groups with 10­30 members (Crook et al. 2007) and are consistent with other cDs of the ATLAS3D sample, but two (NGC 1399 and NGC 1407) follow the relation of giant core E galaxies; of these only NGC 1399 has been classified as a cD (e.g., Schombert 1986) in the center of the Fornax cluster. Werner et al. (2014) recently reported that the entropy profiles of the hot halos of giant E and cD galaxies show a dichotomy. The galaxies displaying extended cold gas as seen


The Astrophysical Journal, 812:127 (14pp), 2015 October 20

Kim & Fabbiano

Figure 6. Left--same as Figure 3(c) but with 5 more well-studied galaxies (stars with galaxy names) taken from the literature (see Table 5). Right--high luminosity cD galaxies with extended [C II] emission are marked by big green squares, while galaxies (mostly giant Es) lacking this emission are identified by large red squares. Table 5 X-Ray Properties of Other Galaxies Name N0507 N1399 N1407 N5044 N7619 d (Mpc) 69.90 19.95 28.84 31.19 52.97 log(LK) 11.69 11.40 11.57 11.23 11.56 log(L
X,GAS)

TGAS 1.03 1.21 0.77 1.00 0.81

Reference O03 NM09 F06 NM09 O03

42.59 41.53 41.00 42.42 41.89

References. O03--O'Sullivan et al. (2003), F06--Fukazawa et al. (2006), NM09--Nagino & Matsushita (2009).

in [C II] emission lines have lower entropies beyond r > 1 kpc than cold gas poor galaxies. It is interesting to note that galaxies with lower entropies are mostly cDs while those with higher entropies are normal core Es in our scaling relations. Three (NGC 4636, NGC 5813, NGC 5846) of the five galaxies with extended [C II] emission in Werner et al. (2014) are in our sample and they are all cDs with excess LX,GAS for their LK and TGAS. Of the remaining two galaxies, NGC 5044 follows the same trend as the other cDs (see Figure 6, Right) while NGC 6868 does not (it was recognized as a different case with a rotating disk by Werner et al. 2014, see their Section 4). On the other hand, three (NGC 4472, NGC 4649, NGC 4261) of the five galaxies with no [C II] emission in Werner et al. (2014) are in our sample and they are normal core Es. We also plot the renaming two galaxies in our scaling relation. NGC 1407, a core giant E, follows the same trend of normal core E as expected. However, NGC 1399, a well-known central dominant galaxy (cD) in the Fornax cluster, also behaves like a normal core E (see below). Because the gas temperatures vary only within a small range (0.8­1 keV), the entropy (kT/n e 2 3) dichotomy (at r > 1 kpc) must be caused by the gas density (ne), i.e., cDs with lower entropies having higher densities (hence higher LX,GAS) than normal core E galaxies, as expected with an additional confinement by ICM/IGM. Werner et al. (2014) further suggested that cooling from the hot phase mainly produced 12

cold gas in galaxies with extended [C II] emission. This can be understood because cDs have a higher density where the cooling rate increases with n e 2 and also the thermal stability parameter (T/n e 2 ) decreases (the gas becomes thermally unstable--see Figure 10 in Werner et al. 2014) The enhanced cooling in higher density (lower entropy) gas in cDs may explain why their gas temperature is almost the same as that of giant Es, while their luminosity is higher by an order of magnitude. A quantitative study is warranted to better understand these observed differences between giant Es and cDs. NGC 1399 is the one exception. While its location in the LX, GAS­TGAS plane is consistent with having no extended cold gas (as other core Es), it is a well-known cD. It may be related to an unusual distribution of DM, in the scale comparable to the hot gas extent. The sample of Nagino & Matsushita (2009) includes six hot gas rich galaxies, shown in our Figure 6: three (NGC 4636, NGC 5044, NGC 5846) with extended cold gas and three (NGC 1399, NGC 4472, NGC 4649) without it. While the mass-to-light ratio (M/LK) is similar in the inner region (<0.5 re), this ratio increases more rapidly with increasing radius in those with extended cold gas where M/ LK(<6re) is higher by a factor of 8­16 than M/LK(<0.5 re), while it only increases by a factor of 4­6 in the other three galaxies (including NGC 1399), indicating that not only the total amount of DM, but also how DM is distributed, is important to determine the amount of hot gas properties and its cooling. Giant ellipticals and central dominant cDs particularly in low-mass groups are not always clearly distinguished. The upward shift in the X-ray scaling relations (higher LX,GAS for a given LK and TGAS) can provide an excellent empirical method to correctly identify them, i.e., the presence or absence of the extended DM associated with IGM/ICM. 5.3. X-Ray Scaling Relations of Coreless ETGs As discussed above, the X-ray scaling relations for coreless galaxies present a large scatter. LX,GAS and LK are weakly correlated among coreless galaxies, indicating that the small


