SLIDING NOT SLOSHING IN A3744: THE INFLUENCE OF RADIO GALAXIES NGC 7018 AND 7016 ON CLUSTER GAS

We made a 75 ks observation of the system in full-window and VFAINT data mode with the Advanced CCD Imaging Spectrometer (ACIS) on board Chandra on 2010 September 11 (ObsID 12241). NGC 7016 was positioned close to the nominal aimpoint of the front-illuminated I3 chip. The other three CCDs of ACIS-I and the S2 chip of ACIS-S were also on during the observations, giving a frame time of 3.14 s. Details of the instrument and its modes of operation can be found in the Chandra Proposers' Observatory Guide.³ Results presented here use ciao v4.5 and the caldb v4.5.6 calibration database. We re-calibrated the data to take advantage of the sub-pixel event reposition routine (edser), following the software threads from the Chandra X-ray Center (CXC),⁴ to make new level 2 events files. Only events with grades 0,2,3,4,6 were retained. After screening to exclude intervals of high background at a threshold appropriate for use of the blank-sky background files, the calibrated dataset has an exposure time of 71.553 ks.

Since the cluster fills a large part of the detector array, background was measured from blank-sky fields following procedures described in the CXC software threads. After cleaning the background data using the same criteria as for the source data, and reprojecting to the same coordinate system, a small normalization correction was applied (2%) so that the count rates matched in the 9.5–12-keV energy band where particle background dominates.

The ciao wavdetect task was used to find point sources with a threshold set to give 1 spurious source per field. Their regions were subsequently masked from the data for the analysis of extended structure. All spectral fits are performed in xspec on binned spectra using χ² statistics over the energy range 0.4–7 keV. The models include Galactic absorption of N_H = 5.27 × 10²⁰ cm⁻² (Dickey & Lockman 1990). Parameter uncertainties are 90% confidence for one interesting parameter unless otherwise stated.

We made sensitive, high-resolution, observations of the field containing NGC 7016 and NGC 7018 using the JVLA in its A configuration at L (1.4 GHz) and C (5 GHz) bands (Table 1). At the time of the observations, only part of the full bandwidth of the JVLA correlator was available. The data were calibrated and flagged for extensive interference before being passed through the normal clean and gain self-calibration cycles in casa. For more diffuse structures we also downloaded archival L-band data taken in C configuration with the VLA, and mapped them using standard procedures in AIPS. Details of the heritage of the data sets and properties of the resulting maps are given in Table 1. For L to C-band spectral-index measurements we made a version of the C-band map with the same restoring beam as the L-band JVLA map.

Despite the lack of spherical symmetry, we have characterized the overall extent of the cluster atmosphere by fitting a β-model profile to a background-subtracted exposure-corrected radial profile centered on the position $21^{\rm h} 07^{\rm m} 24^{\rm s}\llap{.}7, -25^\circ 27^{\prime } 25^{\prime \prime }$ (the approximate centroid of the diffuse emission contained within a circle of radius 227 arcsec) shown as a cross in Figure 2. The result is shown in Figure 3. We note the core radius of 220 arcsec inferred by Bicknell et al. (1990) based on earlier, less sensitive, ROSAT data falls within the 1σ error bound for β ≈ 0.56.

Zoom In Zoom Out Reset image size

We have fitted the spectrum of the brightest extended emission in Figure 2 to a single-temperature thermal (APEC) model absorbed by gas in the line of sight. Emission from the three galaxy/group atmospheres has been excluded. We find very similar results for an on-source circle of radius 227 arcsec centered on $21^{\rm h} 07^{\rm m} 24^{\rm s}\llap{.}7, -25^\circ 27^{\prime } 25^{\prime \prime }$ as for a polygon of similar area tracing better a contour of constant surface brightness. The spectral contours for the circular extraction region and two outer annuli (radii 227 and 300 arcsec, and 300 and 960 arcsec) are shown in Figure 4. There is a weak trend for the best-fit temperature to decrease with increasing radius. The 90% confidence contour for the 227 arcsec radius circle and 1σ contour for the outer annulus do not touch, giving less than 3% probability of the spectral parameters being the same in these regions. The measurable difference is only in temperature, with the abundances consistent with a constant value of roughly 0.3 times solar (i.e., Z/Z_☉ = 0.3).

