EVERY BCG WITH A STRONG RADIO AGN HAS AN X-RAY COOL CORE: IS THE COOL CORE–NONCOOL CORE DICHOTOMY TOO SIMPLE?

The importance of active galactic nucleus (AGN) outflows for cosmic structure formation and evolution has recently been widely appreciated. AGN outflows may simultaneously explain the antihierarchical quenching of star formation in massive galaxies, the exponential cutoff at the bright end of the galaxy luminosity function (LF), the M_SMBH–M_bulge relation, and the quenching of cooling flows in cluster cores (e.g., Scannapieco et al. 2005; Begelman & Nath 2005; Croton et al. 2006; Best et al. 2006). Outflows from radio AGNs are especially important locally because nearly all strong radio AGNs are hosted by early-type galaxies that dominate the high end of the LF and the galaxy population in clusters. Radio AGNs are different from emission-line AGNs selected in optical surveys. Kauffmann et al. (2004) showed that optical emission-line AGNs favor low-density environments and are mainly produced by black holes (BHs) with masses below 10⁸ M_☉. On the other hand, radio-loud AGNs favor denser environments than do normal galaxies, and the integrated radio luminosity density comes from the most massive BHs (Best et al. 2005). Best et al. (2005) also showed that the fraction of galaxies that are radio AGNs increases with the stellar or BH mass (as M^2.5_* or M^1.6_BH), while Shabala et al. (2008) found a similar relation as M^2.1±0.3_* or M^1.8±0.5_BH. The analysis by Best et al. indicates no correlation between radio and optical emission-line luminosities for radio AGNs, and the probability that a galaxy of given mass is radio loud is independent of whether it is an optical AGN. These results drove Best et al. (2005) to suggest that low-luminosity radio AGNs (FR I or fainter) form a distinctly different group from optical emission-line AGNs. They suggested that the radio AGN activity is associated with cooling of gas from the hot halos surrounding elliptical galaxies and clusters (see also Hardcastle et al. 2007). A similar idea has been proposed by Croton et al. (2006). Besides the classical "quasar mode" in AGN feedback, they introduced a "radio mode," which is the result of the X-ray gas accretion onto the central BHs. The inclusion of this "radio mode" in their simulations allows suppression of both excessive cooling and growing of very massive galaxies. It also explains why most massive galaxies are red bulge-dominated systems containing old stars.

Radio activity of galaxies is enhanced in clusters and the distribution of cluster radio AGNs is highly concentrated (e.g., Lin & Mohr 2007; Best et al. 2007). Among all cluster galaxies, the brightest cluster galaxy (BCG) is generally the most massive galaxy and has the biggest impact on the surrounding large cool core (LCC) and the general intracluster medium (ICM) properties. With the SDSS C4 cluster sample, Best et al. (2007) found that BCGs are more likely to host a radio AGN than other galaxies of the same stellar or BH mass (see also Lin & Mohr 2007). This difference implies that either BCGs stay in each radioactive cycle for a longer time or they are more frequently retriggered than non-BCGs. The mutual interaction between radio AGNs (especially those of BCGs) and the surrounding ICM has a significant impact on both of them. The great power of jets from radio AGNs can not only quench cooling in cluster cool cores, but also can drive the ICM properties away from those defined by simple self-similar relations involving only gravity (summarized in Voit 2005). The ICM around the radio AGNs provides a historical chronicle of the SMBH activity. X-ray cavities and shocks serve as calorimeters for the total energy outputs of AGNs that allow us to understand a great deal of AGN feedback and SMBH growth (summarized by McNamara & Nulsen 2007). On the other hand, radio AGNs (FR-I or fainter) may need an enhanced X-ray atmosphere to fuel them. Churazov et al. (2005) used Galactic X-ray binaries as analog for the different evolution states of the SMBH, from radiative-efficient mode with weak outflows (quasars) to radiative-inefficient mode with strong outflows (local giant ellipticals). The SMBHs of local giant ellipticals have very low accretion rates and radiative efficiencies, but are very efficient to heat the surrounding gas. Allen et al. (2006) presented a tight correlation between the Bondi accretion rate and the mechanical power of radio AGN for BCGs in nine groups and clusters. The relation implies that 1%–4% of mass energy of the gas accreted through Bondi accretion is transferred to the feedback energy to heat the surrounding ICM. Hardcastle et al. (2007) further suggested that the accretion of hot gas is sufficient to power all low-excitation radio AGNs (FR-I and some FR-II), while the high-excitation FR-II AGNs are powered by the accretion of cold gas.

As $\dot{M}_{\rm Bondi} \propto M_{\rm BH}^{2} K^{-1.5}$ (K is entropy of the surrounding gas defined as kT/n^2/3_e), strong radio AGNs require low-entropy circumnuclear gas and massive black holes. Although many groups and clusters have large and dense cool cores (e.g., ∼50% in the HIFLUGCS sample; Chen et al. 2007), most cluster galaxies (including many BCGs) are not located in such LCCs. Their surrounding ICM has too high an entropy (> 50 times the value at the center of cluster cool cores) to trigger any significant nuclear activity. Thus, the emerging significant questions are: What fuels the strong radio AGNs in groups and clusters, especially those not in LCCs? Is an X-ray cool core always required for strong radio AGNs in groups and clusters? Is the sufficient hot gas supply the reason that BCGs stay in the radioactive phase longer? The work mentioned in the last two paragraphs only discussed optical and radio data of radio AGNs. Chandra and XMM-Newton allow detailed studies of the X-ray atmosphere around radio AGNs in groups and clusters. Small X-ray thermal halos around the nuclei of strong radio AGNs are often found (e.g., Hardcastle et al. 2001 on 3C 66B; Hardcastle et al. 2002 on 3C 31; Worrall et al. 2003 on NGC 315; Hardcastle et al. 2005 on 3C296; Sun et al. 2005a on NGC 1265; Evans et al. 2006 on a sample of FR-I and FR-II galaxies). Hardcastle et al. (2007) also summarized some recent Chandra and XMM-Newton results for the hot gas atmosphere of nearby radio galaxies but the sample with detailed studies is small and those radio AGNs are mainly in groups.

With different original motivations, Sun et al. (2007, S07 hereafter) presented a systematic analysis of X-ray thermal coronae of early-type galaxies in 25 hot (kT > 3 keV), nearby (z < 0.05) clusters, based on Chandra archival data. Small and cool galactic coronae (⩽4 kpc in radius and kT = 0.5–1.1 keV generally) have been found to be common, >60% in L_Ks > 2 L* galaxies. Although much smaller and fainter than the classical cluster cool cores like those in Perseus, Virgo, and A478, they are in fact mini versions of LCCs (with cooling time of 10 Myr–1 Gyr from the center to the boundary), composed of the ISM gas (from the stellar mass loss) pressure confined by the surrounding ICM. These small coronae managed to survive strong ICM stripping, evaporation, rapid cooling, and powerful AGN outflows (see S07 for detailed discussions). One interesting result noted by S07 is the connection between strong radio AGNs and small coronae (Section 4.4 of S07). However, the S07 sample size is small (nine galaxies with L_1.4 GHz > 10²⁴ W Hz⁻¹) and only small coronae were discussed. Sun et al. (2009, S09 hereafter) added six more examples of BCG coronae with strong radio AGNs, but the combined sample with S07 is still small. The questions raised early need to be addressed with a bigger sample. In this work, we present results of such a sample from the Chandra archive (186 galaxies from 152 groups and clusters, including all BCGs, 74 galaxies with L_1.4 GHz > 10²⁴ W Hz⁻¹). The plan of this paper is as follows. The galaxy sample is defined in Section 2. The data analysis is presented in Section 3, as well as the definition of cool cores in this work. In Section 4, we discuss the X-ray gas atmosphere of BCGs and non-BCGs with luminous radio AGNs from the sample study. After studying the general properties of the sample, we present a detailed analysis of a luminous corona associated with a strong radio AGN in Section 5 (ESO 137-006). Many Chandra data are shallow, so only upper limits of coronal emission exist in some cases. In Section 6, we discuss the faintest corona known that sheds light on the properties of faint embedded coronae. Discussions are in Section 7. Section 8 contains the conclusions. We assumed H₀ = 71 km s⁻¹ Mpc⁻¹, Ω_M = 0.27, and Ω_Λ = 0.73.

