ISOTROPIC ACTIVE GALACTIC NUCLEUS HEATING WITH SMALL RADIO-QUIET BUBBLES IN THE NGC 5044 GROUP

The ACIS images in Figures 1 and 2 show that the central region of NGC 5044 is highly perturbed with many cavities, filaments, and edges. Most of the cavities in NGC 5044 are fairly small, with diameters of only a few kpc, compared to the cavities found in rich clusters, which typically have diameters of tens of kpc (McNamara & Nulsen 2007). Also, the inner cavities in NGC 5044 have a more isotropic distribution about the center of the group compared to the nearly bipolar distribution of the larger, more energetic, cavities seen in richer systems. The central X-ray morphology of NGC 5044 is more similar to that of M87, which has many small cavities and filaments (Forman et al. 2007). The largest filament identified in Figure 1 extends approximately 20 kpc toward the SE.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Beyond the central cavities and filaments, there are two edges toward the SE and one toward the NW (see Figure 2). The spectral analysis presented below (see Section 6) shows that all of these edges are cold fronts and not shocks (i.e., the gas interior to the fronts is cooler than the surrounding gas). The innermost cold front located 30 kpc toward the SE was noted by Gastaldello et al. (2008) based on the earlier Chandra observation. Another cold front is also visible toward the SE in Figure 2 with more of a spiral pattern and extends approximately 50 kpc toward the south. A corresponding spiral shaped cold front is also seen in Figure 2 toward the NW. This spiral feature is more obvious in the surface brightness analysis presented below. Cold fronts are commonly found near the central dominant galaxy in clusters and may be due to merger-induced sloshing of the central galaxy (Markevitch & Vikhlinin 2007; Ascasibar & Markevitch 2006). However, the two-arm spiral pattern in NGC 5044 is more similar to the AGN inflated "hour glass" shape feature observed in NGC 4636 (Jones et al. 2002; O'Sullivan et al. 2005; Baldi et al. 2009). The origin of the cold fronts in NGC 5044 will be discussed in more detail below. The full complexity of the X-ray morphology of NGC 5044 is best revealed in the unsharp masked image shown in Figure 3. This image shows the full extent of the X-ray cavities and the enhancements in the surface brightness around the cavities. There are also several larger and weaker depressions in the unsharp masked image beyond the cavities identified in Figure 3. It is obvious from Figure 3 that the cavities occupy a significant fraction of the total volume within the central 10 kpc.

Zoom In Zoom Out Reset image size

NGC 5044 was recently observed by the Giant Metrewave Radio Telescope (GMRT) at frequencies of 235, 325, and 610 MHz as part of a sample of groups (S. Giacintucci et al. 2009a, in preparation). Previous higher frequency observations of NGC 5044 only detected a compact radio source with a flat spectral index consistent with a core-dominated source (Sparks et al. 1984). The GMRT observation at 610 MHz shows the presence of extended emission toward the SE with a total flux of 8.5 mJy (see Figure 4). The two offset peaks in the 610 Mhz radio image visible in Figure 4 suggest that the 610 MHz radio emission arises from within a torus shaped region with the axis of the torus in the plane of the sky. This figure also shows that the cool gas in the SE filament appears to be threading the center of the torus. This behavior is very similar to that seen in numerical simulations of buoyantly rising AGN inflated bubbles which develop into torus-like structures with cool gas from the center of the cluster being dredged up in their wakes (Churazov et al. 2001; Gardini 2007; Revaz et al. 2008). This phenomenon is observed in the eastern arm of M87 (Forman et al. 2007) and in the dredging up of Hα filaments behind AGN inflated bubbles in the Perseus cluster (Hatch et al. 2006).