Zoom In Zoom Out Reset image size

Beyond a radius of 330 arcsec some of the cluster lies off the ACIS CCDs. We have therefore estimated the total cluster luminosity using a spectral fit to the emission within this circle, with point sources and galaxy emission excluded, and have extrapolated to account for missing flux based on the radial profile. The spectrum gives an acceptable fit to a model with $kT=3.50^{+0.16}_{-0.13}$ keV (1σ error, see also Figure 4). The radial profile (Figure 3) finds that the counts within 330 arcsec should be multiplied by a factor of 2.39  ± 0.09 to account for emission out to r₅₀₀, corresponding to θ ≈ 1200 arcsec for the measured temperature based on Vikhlinin et al. (2006). The result is a bolometric luminosity for the cluster gas of (3.2 ± 0.2) × 10⁴³ erg s⁻¹ (1σ error). The luminosity is too low to be a good match to the luminosity-temperature relations of clusters—see, e.g., Figure 5 of Giles et al. (2012)—or, in other words, the temperature is too hot by kΔT ≈ 1.5 keV for the luminosity. From the sample of Maughan et al. (2012) at 0.1 < z < 1.3, the cluster with X-ray properties most closely resembling those of A3744 is RXJ2247+0337 at z = 0.2, with $kT = 2.7^{+0.7}_{-0.5}$ keV and a luminosity of (4 ± 1) × 10⁴³ erg s⁻¹ (1σ errors). That cluster is described as being neither relaxed nor having a cool core—an appropriate description also of A3744.

A3744 appears as RXC J2107.2-2526 in the REFLEX cluster catalog (Böhringer et al. 2004). The 0.1–2.4 keV cluster luminosity of 1.8 × 10⁴³ erg s⁻¹ is based on 100 detected counts and used an iterative method with an embedded temperature–luminosity relationship which will have settled on a lower temperature than now measured. A better-constrained value for the 0.1–2.4 keV luminosity to r₅₀₀ based on the Chandra data (calculated as above) is (1.81 ± 0.08) × 10⁴³ erg s⁻¹.

We find that the total gas mass out to θ = 1200 arcsec is (1.41 ± 0.09) × 10¹³ M_☉, and under the assumption of isothermal gas in hydrostatic equilibrium we can use Equations (30) and (32) of Worrall & Birkinshaw (2006) to estimate a total mass within this radius of (1.94 ± 0.15) × 10¹⁴ M_☉. The relatively low gas-mass fraction is driven by the unexpectedly high temperature that is measured (total mass is proportional to temperature while the gas mass is largely independent of it). The central proton number density in the X-ray-emitting gas is relatively low, at about 940 m⁻³, which from Figure 5 of Worrall & Birkinshaw (2006) we see corresponds to a long cooling time of about 48 billion years.

The gas morphology indicates a cavity encompassing the filament of NGC 7018 and the swirl of NGC 7016, and so temperature structure in the inner regions might be expected. However, here we find no obvious statistically significant temperature structure and no cool core. In Figure 5 we mark regions selected based on X-ray and radio morphology that lie within the 227 arcsec radius circle and that we refer to as the "arm" and "cavity," and in Figure 6 we show their spectral contours for temperature and abundance.⁵ The axes of Figure 6 are the same as in Figure 4.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The lower average X-ray surface brightness toward the cavity with respect to the arm (a factor of roughly 0.6) is difficult to understand for gas of constant temperature. The apparent opening to the SW seen in Figure 2 is very well aligned with the run of the chip gap, and so, despite exposure corrections having been applied, it is not possible to conclude that the region is unbounded in that direction. We have therefore investigated what scale of spherical cavity is consistent with the data. To do this we have extracted a radial profile centered roughly on the cavity, at $21^{\rm h} 07^{\rm m} 16^{\rm s}\llap{.}8, -25^\circ 26^{\prime } 52^{\prime \prime }$. As shown in Figure 7, the central data are clearly depressed relative to the extrapolation of the best-fit β model fitted at angular radii larger than 120 arcsec (dashed line).

Zoom In Zoom Out Reset image size

Since the angular scale of the counts deficit lies within the core radius, we have tested the simple model of a central evacuated spherical region by subtracting the profile of counts that would otherwise lie there (normalized to the central volume density of the β model) from the profile of the β model. Figure 7 shows that such a central spherical evacuated region is consistent with observations if of angular radius roughly 100 arcsec.