We want to examine the X-ray gas component associated with BCGs and other strong radio AGNs, either LCCs or small coronae. The typically small luminosity of a corona (∼10⁴¹ erg s⁻¹ in the 0.5–2 keV band) limits our studies to local systems. S07 only studied kT > 3 keV clusters at z < 0.05. In this work, we include any Chandra observations of z < 0.065 groups and clusters with an exposure of longer than 5 ks. We also include higher redshift systems with sufficient Chandra data, but with a hard limit of z ⩽ 0.11. As we want to cover more volume in a single group or cluster, we limit our sample to z > 0.01 systems. There are very few strong radio AGNs in z < 0.01 groups and clusters anyway. The group and cluster sample is collected from the Chandra archive. Two X-ray flux-limited samples, B55 (Peres et al. 1998) and HIFLUGCS (Reiprich & Böhringer 2002), are almost 100% covered by Chandra. However, both samples are not big and have few or no groups. As less than 1/3 of the REFLEX and NORAS clusters (Böhringer et al. 2000; Böhringer et al. 2004) have Chandra data, we are forced to include as many systems from the archive as we can. The final sample includes 152 groups and clusters (Table 1). Most of the conclusions of this work should not be much affected by the heterogeneous nature of the cluster sample. We will also discuss the B55 and the extended HIFLUGCS samples in Section 7.4.

Our main goal is to examine the X-ray gas atmosphere around BCGs so all BCGs are included in the galaxy sample. The BCG is defined as the most luminous galaxy in the Two Micron All Sky Survey (2MASS) K_s band within 0.1 r₅₀₀ of the group or cluster. In relaxed groups and clusters, the selection of BCGs is easy. In nine unrelaxed groups and clusters, two BCGs with comparable K_s-band luminosities are selected, including Coma, A2147, A1367, A1142, A3128, RXC J1022.0+3830, SC1329-313, A1275, and A2384. Therefore, the BCG sample includes 161 galaxies. We also want to include all galaxies with a 1.4 GHz luminosity higher than 10²⁴ W Hz⁻¹ (L_1.4GHz,cut). A radio spectral index of −0.8 is assumed in the small radio K-correction. This L_1.4GHz,cut is about 1/3 of the L_* of the radio luminosity function (Best et al. 2005) and is comparable to the luminosities of many nearby 3C or PKS radio galaxies. For comparison, the central radio galaxy of the Centaurus cluster (NGC 4696, PKS 1245-41) has an L_1.4 GHz of 0.60 ×10²⁴ W Hz⁻¹. The giant radio galaxy in our backyard, Centaurus A, has an L_1.4 GHz of 0.46×10²⁴ W Hz⁻¹. The chosen L_1.4GHz,cut corresponds to an average mechanical power of ∼6 × 10⁴³ erg s⁻¹ (but with large scatter; Bîrzan et al. 2008), which is capable of increasing the thermal energy of a 10¹¹ M_☉ gas core (typical for the cool cores of luminous groups and poor clusters) at kT = 1 keV by ∼40% in 100 Myr. For a small corona with a total gas mass of ∼10⁸ M_☉ (S07), this kind of radio outbursts will easily destroy the whole corona if only ∼1% of the total mechanical power is deposited within the corona for 10⁸ yr. The radio fluxes come from the NRAO VLA Sky Survey (NVSS; Condon et al. 1998) and the Sydney University Molonglo Sky Survey (SUMSS; Bock et al. 1999). We also examine the higher resolution images from the Faint Images of the Radio Sky at Twenty Centimeters (FIRST) survey (Becker et al. 1995) and literature if available. The origins of radio sources can be determined in all cases. Because of the proximity, spectroscopic redshifts are available for all radio galaxies in our sample. We generally use the system redshift (Table 1) to calculate distance. For the Centaurus cluster and the NGC 7619 group, we used their surface-brightness-fluctuation distance from Tonry et al. (2001).

Radio AGNs are overrepresented in our BCG sample. There are 50% of BCGs (81 out of 161) with L_1.4 GHz > 10²³ W Hz⁻¹ and 32% of BCGs (52 out of 161) with L_1.4 GHz > 10²⁴ W Hz⁻¹. Lin & Mohr (2007) derived the radioactive fractions of BCGs for 573 groups and clusters selected from the NORAS and REFLEX cluster catalogs. Their fractions at these two thresholds are 33% and 20%, respectively. von der Linden et al. (2007) and Best et al. (2007) examined a sample of 1106 groups and clusters optically selected from Sloan Digital Sky Survey (SDSS). The radioactive fraction of BCGs is ∼30% at L_1.4 GHz > 10²³ W Hz⁻¹, for the typical stellar mass of BCGs in our sample. This fact again demonstrates the heterogeneous nature of our sample, which should be kept in mind when our results are interpreted. In fact, radio AGNs are also overrepresented in the B55 sample and the extended HIFLUGCS sample (Section 7.4). The radio AGN fractions of BCGs in these two samples are similar to ours, ∼50% for L_1.4 GHz > 10²³ W Hz⁻¹ AGNs and ∼30% for L_1.4 GHz > 10²⁴ W Hz⁻¹ AGNs. As X-ray flux-limited samples, the overabundance of radio AGNs of BCGs in both samples is mainly caused by the prevalence of LCC clusters in these samples, as strong radio AGNs are more likely to be associated with luminous cool core clusters (see Section 4).

All observations were performed with the Chandra Advanced CCD Imaging Spectrometer (ACIS). A standard data analysis was performed, which includes the corrections for the slow gain change¹ and charge transfer inefficiency (for both the FI and BI chips). We investigated the light curve of source-free regions (or regions with a small fraction of the source emission) to identify and exclude time intervals with strong particle background flares. We corrected for the ACIS low-energy quantum efficiency (QE) degradation due to the contamination on ACIS's optical blocking filter,² which increases with time and is positionally dependent. The dead area effect on the FI chips, caused by cosmic rays, has also been corrected. We used CIAO3.4 for the data analysis. The calibration files used correspond to Chandra calibration database (CALDB) 3.5.2 from the Chandra X-ray Center. In the spectral analysis, a lower energy cut of 0.4 keV is used to minimize the effects of calibration uncertainties at low energy. The solar photospheric abundance table by Anders & Grevesse (1989) is used in the spectral fits. In the spectral analysis, cash statistics is used for faint sources. Uncertainties quoted in this paper are 1σ. A special data analysis related to ESO 137-006 is discussed in Section 5.1.