The Astrophysical Journal, 812:127 (14pp), 2015 October 20

Kim & Fabbiano

Figure 7. LX,GAS vs. T

GAS

(a) for a ETG sample without core ellipticals (same as Figure 3(c) without red points) (b) for a spiral sample (from Li & Wang 2013).

amount of hot gas near the central region is related to the potential depth determined mostly by stars and to the gas production rate again determined by the stellar mass loss. However, LX,GAS and TGAS are not correlated at all. That is in contrast with core galaxies, which have a strong correlation. The lack of a LX,GAS­TGAS correlation among coreless galaxies may be understood due to the presence of several factors that may affect the retention and temperature of the hot gas. These galaxies are fast rotators with flattened galaxy figures. Both effects may impact the accumulation of hot halos, as shown by the simulations of Negri et al. (2014). They also contain embedded disks with recent SF, which may increase the effect of the stellar feedback, increasing the gas temperature (Negri et al. 2015). Each effect can change the hot gas properties by altering the potential field and energy budget. A similar situation may happen in the ISM of spiral galaxies. In Figure 7 we compare the LX,GAS and TGAS relation among the sample of spiral galaxies taken from Li & Wang (2013). Remarkably the parameter space occupied by the spiral galaxies is similar to that of coreless ETGs. In both cases, there is no clear positive correlation. 5.4. Comparison with Clusters and Groups In Figure 8, we introduce a schematic diagram to compare the LX, GAS­TGAS relations in the ETG samples we have studied with those reported for cDs (O'Sullivan et al. 2003), groups (Helsdon & Ponman 2003), and cluster of galaxies (Pratt et al. 2009). To first approximation, the bigger the system, the hotter and more luminous the gas. The details are more complex, though. Starting from the bottom left corner, the coreless ETGs and spiral galaxies show no clear correlation (see Figures 4(c) and 7). On the other hand, genuine Es show a tight correlation with a slope of 4.5 ± 0.3 (see Figures 4(c) and 5). Also this tight relation is well reproduced among -supported, hot gas rich elliptical galaxies by high-resolution simulations (Negri et al. 2014). The cDs follow a similar trend as the core Es, but are shifted toward higher LX,GAS values for a given TGAS. The larger dark matter halo in which these cDs are embedded may be responsible for the retention of large hot gaseous halos 13
Figure 8. Comparison of the LX,GAS­TGAS relations in various samples. From the bottom left, the coreless ETGs and spirals have no correlation, while the normal (non-CD) core E galaxies have a very tight correlation (LX,GAS TGAS4.5). The cDs and groups have a similar trend with the core Es, but they are shifted toward higher LX,GAS. The clusters at the top right corner have a flatter relation (LX,GAS TGAS3) that other sub-samples. For a reference, the self similar 2 expectation (LX,GAS TGAS ) is shown in dashed lines.

(larger LX). However, the temperature of this gas may not increase because of the presence of cooling connected with the its larger densities (see Section 5.3). The groups also have a similar trend, but they are further shifted upward. The clusters at the top right corner have a strong, but flatter relation (LX, GAS TGAS3). As the exact difference between groups and clusters is somewhat subtle, the distinction in the LX,GAS­TGAS relation of groups and clusters may be ambiguous among the big groups and small clusters, it is possible that the slope continuously changes from 4.5 to 3 (see Sun et al. 2009).