It is not obvious how to construct a static cavity of this type in a gas of constant pressure. It would either need to be a dynamical effect, or there needs to be a source of additional pressure in the cavity. It is likely that extra pressure is provided by radio-emitting relativistic particles (note from Figure 5 that radio emission covers the cavity but not the arm), in which case a power-law X-ray component from inverse-Compton scattering (discussed in Section 7.1) might be detectable. To test how well a power law can be accommodated within the region where the radio emission is brightest, we have defined a further region, extending that covered by the cavity and guided by the radio contours. A spectral fit to either a single-component thermal or power-law model is acceptable, at χ²/dof = 105.1/94 and 106.7/95, respectively. When the two models are combined and the power-law slope is fixed to α = 0.9, we find χ²/dof = 97.8/93 and the flux contribution of the power law is $61^{+34}_{-37}$% (90% confidence). While we would not claim this as strong evidence for a power-law component, it is consistent with a contribution of inverse Compton emission, as discussed further in Section 7.1.

The residuals in the radial-profile fit (Figure 3) show structure at radii between 100 and 400 arcsec. Profiles of different pie slices were made, from which it became apparent that the main feature was an X-ray excess that lies between NGC 7018's tendrils of 1.5 GHz radio emission (Figure 5) outside the gas cavity. We refer to this gas feature as the "plateau" and it is marked in Figure 5. Radial profiles for specific pie slices (Figure 8) show excess counts in the profile corresponding to the position angles of the plateau.

Zoom In Zoom Out Reset image size

After identifying the plateau as a distinct morphological structure, we investigated the spectral properties of the gas on these larger scales in more detail. While the gas in the plateau fits a temperature similar to the average for the cluster, gas to the east at radii beyond 227 arcsec was found to be cooler, and that to both sides between the plateau and E segment was found to be hotter. In particular, the temperature of gas in the E segment (Figure 8) is cooler than that from the combined N and SW segments at high significance (Figure 9).

Zoom In Zoom Out Reset image size

To probe further the locations of temperature changes, we have used the contbin⁶ software of Sanders (2006) to define spectral regions from our adaptively smoothed image, and map xspec model-fitting results onto the image. We used a constant signal to noise (S/N) based on our adaptively smoothed exposure-corrected (but not background subtracted) map, which provided regions for spectral fitting with a S/N ranging between 21 and 40. The S/N was on purpose chosen to be slightly lower than that in the regions of Figure 9, in order to investigate to what extent the sector boundaries of Figures 8 and 9 were supported by a less subjective examination of the data. The fitted spectral model is a single-temperature thermal, and we show results for kT and metallicity in Figure 10.

Zoom In Zoom Out Reset image size

Spectral results are listed by region in Table 2. The fits to the individual regions are acceptable. When the outer regions of most extreme temperature (3, 6, 9, and 11) are combined, we find an unacceptable fit to a single-temperature model, as expected from Figure 9 which shows that gas in the E separates in temperature from that in the N and SW. Our main conclusions are that gas to the E is cool (kT about 2.1 keV) and of metallicity less than about 0.36. Gas within a radius of 227 arcsec and in the plateau is of intermediate temperature (kT about 3.6 keV) and average cluster metallicity, except in a region corresponding to the brightest radio emission from NGC 7017 and NGC 7018, where a possibly depressed abundance might be due to dilution with non-thermal power-law emission (Sections 3.2 and 7.1). Gas wedged between the plateau and the cool gas to the E is significantly hotter (kT about 5 keV) and generally of normal metallicity.

Table 2. Spectral Results for Regions Marked in Figure 10

Region	kTa	Z/Z☉a	χ2/dof	Probabilityb
	(keV)
1	$4.11^{+0.68}_{-0.58}$	$0.91^{+0.62}_{-0.43}$	103.5/92	0.19
2	$3.60^{+0.59}_{-0.49}$	$0.14^{+0.21}_{-0.14}$	103.8/108	0.60
3	$2.09^{+0.50}_{-0.37}$	$0.18^{+0.20}_{-0.11}$	73.9/79	0.64
4	$3.21^{+0.55}_{-0.45}$	$0.14^{+0.19}_{-0.13}$	155.9/138	0.14
5	$3.59^{+0.64}_{-0.54}$	$0.62^{+0.50}_{-0.32}$	86.5/113	0.97
6	$3.86^{+1.13}_{-0.75}$	$0.13^{+0.37}_{-0.13}$	56.3/56	0.46
7	$3.38^{+0.75}_{-0.53}$	$0.77^{+0.68}_{-0.37}$	120.2/102	0.11
8	$3.37^{+0.76}_{-0.52}$	$0.57^{+0.48}_{-0.27}$	85.6/88	0.55
9	$2.18^{+0.68}_{-0.45}$	$0.07^{+0.20}_{-0.07}$	114.8/95	0.08
10	$3.76^{+1.24}_{-0.78}$	$0.26^{+0.53}_{-0.26}$	101.4/105	0.58
11	$5.20^{+2.00}_{-1.80}$	$0.79^{+1.55}_{-0.67}$	73.2/67	0.28
12	$3.98^{+1.71}_{-1.08}$	$0.52^{+0.82}_{-0.39}$	57.8/66	0.75
3+6+9+11	$3.00^{+0.39}_{-0.37}$	$0.18^{+0.14}_{-0.10}$	342.4/292	0.02