In this work, X-ray cool cores include both LCCs and small coronae. How do we define an LCC? A core is considered as an LCC if its central isochoric cooling time is less than 2 Gyr. The X-ray luminosity of an LCC is measured within a radius where the gas isochoric cooling time is 4 Gyr (r_4Gyr). The cooling time profiles come from S09, Cavagnolo et al. (2009)³ and our own work. Note that r_4Gyr effectively means the whole region for small coronae (S07 and Section 5). Reasons behind these thresholds of cooling time are as follows. First, there are systems with central cooling time between 1 Gyr and 10 Gyr, or weak-cool-core clusters (e.g., Mittal et al. 2009; Pratt et al. 2009; Cavagnolo et al. 2009). Some weak cool cores also have a BCG corona, e.g., A3558 with an ICM central cooling time of ∼4 Gyr (S07), A1060 with an ICM central cooling time of ∼3.6 Gyr (Yamasaki et al. 2002), and A2589 with an ICM central cooling time of ∼2.5 Gyr beyond the central source (this work). We want to select a cut of the central cooling time that excludes these sources so 2 Gyr is chosen. Systems with central cooling time of <1 Gyr are usually considered as strong cool cores (e.g., Mittal et al. 2009; Pratt et al. 2009). Our threshold of 2 Gyr includes some weak cool cores, but not many. There are seven clusters in this sample with a central cooling time of 1–2 Gyr: A3571, A2244, A2063, A2142, A4038, A1650, and A2384 (north). None of them has a BCG corona (likely already merged with the large ICM cool core) and their radio AGNs are always faint (L_1.4 GHz < 6 × 10²² W Hz⁻¹). Thus, all LCCs with L_1.4 GHz > 2 × 10²³ W Hz⁻¹ in this work are strong cool cores with a central cooling time of ⩽1 Gyr (see Section 4). Second, we want these two thresholds not too close to reflect the still high luminosities of the weak cool cores, but also not too far to have many systems fall between two thresholds. Thus, 4 Gyr is chosen as the aperture for the luminosity measurement. A somewhat different cooling time (3–5 Gyr) does not affect any conclusions of this work. There are seven systems in this "gray" area with a central cooling time of 2–4 Gyr: A2589, A2657, A3558, A1060, AS405, A3266, and A3562. Based on our definition, they are not LCCs, so any cool cores associated with their BCGs can only be small coronae, which indeed is the case for some of them (A3558, A1060, and A2589). AS405 has the most luminous radio AGNs in this group with an L_1.4 GHz of 1.65×10²³ W Hz⁻¹, while the radio AGNs with other BCGs are all fainter than 3×10²² W Hz⁻¹. As our focus is on BCGs with strong radio AGNs, the exact classification of these systems in the "grey" area have little impact on our conclusions. In fact, we can conclude that weak cool cores do not have strong radio AGNs, unless a corona is present.

Section 3 of S07 detailed the analysis on the identification of faint thermal sources. We always perform a spectral analysis. Most of the sources studied in this work have sufficient counts for a detailed spectral analysis. However, there are faint sources (e.g., with less than 100 counts in the 0.5–2 keV band), so the significance of the iron L-shell hump or the "softness" of the spectrum needs to be tested (see Section 3 of S07). We used the S07 method with Monte Carlo simulations (Section 3.1 of S07) to address the significance of the iron L-shell hump. Only sources with an iron L-shell hump that is more significant at >99.5% level are considered thermal coronae. For faint sources that do not meet the above criteria, we further identified "soft X-ray sources" with a power-law fit as defined in Section 3.2 of S07. The power-law index method is similar to the 1.5–7 keV/0.5–1.5 keV hardness ratio method, as coronae should have strong excess emission above the continuum in the 0.5–1.5 keV band. As argued in S07, most of the "soft X-ray sources" identified by the power-law index method should be genuine coronae. In fact, after the S07 work, deeper Chandra data for A3627 and A2052 had been available. The data cover three sources identified as "soft X-ray sources" by S07, ESO 137-008 in A3627, CGCG 049-092 and PGC 093473 in A2052 (Table 2 of S07). The deeper data clearly reveal a significant iron L-shell hump in each spectrum and confirm the corona nature for all of them. This work identified 73 small coronae. Only seven of them are flagged as "soft X-ray sources" that are not as robust as others (five in Table 2). Thus, this uncertainty has little impact on the conclusions of this work. For sources not identified as coronae or "soft X-ray sources," upper limits on the thermal emission are given with the method listed in S07.

We emphasize that the number of coronae determined in this work is only a lower limit, as Chandra exposures are not optimized for studies of faint galactic emission. As shown in Section 6, the faintest embedded corona known has an X-ray luminosity much lower than any upper limits in this work. About 2/3 of galaxies only with upper limits of coronal emission are detected in X-rays. They are either very faint sources or very bright X-ray AGNs that make the identification of thermal emission difficult. NGC 3862 (or 3C 264) in A1367 is a good example (S07, also in Table 2). If not for an on-axis subarray observation, the weak thermal emission of its corona cannot be confirmed only from the spectral analysis. Different from S07, we also did not use any stacking and always studied the spectrum of a single galaxy. Thus, the identification of thermal coronae in this work was conservative.

The main cool core property we measured is the 0.5–2 keV luminosity. This energy band is where the thermal emission of a corona is significant, even if a strong X-ray AGN is present. For LCCs, the luminosity is derived from the spectral fit to the total spectrum within r_4Gyr. Multi-kT components are often required for bright cores. For coronae, the immediate local background was always used. A power-law component was always included in the spectral fits to account for the nuclear and LMXB emission. The gas abundance cannot be constrained for faint coronae so it was fixed at 0.8 solar as derived by S07.⁴ The derived typical abundance for bright coronae is also consistent with this assumed value (e.g., see Section 5.3 for ESO 137-006). For faint coronae, the uncertainty of the X-ray luminosity mainly comes from the statistical error. The uncertainty of the nonthermal component does not affect the 0.5–2 keV luminosity much, as the identified faint coronae have dominant thermal emission in the soft band. Faint sources with a hard spectrum will not be identified as coronae as stated previously in this section.

We first examine BCGs in the L_0.5-2 keV (r < r_4Gyr)–L_1.4 GHz plane (Figure 1). BCGs in groups (kT < 2 keV), poor clusters (2 keV <kT < 4 keV) and rich clusters (kT > 4 keV) are color-coded. The system temperatures come from BAX,⁵ S09, Cavagnolo et al. (2009), and our own work. As shown in Figure 1, almost all cool cores (including upper limits) can be divided into two classes, marked by a vertical ellipse for small coronae and a tilted ellipse for LCCs. The dividing line between the two classes is L_0.5-2 keV ∼ 4 × 10⁴¹ erg s⁻¹. The gap between two classes is especially significant at L_1.4 GHz > 2 × 10²³ W Hz⁻¹. There are 52 radio sources with L_1.4 GHz > 10²⁴ W Hz⁻¹ and 81 radio sources with L_1.4 GHz > 10²³ W Hz⁻¹. Above L_1.4 GHz > 2 × 10²³ W Hz⁻¹, every BCG has a confirmed cool core, small or large. In fact, there are only two nondetections of cool cores out of 81 BCGs above L_1.4GHz of 10²³ W Hz⁻¹, which is the dividing line between the star formation and AGN components in the local radio LF (e.g., Sadler et al. 2002). Below that threshold, radio emission from star formation begins to dominate the local radio LF. This threshold was also used in the statistical studies by Lin & Mohr (2007) and von der Linden et al. (2007). Upper limits of both nondetections (AS405 and A2572) are high (Figure 1) as both observations are short (7.9 ks) with ACIS-I, especially for AS405 at z = 0.0613. A faint soft X-ray point source is actually detected in the position of A2572's BCG (z = 0.0403) but was rejected as a corona as the error of the power-law index is too large (S07; Section 3). Thus, the current data are consistent with the conclusion that all BCGs with L_1.4 GHz > 10²³ W Hz⁻¹ have a cool core, small or large.