Zoom In Zoom Out Reset image size

The radio emission at 235 MHz is much more extended than the emission at 610 MHz and there is little overlap between the two frequencies (see Figure 5). The lack of any detected emission at 610 MHz in the same region as the 235 MHz emission indicates that the radio spectrum must be very steep with a spectral index of α ≳ 1.6. The lack of any 610 MHz emission in the 235 MHz radio lobes implies that the radio spectrum of the lobes must be fairly flat (α ≲ 1.6). There are two separate extended components in the 235 MHz observation. One of the components originates at the center of NGC 5044, passes through the southern cavity and then bends toward the west just behind the SE cold front (see Figure 6). The bending is probably due to the streaming of gas behind the SE cold front. Beyond the SE cold front, the radio emission sharply bends toward the south, possibly due to the effects of buoyancy. While the cavities in NGC 5044 have a nearly isotropic distribution, Figure 6 shows that all of the extended radio emission lies to the south and SE. Figure 6 also shows that no radio emission is detected in most of the X-ray cavities.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The GMRT observation at 235 MHz shows that there is a second detached radio lobe toward the SE (see Figures 5 and 6). The northwestern edge of the this radio lobe coincides with the SE cold front, suggesting that the relativistic material in the lobe was produced by an earlier radio outburst from NGC 5044 and that it is currently be compressed by the motion of NGC 5044 toward the SE. The GMRT data thus appears to reveal two or possibly three separate radio outbursts. The youngest outburst can be identified with the 610 MHz emission and the oldest outburst with the detached radio lobe. The GMRT observation of NGC 5044 will be discussed in more detail in Giacintucci et al. (2009b, in preparation).

The central AGN and other hard X-ray point sources are easily seen in the raw 2.0–7.0 keV ACIS image (see Figure 7). Using the CIAO tool wavdetect, 12 point sources, in addition to the central AGN, are detected within the central 4' by 4' (44 by 44 kpc) region with an unabsorbed 2.0–7.0 keV flux limit of 1.26 × 10⁻¹⁵ erg cm⁻² s⁻¹ (assuming a power-law spectrum with an index of Γ = 1.4). The hard X-ray band log N–log S relation for background point sources given by Moretti et al. (2003) predicts 7 ± 3 point sources above this flux limit within this region, so there is not a significant excess of point sources over this region. However, within the central 15 kpc, there are eight detected point sources in addition to the central AGN and the predicted background number is 2.3 ± 1. Thus, some of these point sources are probably low-mass X-ray binaries associated with NGC 5044. Assuming these eight point sources are at the distance of NGC 5044, gives 2–7 keV luminosities between 3.3 and 15.3 × 10³⁸ erg s⁻¹, which are typical luminosities for the most luminous low-mass X-ray binaries (LMXBs) in early-type galaxies (e.g., Sarazin et al. 2001; Kraft et al. 2001; Kim & Fabbiano 2004; David et al. 2006; Brassington et al. 2008).

Zoom In Zoom Out Reset image size

Emission from the NGC 5044 group was detected on all five CCDs that were turned on during the ACIS observation and data is available out to a radius of approximately 20' (220 kpc). Figure 14 shows the background-subtracted and exposure corrected surface brightness profile in the 0.3–2.0 keV energy band including data from all five CCDs. The background for each CCD was determined using the method described in Section 2. Separate exposure maps were generated for each CCD at an energy corresponding to the mean source photon energy in the 0.3–2.0 keV energy band. The central point source is evident in the innermost bin of the surface brightness profile shown in Figure 14, as are the inner cavities and a central power-law region extending out to approximately 3 kpc with a slope of ∼ 0.3. Even though there is significant substructure in the central region of NGC 5044, beyond 3 kpc, the azimuthally averaged surface brightness profile is remarkably smooth. We therefore fitted the 0.3–2.0 keV surface brightness profile beyond 3 kpc to the standard β model

$$\begin{eqnarray*} &&{ \Sigma (r) = \Sigma_0 \left[ 1 + \left({ {r} \over {r_0}} \right)^2 \right]^{-3 \beta + 0.5} } \end{eqnarray*}$$

and obtained r₀ = 9.1 ± 0.1 kpc and β = 0.541 ± 0.001 (1σ errors). The outer slope of the best-fit β model shown in Figure 14 is in good agreement with the ROSAT PSPC analysis which included emission out to 40' (β = 0.53; David et al. 2001) and the XMM-Newton PN surface brightness profile fitted by Buote et al. (2003), who obtained β = 0.52.

Zoom In Zoom Out Reset image size

We also generated background-subtracted and exposure corrected surface brightness profiles in the 0.3–2.0 keV energy band in four quadrants to a projected radius of 40 kpc, which is the full extent of the S3 chip (see Figure 15). This figure shows that there are significant azimuthal variations in the surface brightness profile within the central 30 kpc (the location of the SE cold front). Beyond the SE cold front, all four surface brightness profiles are similar. The inner cavities show up as depressions in the surface brightness profiles within the central 10 kpc. Between 10 and 30 kpc, the surface brightness profiles toward the NE and SW (perpendicular to the direction of the SE cold front) closely follow the azimuthally averaged surface brightness profile. The SE surface brightness profile is significantly higher than the azimuthally averaged value between 10 and 30 kpc, while the NW surface brightness profile is significantly lower in this region. The SE cold front is clearly visible as a break in SE surface brightness profile at approximately 30 kpc. There is also a break in the NW surface brightness profile at 30 kpc in the opposite sense (i.e., shallow inner surface brightness and steep outer surface brightness). The azimuthal variations in the surface brightness profiles between 10 and 30 kpc are probably due to the sloshing motion of NGC 5044 toward the SE.