The Astrophysical Journal, 812:127 (14pp), 2015 October 20

Kim & Fabbiano Binggeli, B., Popescu, C. C., & Tammann, G. A. 1993, A&AS, 93, 275 Binney, J. 2004, MNRAS, 347, 1093 Bogdan, A., & Goulding, A. 2015, arXiv:1502.0503 Booth, C. M., & Schaye, J. 2010, MNRAS Lett., 405, L1 Boroson, B., Kim, D. W., & Fabbiano, G. 2011, ApJ, 729, 12 (BKF) Cappellari, M., Emsellem, E., Krajnovic, D., et al. 2011, MNRAS, 413, 813 Cappellari, M., McDermid, R. M., Alatalo, K., et al. 2013, MNRAS, 432, 1862 Choi, E., Ostriker, J. P., Naab, T., et al. 2015, MNRAS, 449, 4105 Civano, F., Fabbiano, G., Pellegrini, S., et al. 2014, ApJ, 790, 16 Cole, S., Norberg, P., Baugh, C. M., et al. 2001, MNRAS, 326, 255 Crook, A. C., Huchra, J. P., Martimbeau, N., et al. 2007, ApJ, 655, 790 Denicolo, G., Terlevich, R., Terlevich, E., Forbes, D. A., & Terlevich, A. 2005, MNRAS, 358, 813­32 Diehl, S., & Statler, T. S. 2008, ApJ, 687, 986 Dullo, B. T., & Graham, A. W. 2013, ApJ, 768, 36 Emsellem, E., Cappellari, M., Krajnovic, D., et al. 2011, MNRAS, 414, 888 Eskridge, P., Fabbiano, G., & Kim, D.-W. 1995, ApJS, 97, 141 Fabbiano, G. 1989, ARA&A, 27, 87 Fukazawa, Y., Botoya-Nonesa, J. G., Pu, J., Ohto, A., & Kawano, N. 2006, ApJ, 636, 698 Gabor, J. M., & Dave, R. 2014, arXiv:1405.1043 Gallagher, J. S., Garnavich, P. M., Caldwell, N., et al. 2008, ApJ, 685, 752 Grevesse, N., & Sauval, A. J. 1998, SSRv, 85, 161 Helsdon, S. F.., & Ponman, T. J.. 2003, MNRAS, 340, 485 Hopkins, P. F., Lauer, T. R., Cox, T. J., Hernquist, L., & Kormendy, J. 2009, ApJS, 181, 486 Howell, J. H. 2005, AJ, 130, 2065 Kim, D.-W. 2012, in Astrophysics and Space Science Library, Vol. 378, Hot Interstellar Matter in Elliptical Galaxies, ASSL, ed. D.-W. Kim & S. Pellegrini (New York: Springer), 121 Kim, D.-W., & Fabbiano, G. 2010, ApJ, 721, 1523 Kim, D.-W., & Fabbiano, G. 2013, ApJ, 776, 116 Kormendy, J., & Ho, L. C. 2013, ARAA, 51, 511 Kormendy, J., Fisher, D. B., Cornell, M. E., & Bender, R. 2009, ApJS, 182, 216 Krajnovi, D., Emsellem, E., Cappellari, M., et al. 2011, MNRAS, 414, 2923 Krajnovic, D., Karick, A. M., Davies, R. L., et al. 2013, MNRAS, 433, 2812 Kuntschner, H., Emsellem, E., Bacon, R., et al. 2010, MNRAS, 408, 97 Lauer, T. 2012, ApJ, 759, 64 Lauer, T. R., Ajhar, E. A., Byun, Y.-I., et al. 1995, AJ, 110, 2622 Lauer, T. R., Gebhardt, K., Faber, S. M., et al. 2007, ApJ, 664, 226­56 Li, J.-T., & Wang, Q. D. 2013, MNRAS, 428, 2085 Mathews, W., & Brighenti, F. 2003, ARA&A, 41, 191 Mathews, W., Brighenti, F., Faltenbacher, A., et al. 2006, ApJL, 652, L17 McDermid, R. M., Emsellem, E., Shapiro, K. L., et al. 2006, MNRAS, 373, 906 Nagino, R., & Matsushita, K. 2009, A&A, 501, 157 Negri, A., Pellegrini, S., & Luca, C. 2015, arXiv:1501.07111 Negri, A., Posacki, S., Pellegrini, S., & Ciotti, L. 2014, MNRAS, 445, 1351 Nipoti, C., & Binney, J. 2007, MNRAS, 382, 1481 O'Sullivan, E., Forbes, D. A., & Ponman, T. J. 2001, MNRAS, 328, 461 O'Sullivan, E., Ponman, T. J., & Collins, R. S. 2003, MNRAS, 340, 1375 Pellegrini, S. 1999, A&A, 351, 487 Pellegrini, S. 2005, MNRAS, 364, 169 Pellegrini, S. 2011, ApJ, 738, 57 Pellegrini, S., Ciotti, L., & Ostriker, J. P. 2012, ApJ, 744, 21 Pratt, G. W., Croston, J. H., Arnaud, M., & Bohringer, H. 2009, A&A, 498, 361 Randall, S., Nulsen, P., Forman, W. R., et al. 2008, ApJ, 688, 208 Rangarajan, F. V. N., White, D. A., Ebeling, H., & Fabian, A. C. 1995, MNRAS, 277, 1047 Sarzi, M., Alatalo, K., Blitz, L., et al. 2013, MNRAS, 432, 1845 Schombert, J. M. 1986, ApJS, 60, 603 Serra, P., Oosterloo, T., Morganti, R., et al. 2012, MNRAS, 422, 1835 Sun, M., Voit, G. M., Donahue, M., et al. 2009, ApJ, 693, 1142 Terlevich, A. I., & Forbes, D. A. 2002, MNRAS, 330, 547 Thomas, D., Maraston, C., Bender, R., & Mendes de Oliveira, C. 2005, ApJ, 621, 673 Tonry, J. L., Dressler, A., Blakeslee, J. P., et al. 2001, ApJ, 546, 681 Tully, B. 1998, Nearby Galaxies Catalog (Cambridge: Cambridge Univ. Press) Werner, N., Oonk, J. B. R., Sun, M., et al. 2014, MNRAS, 439, 2291 Young, L. M., Bureau, M., Davis, T. A., et al. 2011, MNRAS, 414, 940 Young, L. M., Scott, N., Serra, P., et al. 2014, MNRAS, 444, 3408