Zoom In Zoom Out Reset image size

BCGs with small coronae often host radio AGNs as luminous as those BCGs in LCCs. We call these two classes the LCC class and the corona class. The cores in the LCC class are the cool cores that were generally referred to in the cluster papers. The corona class defined by this work also includes ∼5 systems with small cool cores (up to 25 kpc in radius) but larger than typical coronae (less than 5 kpc in radius generally). We present the properties of the X-ray sources associated with L_1.4 GHz > 10²⁴ W Hz⁻¹ radio AGNs in the corona class in Table 2 (also including non-BCGs). Some examples of BCG coronae associated with strong radio AGNs are also shown in Figures 2 and 3. We emphasize that the clusters or groups shown in Figures 2 and 3 were considered "noncool core" systems by previous work (e.g., Mittal et al. 2009) so other mechanisms (e.g., cluster merger) had to be used to explain strong radio AGNs. AWM4 is another example. It was once considered a puzzle as it hosts a radio AGN but lacks an LCC from the XMM-Newton data (Gastaldello et al. 2008). The new Chandra data (released in late 2009 May, after the initial submission of this paper) clearly show the presence of a small (∼2 kpc radius), thermal corona associated with the BCG (Figure 3).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.1. The LCC Class

4.2. The Corona Class

The properties of embedded coronae have been presented and discussed in detail by S07. Jeltema et al. (2008) presented a similar analysis on coronae in groups. We have added many more coronae for BCGs. As shown in Figure 1, X-ray luminosities of coronae show little correlation with the radio luminosities of the galaxy, although coronae with strong radio AGNs are usually luminous. As emphasized in S07 and early in this paper, small coronae in the upper portion of the corona class will be destroyed even if only <1% of AGNs heating is acted inside the corona. Being in rich environments (especially if in hot clusters), it is difficult to rebuild these mini-cool cores from stellar mass loss, once they are destroyed. S07 and this work show that very few coronae of massive galaxies should have been destroyed in this way. Powerful radio jets may simply penetrate the coronal atmosphere with very little energy deposition. However, if radio outbursts have little impact inside a small corona, is radio heating still responsible for quenching cooling inside a corona and turning off the nuclear activity? This is the same cooling/heating question in LCCs. Compared small coronae with LCCs, the global cooling is much weaker (but still strong in the center) while heating is not apparently different from that in LCCs (Figure 1). We will return to this issue in Section 7.

We also present the results for 22 non-BCGs with L_1.4 GHz > L_1.4GHz,cut (Figure 4). Most non-BCGs do not have a confirmed corona but the upper limits are usually high. Section 6 presents the faintest embedded corona known (NGC 4709), which is much fainter than all upper limits in this work. We also present the L_Ks − L_0.5-2 keV plot for all galaxies in this work (Figure 5). The corona class falls around the average relation derived by S07, while the cool cores in the LCC class are much X-ray brighter. We also included the expected emission from cataclysmic variables and coronally active stars (Revnivtsev et al. 2008), which is always small. As discussed in Section 6, the real contribution is even much smaller as we always used the immediate local background that also includes much stellar emission. S07 showed that massive galaxies generally have luminous coronae. There are some nondetections of coronae for massive galaxies in Figure 5. Generally the upper limits are still high. Interestingly, above L_Ks of 3.5 L_Ks,* (or 10^11.61L_Ks,⊙, the dotted line in Figure 5), the galaxies associated with L_1.4 GHz > 10²³ W Hz⁻¹ AGNs are generally more X-ray luminous than those with weak radio AGNs. Massive BCGs with a moderate corona or no corona usually only have weak radio AGNs. Thus, the strong connection between BCGs with strong radio AGNs and X-ray cool cores is not all because of their strong correlations with the galaxy mass.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.3. X-ray AGN Versus Radio AGN

Section 4.1 has depicted the general properties of the corona class. How does a BCG corona associated with a luminous radio AGN look in detail? In this section we present a detailed analysis of a BCG corona with a strong radio AGN, ESO 137-006 in A3627. ESO 137-006 is the brightest (not the most luminous) embedded corona in kT > 3 keV clusters (S07), because of its proximity. A new 57.3 ks ACIS-S observation collects about 3800 source counts in the 0.5–3 keV band (∼96% from the gas). Before this observation, NGC 3842 in A1367 (Sun et al. 2005b) had the most Chandra counts (∼1000 from a 43.2 ks ACIS-S observation) for an embedded corona. Other coronae discussed in the literature (e.g., Coma, Vikhlinin et al. 2001; NGC1265 in Perseus, Sun et al. 2005a) only have ∼500 counts typically. In fact, even including groups and poor clusters, only IC 4296 and NGC 315 have brighter coronae. However, both are in poor groups so the surrounding ICM has a comparable temperature. Both also host bright nuclear sources and there is a bright X-ray jet in NGC 315 (Worrall et al. 2003), which causes some trouble in the analysis of the X-ray gas. Radio AGNs associated with these two galaxies are >17 times fainter than ESO 137-006's. Thus, ESO 137-006 is the best case of a luminous corona associated with a strong radio AGN for a detailed analysis. In S07, we present the results from a 14.1 ks ACIS-I observation targeted on a cluster position that is 5' from ESO 137-006. Here, we present the results of a much deeper targeted ACIS-S observation in cycle 8 (PI: Sun).

At z = 0.01625 (Woudt et al. 2008), A3627 (kT∼ 6 keV) is the closest massive cluster. ESO 137-006 is the BCG of A3627. ESO 137-008 (also with a small corona) has the same K_s band magnitude as ESO 137-006, but its velocity dispersion is much smaller (see Table 2 of S07). The radio WAT source associated with ESO 137-006, PKS 1610-60 (Jones & McAdam 1996), is one of the brightest radio sources in the southern hemisphere and its 1.4 GHz luminosity is 53% higher than NGC 1275's (Figure 1; Table 2). The XMM-Newton mosaic image of A3627 is shown in Figure 6, along with the radio contours from SUMSS. One can observe evidence of the interaction between the radio lobes and the ICM in Figure 6, which implies that ESO 137-006 is not far from the cluster gas core. For the assumed cosmology (Section 1), the angular scale is 0.327 kpc arcsec⁻¹ and the luminosity distance of A3627 is 69.6 Mpc.

Zoom In Zoom Out Reset image size

5.1. Chandra Data Analysis

5.2. The Properties of the Surrounding ICM

Before we study ESO 137-006's corona, the properties of its surrounding ICM were examined to constrain the ambient gas temperature and pressure. As discussed in the Appendix of S07, the soft X-ray background is high in the direction of A3627. S09 developed a method to constrain the local X-ray background with the stowed background. Following the S09 method, we derived a local soft X-ray background flux surface density of (12.1 ± 1.0) ×10⁻¹² erg s⁻¹ cm⁻² deg⁻² in the 0.47–1.21 keV band. The RASS R45 flux was measured from 1–2 deg annulus centered on A3627, excluding a bright PS, is 290 ×10⁻⁶ counts s⁻¹ arcmin⁻². These two fluxes well match their average relation shown in Figure 2 of S09. The hotter thermal component of the local X-ray background has a temperature of 0.33 ± 0.03 keV (see the local background model in S09), which is consistent with the trend that its temperature increases to ⩾0.3 keV in the high R45 flux regions (S09).

With the local X-ray background determined, the temperature profile of the surrounding ICM was derived (∼7 keV, Figure 7). The ICM abundance from the joint fit of the three radial bins is 0.32 ± 0.09 solar, which is much lower than that of the small corona (the next section). The ICM electron density from the ROSAT data is ∼1.9 × 10⁻³ cm⁻³ at the projected position of ESO 137-006 (Böhringer et al. 1996).

Zoom In Zoom Out Reset image size

5.3. The Properties of ESO 137-006's Corona

The 0.5–5 keV Chandra count image of ESO 137-006 is shown in Figure 6. Beyond the bright core, the corona is more extended toward the south, which implies ESO 137-006's motion to the north. This is consistent with the small bending of its radio lobes (Figure 6). To better show the central part of the corona, we applied the Subpixel Event Repositioning tool developed by Li et al. (2004) to the data to make the 1/4 subpixel image (the right bottom panel of Figure 6). Within the central 0.5 kpc, the gas is flattened along the north–south direction, which is also perpendicular to the direction of the radio jets.