Zoom In Zoom Out Reset image size

To further search for any features in the ACIS image of NGC 5044, we computed surface brightness profiles within several concentric annuli as a function of position angle (see Figure 16). The SE spiral shows up as a spike in the surface brightness profiles at a position angle of approximately 140° and is clearly evident to the edge of the S3 chip. There is also a smaller amplitude spike in all four surface brightness profiles shown in Figure 16 at a position angle of approximately 325°. This NW spiral is only marginally visible in the raw image shown in Figure 2, but is quite evident in Figure 16.

Zoom In Zoom Out Reset image size

From our moderately deep Chandra observation, we are able to derive a high spatial resolution temperature profile of NGC 5044 (see Figure 17). The temperatures derived from a single thermal plasma model and a thermal plus power-law model are fully consistent (compare the black and red data points in Figure 17). Within the 0.5–5.0 keV energy band, 85% of the detected photons have energies below 1.5 keV for 1 keV gas, so the spectral fits are heavily weighted by the soft thermal emission and are not significantly affected by the harder emission from unresolved LMXBs. The overall temperature profile shown in Figure 17 is similar to that observed in other groups with a positive temperature gradient at small radii, a peak temperature at approximately 50 kpc, and a declining temperature profile at larger radii (e.g., Finoguenov et al. 2007; E. O'Sullivan et al. 2009, in preparation). Within the central 10 kpc, the azimuthally averaged temperature is nearly isothermal with temperatures between 0.70 and 0.85 keV (consistent with the temperature map shown in Figure 9). There are two sharp temperature jumps at 20 and 30 kpc seen in Figure 17. The outer temperature jump corresponds to the SE cold front. The inner temperature jump is primarily due to the non-azimuthal structure of the central gas temperature (see the temperature map in Figure 9 and Section 6.2). Within 20 kpc, the spherical annuli only contain emission from the coolest gas. Beyond this radius, the spherical annuli also contain some emission from hotter gas. The reduced χ² values for the spectral fits are shown in Figure 18. While the inclusion of a power-law component in the spectral analysis does not affect the derived gas temperatures, it does significantly reduce the χ² value in most spectra. For example, applying a F-test on the spectral fitting results of the innermost five annuli shows that the improvement in the fits is significant at more than 98% in each case. Figure 18 shows that single temperature fits are reasonably acceptable beyond 50 kpc, which corresponds to the peak of the temperature profile and the extent of the cool SE spiral. The degradation in the spectral fits within this region is primarily due to projection effects, azimuthal variations in the gas temperature and the presence of multi-phase gas, as can be seen in the temperature map shown in Figure 9.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We also fitted the azimuthally averaged spectra to an absorbed thermal plasma model in three additional energy bands to determine the influence of the emission from unresolved LMXBs: (1) a soft energy band from 0.5–1.5 keV which contains mostly Fe–L emission lines plus some weaker emission from O, Ne, and Mg; (2) an energy band from 1.5–2.2 keV, which contains emission lines from He-like and H-like Si; and (3) a harder energy band from 2.2–5.0 keV, which contains mostly continuum emission plus emission lines from S, Ar, and Ca (see Figure 19). For the soft energy band we used a set of spectra with 20,000 net counts each and for the harder energy bands we used a set of spectra with 40,000 net counts each. Since most of the photons in the 0.5–5.0 keV energy band have energies below 1.5 keV, the temperature profile derived in the broad and soft energy bands are in good agreement. The temperature profiles derived from the Si line ratio and the hard energy band also agree with the temperature profile derived from the broad and soft energy bands beyond 30 kpc. The influence of the LMXBs on the derived gas temperatures is most notable in the hard energy band within the central 30 kpc.