The LX,GAS­TGAS relation expected by the self similar case (where gravity dominates) has a slope of 2 (dashed lines in Figure 8). The steep slope (3) in clusters indicates that baryonic physics is already important even in this largest scale. In galaxies, the slope is even steeper (4.5), further indicating the increase of importance of non-gravitational effects (including SF, AGN and their feedback). 6. SUMMARY (1) Using the well-characterized ATLAS3D sample of 61 ETGs, we confirm our previous results (BKF) of the existence of two correlations in ETG samples: LX,GAS­LK and LX,GAS­TGAS. The best fit relations among the entire ATLAS3D X-ray sample are LX,GAS LK 3 and LX,GAS­ TGAS5, but the rms deviations are still large in both relations (0.7­0.8 dex or a factor of 5­6). (2) Among normal (non cD) core galaxies which are considered as passively evolving (no SF), --supported (slow rotation) Es, the best fit relations are LX,GAS LK 2.8 and LX,GAS ­TGAS4.5. The most significant result is the tight correlation between LX,GAS and TGAS. The rms deviation is 0.2 dex. If we exclude three galaxies with indirect signatures of recent SF, the rms deviation of 11 genuine Es is further reduced to 0.13 dex (or a factor of 1.3). (3) For the gas-rich galaxies (LX > 1040 erg s-1), the tight LX, GAS­TGAS relation is consistent with the recent simulations of Negri et al. (2014). However, the tight relation unexpectedly extends to gas-poor galaxies (LX, GAS = a few â1038­1040 erg s-1), where the gas is expected to be in a wind/outflow state. Our result may suggest the presence of small bound hot halos even in this low luminosity range. (4) We found that cD-type galaxies (central dominant galaxies in groups and clusters) follow similar scaling relations, but have an order of magnitude higher LX,GAS than giant elliptical galaxies with similar LK and TGAS. Based on the presence of extended cold gas in several cDs, we suggest that enhanced cooling in cDs, which have higher hot gas densities and lower entropies, could lower TGAS to the range observed in giant Es. (5) Among coreless ETGs LX,GAS and TGAS do not correlate at all, similar to what is found among spiral galaxies, suggesting that various additional mechanisms (e.g., rotation, flattening, SF) play a significant role to disrupt the hot ISM. We thank Silvia Pellegrini for helpful discussions. The data analysis was supported by the CXC CIAO software and CALDB. We have used the NASA NED and ADS facilities, and have extracted archival data from the Chandra Data Archive. This work was supported by the Chandra GO grant AR4-15005X, by 2014 Smithsonian Competitive Grant Program for Science and by NASA contract NAS803060 (CXC). REFERENCES
Annibali, F., Bressan, A., Rampazzo, R., et al. 2010, A&A, 519, A40 Bender, R., Surma, P., Doebereiner, S., Moellenhoff, C., & Madejsky, R. 1989, A&A, 217, 35

14