We derived the profiles of the X-ray surface brightness, temperature (projected and deprojected), density, entropy, cooling time, and pressure for the corona, as shown in Figure 7. In the spectral analysis, the local background is extracted from the 34''–120'' (11.1–39.2 kpc) annulus. VAPEC was used in each annulus, as it fits the spectral lines better. We used the classical "onion-peeling" method for deprojection. For such a massive galaxy, the LMXB emission should be significant. We used the L_X,LMXB–L_Ks scaling relation derived by Kim & Fabbiano (2004). As there are no HST data of ESO 137-006, we simply used the HST stellar light profile of NGC 3842 for the stellar light profile of ESO 137-006 (Sun et al. 2005b), as two BCGs have similar 2MASS K_s luminosities (0.22 mag difference). After adjusting to ESO 137-006's K_s-band luminosity and A3627's local background, one can see that the LMXB light + a flat local background well describes the observed profile beyond ∼4.2 kpc radius. In fact, the net spectrum in the 13''–34'' annulus (with the 34''–120'' spectrum as the local background) can be well fitted by a power law with a photon index of 1.7. Its flux is also consistent with the expected LMXB flux in the region (Table 4). Thus, we conclude that the corona is pressure confined at ∼4.2 kpc radius. In fact, the electron pressure ratio across the 4.2 kpc boundary is 1.16 ± 0.51, consistent with pressure equilibrium. Obviously, there are large jumps of entropy and cooling time across the boundary. Across the coronal boundary, heat conduction has to be suppressed by a factor of at least ∼230 from the Spitzer value, using the ratio of the heat conduction flux and the total X-ray flux (the method used in S07). Without any suppression, this small corona will be evaporated in ∼8 Myr (S07). Inside the boundary, the corona has low entropy and short cooling time (Figure 7) that is typical for the center of LCCs. The properties of ESO 137-006's corona are summarized in Table 4.

Table 4. Coronae of ESO 137-006 and NGC 4709

Propertya	ESO 137-006	NGC 4709
Cluster (z)	A3627 (0.0162)	Centaurus (0.0114)
log(LKs/L☉)	11.77	11.27
log(L1.4 GHz/W Hz−1)	25.39	<20.88
rcut (kpc)	4.24 ± 0.20	∼2.0
kT (keV)	0.91 ± 0.02b	0.78 ± 0.09
O (solar)	0.52+0.63−0.40	(0.8)
Ne (solar)	4.14+2.38−1.51	(0.8)
Mg (solar)	2.27+1.19−0.70	(0.8)
Si (solar)	1.66+0.82−0.48	(0.8)
S (solar)	2.76+1.55−0.99	(0.8)
Fe (Ni) (solar)	1.13+0.49−0.28	(0.8)
f0.5-2 keV,thermal,obs (erg cm−2 s−1)	(1.77 ± 0.09)×10−13	(9.5 ± 1.8)×10−15
L0.5-2 keV,thermal (erg s−1)	(1.69 ± 0.08) ×1041	(1.89 ± 0.36) ×1039
Lbol,thermal (erg s−1)	(2.76 ± 0.14) ×1041	(3.1 ± 0.6) ×1039
L0.3-8 keV,LMXB (r < rcut) (erg s−1)	(5.2±1.1) × 1040	∼ 5.3 × 1039
L0.3-8 keV,LMXB (r = rcut − 11.1 kpc) (erg s−1)	(4.1±0.8) × 1040	⋅⋅⋅
L0.3-10 keV,nucleus (erg s−1)	<4 × 1039	<5 × 1038
ne,center (cm−3)	∼0.5	∼0.034
tcooling,center (Myr)	∼7	∼130
Mgas (M☉)	(1.78 ± 0.15) ×108	∼ 2.9 × 106

Notes. ^aThe energy bands are all measured in the rest frame. ^bSee Figure 7 for ESO 137-006 temperature profile.

Download table as:  ASCIITypeset image

As the abundances in each annulus are all consistent with the same value (from the VAPEC model), we fix them together. The best-fit abundances from the VAPEC model are listed in Table 4. We also derived the abundance ratios: Si/Fe = 1.47^+0.94_−0.58, Fe/O = 2.17^+3.60_−1.42, and Fe/Mg = 0.50^+0.35_−0.20. These ratios can be compared with the theoretical estimates from Iwamoto et al. (1999): Fe/O = 0.26, Si/Fe = 3.03, and Fe/Mg = 0.26 for SN II, Fe/O = 27–75, Si/Fe = 0.54–0.62, and Fe/Mg = 31–69 for SN Ia. Thus, the abundances of the corona are not enriched by SN Ia only.

As shown in Figure 7, the central density is high. Can we constrain the gas properties around the Bondi radius? The mass of the central SMBH can be estimated from the K_s-band luminosity of the galaxy (Marconi & Hunt 2003; see Section 7.1), 1.61×10⁹ M_☉. The Bondi radius is then 62 pc (or 019) for a central temperature of ∼0.8 keV (Figure 7). This scale is unresolved. However, we can still constrain the range of gas density and entropy around the Bondi radius, from the measured total X-ray emission within the innermost bin. We assume n_e = n_e,0(r/r₀)^−α, where r₀ is the Bondi radius and n_e,0 is the electron density at r₀. Of course, α < 1.5. We integrated the density model to compare with the observed X-ray luminosity within the innermost bin, assuming a constant emissivity. As shown in Figure 8, the gas density at the Bondi radius is always ⩽1.8 cm⁻³, even when the point-spread function (PSF) correction is considered. Similarly, to compare the model with the emission-weight temperatures observed, we find that the gas temperature at the Bondi radius is 0.8–0.9 keV. These results will be used in Section 7.1 to examine whether Bondi accretion is sufficient to power the radio AGNs of ESO 137-006.

Zoom In Zoom Out Reset image size

In spite of its high central density, the size of ESO 137-006's corona is small, so the total gas mass is very small, (1.78 ± 0.19) ×  10⁸ M_☉ within 13'' (or 4.24 kpc) radius. This can be compared with the typical gas mass within 10 Gyr cooling radius for luminous cool cores in groups and clusters, ∼ 10¹¹–5 × 10¹² M_☉ (S09; Cavagnolo et al. 2009). Owing to the vast contrast, S07 named the embedded coronas as mini-cooling cores. As discussed in Section 7, this kind of mini-cooling cores is sufficient to fuel powerful FR-I radio galaxies. As shown in Figure 1 and Table 2, the X-ray luminosity of ESO 137-006's corona is not particularly high in the corona class. Thus, we expect that other luminous BCG coronae would have similar properties to those of ESO 137-006's.

There are upper limits of the coronal emission in Figures 1, 4, and 5, so it is interesting to know how faint an embedded corona can be. Strong ram pressure in groups and clusters generally only allows coronae with dense cores to survive (S07). While many BCGs have luminous coronae (L_0.5-2 keV ∼ 10⁴¹ erg s⁻¹; Figures 4 and 5), faint embedded coronae with moderate cool cores do exist. For BCGs associated with luminous radio AGNs, the faintest corona known is the one associated with NGC 3862 (or 3C 264) in A1367 (L_0.5-2 keV ∼ 1.4 × 10⁴⁰ erg s⁻¹, Table 2; S07). One can see that its luminosity is comparable with or smaller than all upper limits for nondetections in this work (Figures 4 and 5). How faint can a BCG corona be? In this section, we discuss the corona of NGC 4709, which has the lowest X-ray luminosity known for an embedded corona in a galaxy more luminous than L_*.