Zoom In Zoom Out Reset image size

To unambiguously determine if the harder spectral component in the center of NGC 5044 arises from an extra component of hotter gas or LMXBs, we fitted the 1.5–7.0 keV emission within the central 8 kpc to a single temperature model, a two-temperature model (with the abundances linked) and a single temperature plus power-law model (see the results in Table 1). As can be seen in Table 1, the thermal plus power-law model provides the best fit to the emission within the central 8 kpc. In addition, the best-fit temperature is consistent with the central gas temperature derived above and the best-fit power-law index is consistent with the characteristic power-law emission from LMXBs. The unabsorbed 2.0–7.0 keV flux of the power-law component is F(2–7 keV) = 6.6 × 10⁻¹⁴ erg cm⁻² s⁻¹. Combining this with the 2.0–7.0 keV flux of the resolved point sources within the central 8 kpc gives a total 2.0–7.0 keV flux and luminosity for the LMXBs of F(2–7 keV) = 7.5 × 10⁻¹⁴ erg cm⁻² s⁻¹ and L(2–7 keV) = 4.5 × 10³⁹ erg s⁻¹. The 2.0–7.0 keV flux from the LMXBs within the central 8 kpc is 17% of the total flux from the galaxy. In David et al. (2006), we showed that there is a strong correlation between the total 0.5–2.0 keV luminosity of the LMXBs in early-type galaxies and the K-band luminosity of the galaxies. An analysis of the 2MASS K-band image of NGC 5044 gives m_K = 8.00 and L_K = 1.96 × 10¹¹ L_K☉ within the central 8 kpc. Using Equation (1) in David et al. (2006), we find a predicted luminosity of L(0.5–2.0 keV) = 4.1 × 10³⁹ erg s⁻¹. A power-law spectral model with Γ = 1.4 gives L(2–7 keV) = 8.1 × 10³⁹, which is consistent with the observed 2–7 keV luminosity, given the observed scatter in the relationship. This analysis shows that the harder X-ray component in the center of NGC 5044 is most likely due to unresolved LMXBs and not a second component of hotter gas. In general, since the X-ray emitting gas has a more extended distribution than the stars and LMXBs in early-type galaxies and the gas is the coolest in the center, the harder X-ray emission from LMXBs should be the most prevalent in the center of these galaxies.

The projected temperature profiles within four quadrants out to 40 kpc (the field of view of the S3 chip) are shown in Figure 20. The SE cold front shows up as a sharp temperature jump in the SE temperature profile at a radius of 30 kpc. Interior to the SE cold front, the gas is essentially isothermal. The NW temperature profile shows a temperature jump at 20 kpc which coincides with the break in the NW surface brightness profile (see Figure 15). There is a also a significant temperature jump in the SW temperature profile at a radius of 20 kpc, but there is no obvious break in the surface brightness at this location (see Figure 15). This temperature jump is likely due to non-azimuthal variations in the gas temperature within the SW quadrant. A further analysis of the temperature map shows that the temperature varies by 15% at a radius of 20 kpc within the SW quadrant (i.e., between position angles of 180° and 270°). There are also weaker temperature jumps in the NE and NW directions at approximately 14 kpc. These jumps are also probably due to non-azimuthal temperature variations within these quadrants.

Zoom In Zoom Out Reset image size

The presence of cool gas in the SE and NW spirals can best be shown by plotting the azimuthal variations in gas temperature within concentric annuli between 20 and 40 kpc from the center of NGC 5044 (see Figure 21). This figure shows two sharp temperature drops at position angles of 135° and 315°, which are coincident with the SE and NW spirals seen in the surface brightness profiles shown in Figure 16. Within the inner annulus, there is also a broader dip in gas temperature between P.A. = 0 and P.A. = 270°, due to the greater extent of cool gas in the SE direction (see the temperature map in Figure 9).