Notes. ^a90% errors for one interesting parameter. ^bNull hypothesis probability. Unacceptable when 3, 6, 9, and 11 are combined.

Download table as:  ASCIITypeset image

NGC 7016 is a relatively isolated galaxy (Figure 11). The 886 net counts from a 12 arcsec radius source-centered circle (Figure 12) give a poor fit to a thermal model alone (χ²/dof = 79.9/30), but the fit is acceptable when a power law (with no excess absorption) is added to account for emission from the active nucleus. The abundance is poorly constrained and was fixed at 0.3 times solar in the two-component fit. Results are given in Table 3. The temperature is cooler than that of the surrounding cluster, as expected for a galaxy atmosphere.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 3. Galaxy X-Ray Spectral Results

Region	za	kT	Z/Z☉	Nb	αx	S1 keV	χ2/dof
		(keV)		(cm−5)		(nJy)
NGC 7016	0.03685	$0.87^{+0.08}_{-0.06}$	0.3 (fixed)	$5.1^{+1.2}_{-1.1} \times 10^{-5}$	$0.72^{+0.38}_{-0.45}$	$8.6^{+4.0}_{-3.5}$	29.9/29
NGC 7017	0.03465	$1.14^{+0.18}_{-0.21}$	$0.3^{+1.2}_{-0.2}$	$2.5^{+2.3}_{-1.8} \times 10^{-5}$	⋅⋅⋅	⋅⋅⋅	7.2/8
NGC 7018-nucleus	0.03842	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	0.75 ± 0.14	15.8 ± 1.6	26.9/25
NGC 7018-gas	⋅⋅⋅	$1.01^{+0.14}_{-0.10}$	$0.14^{+0.11}_{-0.06}$	$7.0^{+2.4}_{-2.0} \times 10^{-5}$	⋅⋅⋅	⋅⋅⋅	17.5/14

Notes. ^aRedshift from NED of galaxy with the brightest X-ray emission (western component of NGC 7017 and eastern component of NGC 7018). ^b$10^{14} N = ({(1+z)^2 \int n_{\rm e} n_{\rm p} dV / 4\pi D_{\rm L}^2})$, where n_p, n_e are densities of hydrogen nuclei and electrons, respectively, V is volume and D_L is luminosity distance.

Download table as:  ASCIITypeset image

From a smaller circle of radius 1.5 arcsec, again centered on the radio-core position of $21^{\rm h} 07^{\rm m} 16^{\rm s}\llap{.}28, -25^\circ 28^{\prime } 08^{\prime \prime }\llap{.}5$, the counts are more highly dominated by the power-law component and spatially sharply peaked, but the fit was still improved by a small contribution from thermal gas (from χ²/dof = 26.9/17 to 10.7/15). The power-law parameter values were consistent with those from the larger extraction region, demonstrating an active-nucleus origin of the non-thermal emission. The power-law X-ray flux density (Table 3) and 4.96-GHz core flux density of 58.3 mJy lie within the scatter of the correlation of unabsorbed nuclear emission components of Evans et al. (2006), arguing in favor of a common non-thermal origin. An extension of the X-ray emission in the direction of the radio jet (Figure 12) strongly suggests X-ray synchrotron emission from an inner radio jet, found by Chandra to be a common feature of FR I radio galaxies (Worrall et al. 2001).

There is a bright pair of galaxies associated with NGC 7017 (Figure 13), although their velocity difference of 2505 ± 36 km s⁻¹ (Quintana & Ramírez 1995) is large. The centers of each were detected as X-ray point sources by wavdetect. We excluded the counts in circles of radii 3 and 1.2 arcsec around the centers of the western and eastern galaxy, respectively, and 176 net counts were then detected from an ellipse of semi-axis lengths of 17.4 and 14.2 arcsec shown in Figure 13. A good fit to a thermal model (Table 3) was found.