NGC 4709 is the dominant galaxy of a shock-heated subcluster falling into the Centaurus cluster (Churazov et al. 1999). A faint corona was detected near the edge of the S1 chip in a 34.3 ks Chandra observation (S07). Here we present the results from our new targeted observation. We adopt a distance of 35.3 Mpc for NGC 4709 (Tonry et al. 2001), which is much smaller than the distance we used in S07 (49.4 Mpc). The angular scale is 0.167 kpc arcsec⁻¹. The observation of NGC 4709 was performed with ACIS-S on 2007 March 25. No background flares were present. The effective exposure is 29.7 ks. An extended but faint source is detected at the position of NGC 4709. A strong iron L-shell hump centered at ∼0.9 keV clearly shows the presence of the thermal gas. The new targeted observation allows better removal of the local background. Half of the 0.5–2 keV counts are from a hard power-law component (likely LMXBs) and about 90 counts are from the emission of the thermal gas. The gas temperature is 0.78 ± 0.09 keV with a fixed abundance of 0.8 solar (see S07). The temperature difference from S07 comes from the different contribution of the hard power-law component in the spectral fits. The S07 fit to the old S1 data allows little flux from the hard component. The rest-frame 0.5–2 keV luminosity is 1.8×10³⁹ erg s⁻¹, which makes it the faintest corona known in clusters (S07 and this work). We compare its Chandra surface brightness profile with that of ESO 137-006 in Figure 9. The contribution of the LMXB light has been removed, assuming that it follows the stellar light. One can see the vast difference. The corona boundary is ∼2.0 kpc. Assuming a constant emissivity for the corona, the central electron density of the gas is 0.034 cm⁻³ and the total gas mass is 2.9 × 10⁶ M_☉ within a 2 kpc radius (Table 4). The central cooling time is ∼0.13 Gyr so it is still a moderate cool core. The properties of NGC 4709's corona are summarized in Table 4 to compare with those of ESO 137-006's corona.

Zoom In Zoom Out Reset image size

We indeed note that faint thermal emission can also come from the integrated emission of cataclysmic variables and coronally active stars (e.g., Revnivtsev et al. 2008). This emission component is linearly scaled with galaxy's K_s-band luminosity (Revnivtsev et al. 2008). As shown in Figure 5, the expected luminosity of this component gets close to that observed for NGC 4709, although it is far smaller than those of luminous coronae. However, one needs to remember that in our analysis of small coronae, local background is always used. For NGC 4709, the global spectrum is extracted from an aperture of 2.23 kpc in radius, while the local background is from the 2.23–5.82 kpc annulus. We analyzed the 2MASS K_s band image and found that the net stellar emission within 2.23 kpc radius is only ∼3.5% of galaxy's total light, after subtracting the local background from the 2.23–5.82 kpc annulus. Moreover, NGC 4709's X-ray source is more peaked than the optical light. Thus, we are confident that NGC 4709 has an X-ray gaseous halo.

Coronae like NGC 4709's are so faint that they are easily overlooked at z > 0.03. Coronae like 3C264's can also be easily overlooked at z > 0.07, especially when a bright X-ray nuclear source is present (the case for 3C264). This fact needs to be kept in mind when z ⩾ 0.05 FR-I radio galaxies or cluster BCGs are studied. A significant X-ray gas component that is sufficient to fuel the central AGNs may elude detection easily.

7.1. Fueling Radio AGNs

What fuels the radio AGNs in groups and clusters? We focus on radio AGNs associated with a small corona, as radio AGNs in the LCC class have been widely discussed in the literature. Even for small coronae, there is sufficient amount of gas cooled from the hot phase to fuel the radio AGNs. The maximum mass deposition rate from cooling (assuming steady and isobaric) is: $\dot{M}_{\rm cooling} \approx 2\,\mu$m_pL_X,bol/5 kT = 0.44 (L_bol/10⁴¹ erg s⁻¹) (kT/0.9 keV)⁻¹ M_☉ yr ⁻¹. We calculate $\dot{M}_{\rm cooling}$ for each corona in the upper portion of the corona class (Figure 10). The bolometric correction is made case by case. For upper limits, we assume a temperature of 0.7 keV (S07). The required SMBH accretion rate is estimated from the L_1.4 GHz-jet power relation by Bîrzan et al. (2008), assuming a mass-energy conversion efficiency of 0.1. As shown in Figure 10, the required SMBH accretion rate is always small and there should be enough amount of cooled gas in coronae. The reality is however more complicated. First, the above cooling rate is reduced if there are heat sources. Even if strong radio outbursts deposit very little heat inside a corona, SN heating can be significant as discussed in S07. On the other hand, the stellar mass loss rate from evolved stars is significant inside coronae, 0.2– 0.8 M_☉ yr⁻¹ (Faber & Gallagher 1976; S07). However, both the SN heating and the stellar mass loss should follow the stellar light profile, which is much shallower than the X-ray emission profile of coronae (S07). The detailed energy balance and evolution of gas in different phases is unclear. However, as argued in S07, cooling in the central kpc of a luminous corona (where most X-ray emission comes from) should overwhelm heating so the actual mass deposition rate should not be reduced much from the above estimates. Second, the L_1.4 GHz-jet power relation in Bîrzan et al. (2008) was derived from radio AGNs in the LCC class, where the radio luminosity of lobes may be enhanced by the high ICM pressure of LCCs (e.g., Barthel & Arnaud 1996). Moreover, the Bîrzan et al. (2008) relation likely underestimates the jet power by missing power from undetected weak shocks and cavities. These two factors should not increase the jet power by more than a factor of a few (McNamara & Nulsen 2007), so cooling of small coronae should provide enough fuel for their radio AGNs.

Zoom In Zoom Out Reset image size

Are radio AGNs fueled by hot gas (e.g., Best et a. 2005; Croton et al. 2006; Allen et al. 2006; Hardcastle et al. 2007)? Typically for hot accretion, Bondi accretion is assumed (e.g., Hardcastle et al. 2007). The Bondi accretion rate is $\dot{M}_{\rm Bondi} = 0.0042 (K / 2$ keV cm²)^−3/2(M_BH/10⁹M_☉)²M_☉ yr⁻¹. Here, K is the gas entropy at the Bondi radius. The Bondi radius in this sample (10–150 pc) is always unresolved. However, as shown in Section 5.3, the gas entropy can still be constrained from the observed X-ray luminosity of the innermost bin. For the best-studied case, ESO 137-006, we take the gas entropy at the Bondi radius as 0.8 ± 0.3 keV cm² (Section 5.3). There are four luminous coronae for which we performed detailed studies before: NGC 3842 and NGC 3837 (Sun et al. 2005b), NGC 1265 (Sun et al. 2005a), and 3C 465 (S07). Their central entropy values are ⩽1.1–1.8 keV cm². For the faintest corona of a massive galaxy, NGC 4709's corona, the central entropy is ∼6.5 keV cm² (Section 6). As we cannot perform a similar analysis for ESO 137-006's corona on other coronae, we simply assume an average entropy of 1.5 keV cm², with an uncertainty range of 0.5–3 keV cm² (note that Hardcastle et al. 2007 used an entropy value of 1.11 keV cm²). The SMBH mass is estimated from the 2MASS K_s band luminosity of the galaxy (Marconi & Hunt 2003), log(M_SMBH/M_☉) = 9.47 + 1.13 log(L_Ks/10¹²L_☉). The results are also shown in Figure 10. A large Bondi accretion rate needs a massive black hole and low-entropy surrounding gas. Both requirements point to massive galaxies (recall the L_X–L_Ks correlation for coronae, e.g., S07). Thus, low-mass galaxies with a faint corona may not power their radio AGNs through hot accretion (Figure 10), especially if the required SMBH accretion rate is higher for the reasons presented in the last paragraph. However, note that both the Marconi & Hunt (2003) relation and the Bîrzan et al. (2008) relation have large scatters (∼0.5–1 dex), which makes it impossible to reject hot accretion for any sources in the plot. On the other hand, SMBH spin can reduce the required accretion rate (e.g., Wilson & Colbert 1995; McNamara et al. 2009).