Zoom In Zoom Out Reset image size

To deduce the spectral properties of the AGN in NGC 5044, we extracted a spectrum from within a 12 (220 pc) radius region centered on the AGN. Table 2 shows that the spectrum is poorly fitted with a thermal or power-law model, but is well fitted with a thermal plus power-law model. The best-fit temperature in the thermal plus power-law model is consistent with the central gas temperature shown in the temperature map (see Figure 9) and temperature profile (see Figure 17). We also allowed the hydrogen column density to vary in the fitting process, but there is no evidence for excess absorption above the galactic value. The best-fit power-law index in the two component model is also quite typical of AGN. Based on the best-fit thermal plus power-law model, the unabsorbed 0.5–5.0 keV flux and luminosity of the AGN are 1.98 × 10⁻¹⁴ erg s⁻¹ cm⁻² and 3.57 × 10³⁹ erg s⁻¹ and the bolometric X-ray luminosity is 6.64 × 10³⁹ erg s⁻¹. The expected bolometric X-ray luminosity of the LMXBs based on the K-band luminosity of the galaxy within the central 220 pc is 3.3 × 10³⁷ erg s⁻¹, which is less than 1% of the total luminosity of the power-law component within this region. The central radio point source was detected in our GMRT observations at frequencies of 235 MHz, 327 MHz, and 610 MHz, with flux densities of 35 ± 2 mJy, 32 ± 2 mJy, and 30 ± 2 mJy, respectively (S. Giacintucci et al. 2009a, in preparation). Integrating the radio flux density between 5 MHz and 10 GHz, assuming S_ν ∝ ν^−α with α = 0.16 gives a radio luminosity of L_R = 4.1 × 10³⁸ erg s⁻¹. This shows that the radiative efficiency of the AGN in the radio band is only 6% of that in the X-ray band. Using a stellar velocity dispersion of σ_* = 237 km s⁻¹ for NGC 5044 (which is the average of 4 measurements listed in HyperLeda) and the relation between stellar velocity dispersion and central black hole mass in Gebhardt et al. (2000), gives a central black hole mass of M_bh = 2.27 × 10⁸ M_☉. This gives a ratio between the bolometric and Eddington luminosity of L_bol/L_Edd = 2.3 × 10⁻⁷.

The bolometric X-ray luminosity of the gas within the central 220 pc is 1.56 × 10³⁹ erg s⁻¹ based on the best-fit thermal plus power-law spectral model. Using the best-fit emission measure of the thermal component, assuming a uniform gas density within the central 220 pc, gives a density of n_e = 0.21 cm⁻³ and a total gas mass of M_gas = 2.64 × 10⁵ M_☉. The isobaric cooling time of the gas is t_c = 5kTM_gas/(2μm_pL_bol) = 3.5 × 10⁷ yr. Alternatively, if we assume the gas density follows a power-law profile given by n_e ∝ r^−0.8, consistent with the density distribution determined from a deprojected spectral analysis (L. David et al. 2009, in preparation), we obtain n_e = 0.14(r/r₀)^−0.8 cm⁻³, with r₀ = 220 kpc.

The Bondi accretion radius of the central black hole is R_a = GM_bh/c²_a, where c_a is the adiabatic sound speed at the accretion radius. Using a gas temperature of 0.75 keV, gives R_a = 4.9 pc. The Bondi accretion rate is given by ${\dot{M}_B = 4 \pi R_a^2 \rho (R_a) c_a}$. The greatest uncertainty in this expression is the gas density at the Bondi radius. Even in the nearby NGC 5044 group, an extrapolation over a factor of 40 in radius is required to estimate the density at R_a. If the gas is isothermal and the density follows the same power-law expression given above, then ${\dot{M}_B = 0.011\, {M}_{\odot }}$ yr⁻¹. If the X-ray luminosity of the AGN is powered by Bondi accretion, the resulting efficiency $({\epsilon = L_{\rm bol}/(\dot{M}_B c^2})$ is 1.1 × 10⁻⁵, which is typical of the radiatively inefficient AGN in early-type galaxies (Loewenstein et al. 2001; Di Matteo et al. 2003; David et al. 2005).

Spectra were extracted from the cavities and filaments using the regions shown in Figure 22 and fitted to absorbed apec and vapec models in the 0.5–5.0 keV energy band. A power-law component was added in both cases to account for the emission from unresolved X-ray binaries with the power-law index frozen at Γ = 1.4. Table 3 shows that the apec model provides a poor fit to the spectra extracted from both the cavities and filaments. A significant improvement is obtained with the vapec model indicating that either the abundance ratios differ from those in the abundance table of Grevesse & Sauval (1998), or the gas is multi-phase (especially as seen in projection), and that different lines are being excited by different temperature gas. However, as is also evident from the temperature map (in which the gas temperature is derived solely from the centroid of the Fe–L lines), the cavities have higher temperatures than the filaments, but only by 0.05–0.1 keV. Both the temperature map and temperature profile show that the gas is nearly isothermal within the central 10 kpc. Most of the cavities are only a few kpc in extent, so the gas that was displaced likely had a temperature similar to the surrounding gas seen in projection. The coolest gas within the central region of NGC 5044 is located within the SW filament with kT = 0.65 keV (see Table 3). This is also evident in the temperature map derived from the earlier 20 ks ACIS observation presented by Gastaldello et al. (2008).