Zoom In Zoom Out Reset image size

NGC 7018 is also a galaxy pair. The eastern member hosts the radio-galaxy nucleus, and here point-like nuclear X-ray emission that strongly outshines diffuse gas emission is seen (Figure 14). Seven hundred nine net counts from a circle of radius 5.6 arcsec centered on $21^{\rm h} 07^{\rm m} 25^{\rm s}\llap{.}65, -25^\circ 25^{\prime } 43^{\prime \prime }\llap{.}3$ can be fitted well to a power-law model with no excess absorption (Table 3). As for NGC 7016 the power-law X-ray flux density (Table 3) and 4.96 GHz core flux density of 36.0 mJy lie within the scatter of the correlation of unabsorbed nuclear emission components of Evans et al. (2006), arguing in favor of a common non-thermal origin.

Zoom In Zoom Out Reset image size

Quintana & Ramírez (1995) find a velocity difference of only 117 ± 30 km s⁻¹ between the radio host galaxy and its western companion. Diffuse X-ray emission, associated largely with the companion galaxy, is seen to the west of the radio nucleus. 385 net counts were extracted from an ellipse of semi-axis lengths 20.6 and 11.7 arcsec, excluding the circle described above and shown in Figure 14. The data give a good fit to a thermal model (Table 3). We note that there appears to be a weak excess of radio emission associated with the western galaxy.

The radio core of NGC 7018 is within its northern lobe. While a broad trunk of radio emission connects it to the southern lobe, the jet to the north is narrow and more distinct, terminating in a double hotspot (Figure 1). There is excess X-ray emission to the north of the nucleus associated with the tip of the inner jet (Figure 14; 14 counts detected in a circle of radius 2 arcsec where an average of 2.5 is expected based on local background). The northern hotspot region is also detected in X-rays, with 18 counts detected in a circle of radius 3.5 arcsec where five counts are expected based on local background. Hotspot counts appear to be associated with each component of the double system, with a small misalignment between X-ray and radio that is probably within the tolerance of off-axis mapping uncertainties in the radio (Figure 15). The southern hotspot region is undetected in X-rays.

Zoom In Zoom Out Reset image size

The small radio source in the NW of the field of Figure 1 has a curious structure, as shown in Figure 16. wavdetect finds an X-ray source at the NW tip of the SE lobe with about 7 excess counts (0.3–5 keV). This seems to be chance coincidence, since it does not match the location of the radio core or any other radio feature. There is possible excess X-ray emission associated with bright parts of the lobes, but it is not possible to establish a level of significance since X-ray counts are generally sparse in this part of the image. There is no strong 2MASS association of the radio source.

Zoom In Zoom Out Reset image size

The X-ray cavity is a region of distinctly reduced X-ray surface brightness in the direction where the radio emissions from NGC 7018 and NGC 7016 overlap. Due to coincidence in projection, we assume the cavity is co-located within the core radius of the cluster, and we have demonstrated that the contrast in X-ray surface brightness is consistent with such an interpretation.

Many clusters and groups are now known to harbor gas cavities containing the radio plasma of active galaxies (e.g., Bîrzan et al. 2004). Since the pressure in relativistic plasma should not be less than that of surrounding X-ray gas, insight can be gained into the extent to which the relativistic plasma departs from a state of minimum energy or requires extra pressure to be provided by non-radiating particles or a reduced filling factor (e.g., Dunn et al. 2005; Bîrzan et al. 2008; O'Sullivan et al. 2011). Where a detection of inverse-Compton radiation from the cavities can be used to measure the magnetic field, the minimum-energy assumption is tested, and results concerning the other pressure components are more secure (e.g., Worrall et al. 2012). In A3744, while there is not a clean spectral separation of inverse-Compton and thermal emission in the cavity, the suggestion of lower metallicity implies that inverse Compton emission is not completely negligible, providing a modeling constraint.