Besides hot gas, dust and molecular gas are also present in early-type galaxies and BCGs. Dust with mass ⩽10⁴–10⁵ M_☉ was detected in ∼80% of nearby large early-type galaxies (e.g., van Dokkum & Franx 1995). The origin of dust includes mergers with dust-rich dwarfs and a stellar origin (Mathews & Brighenti 2003). Interestingly, the dust detection rate is higher in radio-jet galaxies than in non radio-jet galaxies (e.g., de Koff et al. 2000; Verdoes Kleijn & de Zeeuw 2005; Simões Lopes et al. 2007). For galaxies with dust detections, FR-I galaxies also have more dust than radio-quiet ellipticals. The dust in FR-I galaxies is generally situated in sharply defined disks on small (<2.5 kpc) scales, while radio jets are generally perpendicular to the dust disk (e.g., de Koff et al. 2000). Sometime the dust disk is warped (e.g., 3C 449 in our sample; Tremblay et al. 2006), likely in the process of settling down to be regular disks or are being perturbed (Verdoes Kleijn & de Zeeuw 2005). The dust-jet connection is still not well understood (e.g., which impacts which?) but dust may play a role to fuel the central SMBH. MIR emission from PAHs and dust in some FR-I galaxies were also found from the Spitzer data (e.g., Leipski et al. 2009).

There is also mounting evidence of CO detections in radio galaxies (e.g., Lim et al. 2000; Evans et al. 2005; Prandoni et al. 2007). Three galaxies in our sample belong to this growing class, 3C 31 (a molecular gas mass of (4.8–10)×10⁸ M_☉, Lim et al. 2003; Evans et al. 2005; Okuda et al. 2005), 3C 264 or NGC 3862 (a molecular gas mass of 2.6×10⁸ M_☉, Lim et al. 2003), and 3C 449 (a molecular gas mass of 2.4 × 10⁸ M_☉; Lim et al. 2003). The CO emission of 3C 31 and 3C 264 exhibits a double-horned line profile characteristic of a rapidly rotating disk (Lim et al. 2000; Okuda et al. 2005). At least eleven other galaxies in our sample were undetected in these CO observations, with upper limits of (1–5)×10⁸ M_☉ (all mass values have been adjusted to our cosmology). It is intriguing that the estimated mass values or upper limits of the molecular gas are comparable or even larger than the mass values of the X-ray gas of their coronae, although the cold gas can in principle be produced through cooling over <1 Gyr. Note that CO detections are also present in the general samples of BCGs in LCCs (Edge 2001; Edge & Frayer 2003; Salomé & Combes 2003). Some of the BCGs with significant CO detections only have weak radio AGNs (e.g., A262, 2A 0335+096, and A2657 with L_1.4 GHz = 0.09–1.0 × 10²³ W Hz⁻¹). More detailed molecular data of BCGs with and without strong radio AGNs will better reveal the connection between the radio activity and the cold gas component. A related question is on the nuclear star formation in small coronae. Rafferty et al. (2008) and Cavagnolo et al. (2008) showed that star formation turns on when the central cooling time of the gas falls below ∼0.5 Gyr or the gas entropy falls below ∼30 keV cm² (at r = 3–4 kpc for ESO 137-006, Section 5). Will this be the case for small coronae? On the other hand, the energy balance and transfer between gases in different phases (molecules, dust, stellar winds, and 10⁷ K X-ray gas) is an interesting problem to explore.

To sum up, cooling of the X-ray coronae can provide enough fuel to the central SMBH, provided that the cooled materials can reach the very center. Bondi accretion may be sufficient for the massive black holes in a luminous X-ray corona, but likely not the only answer for the strong radio AGNs in the corona class. Figure 10 also does not necessarily argue for hot accretion for massive galaxies as there are many uncertainties. Besides, note that a small amount of angular momentum of the hot gas can largely reduce the accretion rate (e.g., Proga & Begelman 2003). The large scatter of the L_1.4 GHz-jet power relation also implies that some strong radio AGNs will require much higher mass accretion rates than the average values. On the other hand, dust and molecular gas coexist with the 10⁷ K X-ray gas in radio AGNs (including some in our sample), which may bring cold accretion into play. One issue we did not discuss is galaxy merger as many BCGs are dumbbells. Although the connection between the various gas components is unclear, we suggest that the existence of an X-ray corona with high pressure can effectively shield dust and cold gas from strong evaporation and stripping by the ICM, especially in hot clusters.

7.2. Are Coronae Decoupled from the Radio Feedback Cycle?

7.3. The Implications of the Corona Class

7.4. The B55 and the Extended HIFLUGCS Samples

Our sample (Table 1) includes nearly all systems in the B55 and the extended HIFLUGCS samples. The B55 sample (55 clusters; Peres et al. 1998) is a hard X-ray selected sample that has been fully covered by Chandra. There are no groups (kT < 2 keV) in the B55 sample. The basic HIFLUGCS sample has 63 groups and clusters (Reiprich & Böhringer 2002) that have been fully covered by Chandra. Reiprich & Böhringer (2002) also listed 11 systems that are brighter than the HIFLUGCS flux limit but within 20 deg from the Galactic plane, so they are not included in the HIFLUGCS sample. We include them and call the resulting sample the extended HIFLUGCS (HIFLUGCS-E) sample. Only the Antlia group does not have Chandra data, but its BCG is radio quiet anyway. Instead of marking the B55 and the HIFLUGCS-E systems in Figure 1, we plot the cooling time at 10 kpc radius of the BCG versus the 1.4 GHz luminosity of the BCG (Figure 11). The cooling time profiles come from S09, Cavagnolo et al. (2009), and our own work. We choose a radius of 10 kpc to well separate small coronae from LCCs. Mittal et al. (2009) presented a similar plot for the HIFLUGCS sample, but they used isobaric cooling time at 0.004 r₅₀₀ (2–6 kpc) and the total radio luminosity. They ignored the presence of small coronae so other mechanisms (e.g., cluster merger) were used to explain strong radio AGNs in some "noncool core" clusters that are all found to host small coronae in this work. Cavagnolo et al. (2008) presented a plot of the central entropy versus the radio luminosity. Their conclusion of the association between low entropy gas and the strong radio AGN is still consistent with ours. In Figure 11, we did not include A1367 and A754 as they are irregular clusters with the X-ray peak far from their BCGs. In the plot for the HIFLUGCS-E sample, A2163, as the most distant system (z = 0.203), was excluded.

Zoom In Zoom Out Reset image size

As shown in Figure 11, there is a general anticorrelation between the radio activity and the central gas cooling time (e.g., at 10 kpc), especially if groups are excluded. However, there are three outliers in the B55 sample and seven outliers in the HIFLUGCS-E sample. All these BCGs have small coronae. More exactly, the BCGs of A3532 and A3376 are only flagged as soft X-ray sources (or candidates of coronae; see Section 3) as the observations are not deep enough (9.5–64 ks with ACIS-I) at their redshifts (z = 0.046–0.055). As emphasized in Section 3, we are confident that most soft X-ray sources are genuine thermal coronae. At L_1.4 GHz > 10²⁴ W Hz⁻¹, the fraction of coronae increases when the cluster flux limit of the sample decreases, 3 out of 16 in the B55 sample versus 7 out of 21 in the HIFLUGCS-E sample. The fraction is 30 out of 49 (14 out of 33 for kT > 2 keV systems) in our sample, although ours is not a flux-limited sample. This is not surprising as small flux-limited samples are biased to the local luminous LCC clusters. Owing to its high flux limit, groups in the extended HIFLUGCS sample are luminous groups with LCCs and none of their BCGs has a strong radio activity (L_1.4 GHz > 10²⁴ W Hz⁻¹), while there are many such groups in our sample (Figure 1).