Zoom In Zoom Out Reset image size

We also searched for evidence of hot thermal gas entrained within the cavities by extracting a spectrum of the combined emission from all four cavities shown in Figure 22. The extracted spectrum shows no evidence for any Fe–Kα emission, which would be an unambiguous sign of shocked hot gas in a group of galaxies. The net count rate between 5 and 7 keV is actually consistent with zero with a 3σ upper limit of 1.1 × 10⁻⁴ ct s⁻¹. If we scale the expected emission from X-ray binaries according to the K-band luminosity of the galaxy within the four cavities we find a predicted 5–7 keV count rate from X-ray binaries of 1.0 × 10⁻⁴ counts s⁻¹, which is just below our 3σ upper limit. Since the projected cavity spectrum contains thermal emission from gas with temperatures between 0.8 and 1.4 keV, and since we are only interested in detecting gas hotter than that encountered along the line of sight, we fitted the spectrum in the 2.5–5.0 keV energy band to an absorbed vapec plus power-law model (to account for the emission from unresolved LMXBs) and obtained a best-fit temperature of kT = 0.90 ± 0.30 keV (90% errors), which is consistent with emission from ambient gas along the line-of-sight.

To place constraints on the amount of hot gas entrained within the cavities, we computed the predicted count rate in the 5–7 keV energy band for gas in pressure equilibrium with the surrounding medium assuming a solar abundance of Fe (see Table 4). This table shows that the predicted count rates for gas hotter than 2 keV are all well below the 3σ upper limit and the count rate expected from unresolved LMXBs. Thus, we cannot exclude the presence of gas with temperatures as low as 2 keV in pressure equilibrium within these cavities.

NGC 5044 hosts many small X-ray cavities with a nearly isotropic distribution and one larger cavity toward the south. The smaller cavities are all radio quiet, even at low frequencies (sensitivity level of 0.25 mJy/b at 235 MHz; Giacintucci et al. 2009b, in preparation). This suggests that these cavities are no longer powered by the central AGN and are primarily driven by buoyancy. The isotropic distribution of the small, buoyantly driven cavities may be due to group "weather" as in the simulations of Bruggen et al. (2005) and Heinz et al. (2006). The origin of the group weather could be due to previous AGN outbursts or the sloshing of NGC 5044 with respect to the center of the group potential.

The enthalpy in a cavity is given by H = γPV/(γ − 1) and varies between 2.5 PV and 4 PV for γ = 5/3 and γ = 4/3. There are many uncertainties in estimating the ages of a cavities and McNamara & Nulsen (2007) list three possible methods for computing their ages, including: the sound crossing time, the buoyancy time and the refill time. We computed the three different age estimates for each cavity and found that the age estimates agree to within 30%. As a reasonable approximation, we compute the cavity power as P_c = 4PV/t_s, where t_s is the sound crossing time between the AGN and the center of the cavity. This gives a total cavity power for the three small radio-quiet cavities (i.e., the SW, NW, and NE cavities) of 9.2 × 10⁴¹ erg s⁻¹. The cavity power of the larger southern cavity is 5.6 × 10⁴² erg s⁻¹. For comparison, the bolometric luminosity of the hot gas within the central 10, 20 and 30 kpc is 2.1 × 10⁴² erg s⁻¹, 4.7 × 10⁴² erg s⁻¹, and 7.1 × 10⁴² erg s⁻¹, respectively. This shows that the cavity power of the three smaller cavities, currently located between 4 and 7 kpc from the center of NGC 5044, can balance about one-half of the radiative losses within the central 10 kpc. However, the combined cavity power of the three small cavities plus the larger southern cavity is sufficient to balance all radiative losses within the central 25 kpc. As discussed below, the presence of Hα emitting gas within the central 5 kpc shows that as least some of the hot gas is able to cool to low temperatures.

Studies have shown that the mechanical cavity power far exceeds the radio luminosity of the X-ray cavities in groups and clusters of galaxies and that the ratio of cavity power to radio luminosity increases with decreasing radio luminosity (Birzan et al. 2004, 2008). In this paper we concentrate on the properties of the radio lobe toward the SE observed at 610 MHz. The analysis of the extended radio emission at 235 MHz within the southern cavity is more complicated due to the larger beam size at 235 MHz (22'' by 16'') and the 20'' separation between the southern X-ray cavity and the central AGN. The GMRT data at 235 MHz will be discussed in more detail in S. Giacintucci et al. 2009b, in preparation).

The monochromatic 610 MHz luminosity of the SE radio lobe in NGC 5044 is L₆₁₀ = 9.3 × 10³⁶ erg s⁻¹ (S. Giacintucci et al. 2009a, in preparation). Assuming a spectral index of α = 1 and integrating between 10 MHz and 5 GHz (as in Birzan et al. 2004) gives a radio luminosity of L_R = 5.7 × 10³⁷ erg s⁻¹, which is a factor of approximately 100 less than the least luminous radio filled cavity studied in the Birzan et al. (2004) sample. The ratio of cavity power to radio luminosity, assuming the cavity is the same size as the radio lobe, is P_c/L_R = 1.3 × 10³ for the SE radio lobe in NGC 5044. Using the relation between cavity power and radio luminosity (Equation (12) in Birzan et al. 2008) along with the observed radio luminosity of the SE radio lobe, gives a predicted cavity power of P_c = 4.0 × 10⁴¹, which is only 40% less than the observed value. While this is only one extra data point, it shows that there is some evidence that the relation between P_c and L_R derived in Birzan et al. (2008) extends over 8 orders of magnitude in L_R.

We showed above that we cannot place very tight constraints on the amount of hot thermal gas within the cavities in NGC 5044 due to the hard X-ray emission from unresolved LMXBs. Constraints on the particle content within the radio lobes observed in groups and clusters can be obtained by assuming the lobes are in pressure equilibrium with the ambient gas (Dunn & Fabian 2004; Birzan et al. 2008; Croston et al. 2008). The total particle energy within a radio lobe can be written as E_p = kE_e, where E_e is the energy in relativistic electrons emitting synchrotron radiation at frequencies greater than 10 MHz. Using the observed 610 MHz radio luminosity from the SE radio lobe along with the expression for E_e in Pacholczyk (1970), assuming α = 1, gives

$$\begin{eqnarray*} &&{ E_p = 6.0 \times 10^{54} \left({ {B} \over {1 \;\mu {\rm G}}} \right) k \;\rm {erg}. } \end{eqnarray*}$$

The energy in the magnetic field is E_B = B²ϕV/(8π), where V is the volume of the radio lobe and ϕ is the volume filling factor. For the SE radio lobe, this gives

$$\begin{eqnarray*} &&{ E_B = 1.4 \times 10^{54} \left({ {B} \over {1 \;\mu {\rm G}}} \right)^2 \phi \;\rm {erg.} } \end{eqnarray*}$$

Assuming energy equipartition then gives

$$\begin{eqnarray*} &&{ B_{eq} = 1.5 \left({ {k} \over {\phi }} \right) \;\mu {\rm G}. } \end{eqnarray*}$$

Combining the assumptions of energy equipartition and pressure equilibrium between the ambient medium and the radio lobe (P_T = kE_e/(3Vϕ) + E_B/(Vϕ)) gives k/ϕ = 1.2 × 10⁴ and B_eq = 23 μG.

Performing a similar analysis, Dunn & Fabian (2004) and Birzan et al. (2008) obtained values of k/ϕ up to approximately 4000 based on samples of radio lobes in mostly rich clusters of galaxies. This comparison shows that the non-radiating energy content of the SE radio lobe in NGC 5044 is significantly greater than that in the larger radio lobes analyzed in these two samples. Birzan et al. (2008) found a correlation between k/ϕ and the synchrotron age of the radio lobe. Based on an analysis of radio lobes in a sample of groups of galaxies, Croston et al. (2008) noted that "plumed" radio lobes were more under-pressurized than radio lobes connected with a jet. Both of these results suggests that k/ϕ increases with time after the radio lobes are no longer powered by the central AGN. The most likely origins for this increase in k/ϕ are the entrainment of thermal gas and aging of the relativistic electron population. While we cannot estimate the synchrotron age for the SE radio lobe without knowing the break frequency in the radio spectrum, the sound crossing time between the AGN and the SE lobe is only 2 × 10⁷ yr, so it must be fairly young. However, the torus-like structure of the SE radio lobe and the fact that a cool filament of gas is currently threading the torus suggest that it may have already entrained a significant amount of ambient gas.

Most of the Hα emission from NGC 5044 is contained in a broad filament aligned in a north to south direction (Macchetto et al. 1996). The northern portion of the Hα filament has an extent of approximately 7 kpc and lies along the eastern rim of the NW cavity. The southern portion of the filament trails the southern cavity, but unfortunately, a "ghost image mutilated the Hα+[N ii] emission map" beyond approximately 5 kpc, which is only half of the distance to the center of the SE radio torus. Temi et al. (2007) showed that there is some evidence for distributed polycyclic aromatic hydrocarbon (PAH) emission with the same morphology as the southern extension of the Hα filament from the Spitzer observation of NGC 5044. Spatial correlations between the Hα emission, PAH emission and X-ray morphology were also noted by Gastaldello et al. (2008) based on the earlier 20 ks Chandra observation.