We present a systematic study to search for X-ray cool cores associated with 161 BCGs and 74 strong radio AGNs (L_1.4 GHz > 10²⁴ W Hz⁻¹) in 152 nearby groups and clusters (186 galaxies in total), selected from the Chandra archive, including nearly all systems in the B55 and the extended HIFLUGCS samples. The main conclusions of this work are the following.

• 1.  
  All 69 BCGs with strong radio AGNs (L_1.4 GHz > 2 × 10²³ W Hz⁻¹) have X-ray cool cores with a central isochoric cooling time of <1 Gyr (Section 4 and Figure 1). In fact, there are only two nondetections of cool cores out of 81 BCGs above L_1.4GHz of 10²³ W Hz⁻¹. Upper limits of both nondetections are high, so we claim that every BCG with an L_1.4 GHz > 10²³ W Hz⁻¹ AGN has an X-ray cool core. This conclusion also holds in the B55 and the extended HIFLUGCS samples (Section 7.4). The BCG cool cores can be divided into two classes, large (r_4Gyr > 30 kpc) and luminous (L_0.5-2 keV ⩾ 10⁴² erg s⁻¹) cool cores like Perseus cool core, or small (⩽4 kpc in radius typically) coronae like ESO 137-006 in A3627 (Section 5, Figures 6–7 and S07, also see Figures 2–3 for more examples). We call them the LCC class and the corona class. The gas of the former class is primarily of ICM origin, while the latter one is of ISM origin. For 22 non-BCGs with L_1.4 GHz > 10²⁴ W Hz⁻¹ AGN, there are only six confirmed corona detections, but most upper limits are high.
• 2.  
  We emphasize that the above result is different from the well-known result that almost every group or cluster with a strong LCC has a radio AGN (e.g., Burns 1990; Eilek 2004). Our result also shows that the traditional cool core/noncool core dichotomy is too simple. A better alternative is the cool core distribution function with an enclosed X-ray luminosity or gas mass (e.g., the histogram in Figure 1).
• 3.  
  Small coronae, easily overlooked or misidentified as X-ray AGNs at z > 0.1, are mini-cool cores in groups and clusters (Sections 4–6). They can trigger strong radio outbursts long before LCCs are formed. The triggered outbursts may destroy embryonic LCCs and thus provide another mechanism besides mergers to prevent the formation of LCCs (see Burns et al. 2008). The outbursts triggered by coronae can also inject extra entropy into the ICM and modify the ICM properties in systems without LCCs. For BCGs with strong radio AGNs in our sample, the corona fraction is at least comparable with that of LCCs.
• 4.  
  There are no groups with a luminous X-ray cool core (L_0.5-2 keV > 10^41.8 erg s⁻¹) hosting a strong radio AGN with L_1.4GHz > 10²⁴ W Hz⁻¹ (Section 4 and Figure 1). This is not observed in clusters (kT > 2 keV). The absence of low-mass systems with strong radio AGNs creates a gap between the two classes at high radio luminosities (Figure 1). Although this result needs to be examined with a larger, representative group sample (e.g., purely optically selected), it may point to a greater impact of feedback on low-mass systems than clusters. We suggest that some groups with a luminous cool core may have been transferred into the corona class in a strong radio outburst.
• 5.  
  In the LCC class, there is a general trend (albeit with large scatter) that more luminous cool cores host more luminous radio AGNs. We suspect that the environmental boosting may play a role to create the trend. BCGs in weak cool cores and noncool cores only have weak radio AGNs (Figures 1, 11 and Section 4).
• 6.  
  Only ∼16% of radio AGNs (L_1.4GHz > 10²⁴ W Hz⁻¹) have luminous X-ray AGNs (L_0.5-10 keV > 10⁴² erg s⁻¹), while the X-ray AGN fraction is even smaller for BCGs (∼4%). On the other hand, ⩾78% of strong radio AGNs have a confirmed cool core (⩾90% for BCGs), which implies their tight connection. For BCGs, all detected X-ray AGNs (L_0.5-10 keV > 10⁴² erg s⁻¹) are also a radio AGN (L_1.4GHz > 10²⁴ W Hz⁻¹; Section 4.3). Thus, a strong X-ray AGN of a BCG only emerges at the stage of its strong radio activity, occasionally.
• 7.  
  Luminous coronae may be able to power their radio AGNs through Bondi accretion (Section 7.1; and also see Hardcastle et al. 2007), while the hot accretion may not work for faint coronae in less massive galaxies. However, a complete inventory of cold gas in embedded coronae is required to address the question of the accretion mode. We note that cold ISM and dust indeed exist in some coronae with strong radio AGNs.
• 8.  
  While coronae may trigger radio AGNs, strong outbursts have to deposit little energy inside coronae to keep them intact (also emphasized in S07). Thus, it is unclear whether coronae are decoupled from the radio feedback cycle (Section 7.2). On the other hand, weak outbursts can provide gentle heating required to offset cooling in coronae. The existence of coronae around strong radio AGNs in groups and clusters also affects the properties of radio jets, e.g., extra pressure to decelerate jets and maintain the collimation of jets. Its small size also allows the bulk of the jet energy to transfer to the outskirts of the system.
• 9.  
  We also present detailed analyses on coronae associated with ESO 137-006 (in A3627) and NGC 4709 (in the Centaurus cluster), as an example of luminous coronae associated with a strong radio AGN (Section 5) and an example of faint coronae (Section 6), respectively.

We stress the importance of small X-ray cool cores like coronae. There has been a lot of attention to understand LCCs like Perseus's. However, most of the local massive galaxies (e.g., more luminous than L_*) are not in LCCs. Small X-ray cool cores like coronae may be more typical gaseous atmosphere that actually matters in the evolution of massive galaxies to explain their colors and luminosity function (see the beginning of Section 1). There has also been a lot of discussions on "bimodality" of the cluster/group gas cores (e.g., Cavagnolo et al. 2009), basically cool cores (with low entropy and power-law distribution) and noncool cores (with high entropy and flat distribution). However, one should not misunderstand the bimodality of the ICM entropy at ⩾5–10 kpc scales with the bimodality of the gas entropy at ∼ Bondi radius of the central SMBH. The formal "bimodality" has been confirmed (e.g., Cavagnolo et al. 2009), although we now know that clusters are identified as "noncool core" clusters at ⩾5–10 kpc scales can still have low-entropy gas around their BCGs. The properties of the gaseous atmosphere around the central SMBH are not necessarily correlated with the gas properties at larger scales (e.g., r⩾ 5–10 kpc).

We are very grateful to Paul Nulsen, Megan Donahue, Christine Jones, and Mark Voit for comments and suggestions on an early draft of this paper. We also want to thank Alexey Vikhlinin and Bill Forman for inspiring discussions since the discovery of embedded coronae. We also thank Craig Sarazin, Ken Cavagnolo, and Greg Sivakoff for discussions. We thank Ken Cavagnolo for providing his results on some cool core clusters. We thank Alexey Vikhlinin for providing us with the proprietary Chandra data of A3135, A3532, and 3C 88. We also thank the referee for helpful comments. This research has made 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. The financial support for this work was provided by the NASA grants GO7-8081A, GO8-9083X and the NASA LTSA grant NNG-05GD82G.