Caon et al. (2000) obtained long-slit spectroscopy at three position angles across the Hα filament. The gas in the northern extension of the filament is mostly redshifted with respect to NGC 5044 with a peak velocity of approximately 100 km s⁻¹. The kinematics of the gas in the southern extension of the Hα filament are more chaotic with red and blue shifted velocities between ±100 km s⁻¹. Without knowing the inclination of the Hα filament relative to the plane of the sky, we cannot say which gas is in-falling and which gas is outflowing. However, the kinematics of the gas in the Hα filament are very similar to the kinematics of the Hα filaments in NGC 1275 observed by Hatch et al. (2006) and are probably due to dredge up and subsequent infall behind the buoyantly rising cavities. Temi et al. (2007) also argued that the extended dust distribution, as highlighted by the PAH emission, could be due to the break up of a central dusty disk followed by buoyant uplifting behind the southern X-ray cavity.

The kinematics of the warm 10⁴ K gas are inconsistent with an external origin for the gas (Caon et al. 2000), so the gas could arise from either stellar mass loss or radiative cooling of the hot ambient gas. The total mass of Hα emitting gas is 8.5 × 10⁵ M_☉ (Macchetto et al. 1996). From the Hα image, it appears that most of the warm gas originates from within the central 2 kpc, and is then dredged outward. Fitting a de Vaucouleurs profile to the 2MASS K-band image of NGC 5044 gives a deprojected K-band luminosity within the central 2 kpc of L_K = 6.5 × 10⁹ L_☉ (L. David et al. 2009, in preparation). Using the stellar mass loss rate in Athey et al. (2002) gives $\dot{M}_* = 0.012 {\;M}_{\odot }$ yr⁻¹ within this region. The cooling time of the hot gas within the central 2 kpc is 4.1 × 10⁷ yr and the mass cooling rate is approximately $\dot{M}_h = 0.9 {\;M}_{\odot }$ yr⁻¹. Thus, even if only 2% of the hot gas is able to cool to low temperatures, radiative cooling of the hot gas would still be the dominant mass supply mechanism for the warm gas in NGC 5044.

NGC 5044 has two semi-circular cold fronts at projected distances of 30 and 50 kpc from the center of the group. Cold fronts are commonly found near the central dominant galaxy in clusters and may be due to merger induced sloshing (Ascasibar & Markevitch 2006; Markevitch & Vikhlinin 2007). Based on an optical study of the galaxies in the NGC 5044 group, Cellone & Buzzoni (2005) found that NGC 5044 has a peculiar velocity of approximately 140 km s⁻¹ with respect to the velocity centroid of the other galaxies in the group. One possible candidate for perturbing NGC 5044 is a sub-group of galaxies toward the NE identified by Mendel et al. (2008), but this group lies at a projected distance of 1.5 Mpc. Another possible candidate is the optically perturbed spiral galaxy NGC 5054 which is located at a projected distance of 300 kpc toward the SE, has a blue-shifted velocity of 960 km s⁻¹ and is only two magnitudes fainter than NGC 5044.

Single spiral features have been observed in the X-ray images of the Ophiuchus cluster (Sanders & Fabian 2002), the Perseus cluster (Churazov et al. 2000; Fabian et al. 2006) and A2029 (Clarke et al. 2004). Ascasibar & Markevitch (2006) have shown through numerical simulations that single spiral arms of cool gas are commonly produced by sloshing of the central galaxy, however, none of their simulations produce two spiral arms like those observed in NGC 5044. The spiral arms in NGC 5044 are more similar to the spiral arms (or "hour glass" shaped feature) in NGC 4636 (Jones et al. 2002; Baldi et al. 2009). X-ray spectral analysis of the gas in the spiral arms in NGC 4636 shows that this gas was recently shocked, probably during the inflation of the radio emitting plasma currently observed between the spiral arms (Jones et al. 2002; Baldi et al. 2009). The spiral arms in NGC 5044 are cooler than the surrounding gas. Also, no radio emission is detected at either 235 MHz or 610 MHz interior to the spiral arms like that observed in NGC 4636. Thus, if the spiral arms in NGC 5044 were created by the inflation of radio bubbles, this must have occurred during an earlier radio outburst. This would also give the gas in the spiral arms time to cool by adiabatic expansion if the gas were initially shocked.