The average gas pressure surrounding the X-ray cavity can be calculated from the X-ray radial profile, giving⁸ (9.8 ± 0.5) × 10⁻¹³ Pa (1σ error). A sphere of radius 100 arcsec evacuated of X-ray-emitting gas provides a model for the cavity that is consistent with the surface-brightness distribution (Figure 7). We model the cavity's radio emission from our map using data from program AC105 as synchrotron radiation from electrons with a power-law distribution of α_r = 0.9 extending to a minimum Lorentz factor of γ_min = 10, to find a minimum-energy field strength⁹ in the cavity of B_me = 0.85 nT (no protons) or 1.0 nT (equal energy in protons and electrons). We do not have radio data to constrain α_r, but the adopted value is supported by the X-ray data if modeled as power-law emission from inverse Compton scattering by the electron population (Section 3.2). The resulting pressure is a factor of 5.0 (3.5) too low to match that in surrounding X-ray gas, where here and in what follows the first value is for no protons and the second for equal energy in protons and electrons.

While the cavity pressure can be increased by adding further non-radiating particles or reducing the filling factor, the prediction for inverse Compton scattering on the cosmic microwave background radiation is still then less than 6% of the X-ray counts from the cavity, not explaining the reduced metallicity when the spectrum is fitted entirely to thermal emission. In contrast, a small departure from minimum energy in the sense of a reduced magnetic field can both provide the required pressure and explain the metallicites. Specifically, reducing the magnetic field strength to 30 (37)% of the minimum-energy value, i.e., to 0.26 (0.37) nT, matches the pressure and provides an increase in electron number density (to match the radio synchrotron emission) such that 52 (27)% of the 0.3–5-keV X-ray counts from this region would be from inverse Compton emission. A modest departure from minimum energy is consistent with findings elsewhere, and in particular with the lobes of radio galaxies for which X-ray inverse Compton is detected and modeled using the same value of γ_min as adopted here (Croston et al. 2005). We then note that the cavity shape would need to be somewhat spheroidal and elongated along the line of sight, to explain the surface brightness contrast of Figure 7 in the presence of extra (inverse Compton) X-ray counts in the cavity.

The sound speed in X-ray gas of temperature kT (keV) is roughly $516 \sqrt{kT}$ km s⁻¹, meaning that radio plasma should have been developing and filling the cavity for at least 10⁸ yr for the cavity not to have collapsed. This is within the lifetime of typical radio galaxies based on spectral aging measurements (e.g., Alexander & Leahy 1987). NGC 7018, whose southern lobe is currently at the edge of the cavity, would have taken a comparable time (1.7 × 10⁸ yr) to cross the cavity at a speed characteristic of the cluster velocity dispersion, and faster speeds are expected here, near the cluster center. Thus it seems likely that the radio plasma of NGC 7018 is primarily responsible for carving out the cavity. In contrast, the radio emission of NGC 7016 is broken into the swirl inside the cavity. While this might be due to collision with radio plasma of NGC 7018, it may also result from interaction with a wake left by the motion of NGC 7018 and its companion galaxy (see, e.g., Worrall & Birkinshaw 2005).

If NGC 7018 is responsible for the cavity, its lobes should be at least mildly overpressured with respect to the surrounding gas. The S lobe lies at the entrance to the cavity where the average gas pressure (calculated using the X-ray radial profile for an off-center angular distance of 65 arcsec) is (1.11  ±  0.05) × 10⁻¹² Pa (1σ error). We model the lobe using our 1.39 GHz JVLA data with parameters as for the cavity,¹⁰ but with a sphere of radius 18 arcsec, to find a minimum-energy field strength of B_me = 2.2 (2.6) nT and a pressure of 1.3(1.8) × 10⁻¹² Pa, slightly above that in the external gas. If the magnetic field is below the minimum-energy value by the amount we argue is likely for the cavity, then the S lobe is overpressured by a factor of 4.9 with respect to its surrounding gas. Such an overpressure would lead to a shock of Mach number 2 being driven into the external medium (e.g., Equation (49) of Worrall & Birkinshaw 2006) and direct heating in the immediate vicinity of the lobe, not ruled out by the data due to masking by much foreground and background gas along the line of sight. Direct evidence of shock heating is difficult to verify around cluster-embedded radio lobes, but there is growing evidence for moderately strong shocks around intermediate-power radio galaxies like NGC 7018 (Worrall et al. 2012).

It is striking that the radio filament of the S lobe of NGC 7018 (Figure 1) runs along the northern inside edge of the cavity (Figure 2). The structure contains strands, and the brightening is suggestive of preferential particle acceleration along the interface between the cavity and the external X-ray gas—something which in principle could be tested by radio spectral index mapping.

The tendrils of NGC 7018 run along the outside of the plateau and look like they are buoyant, so that their deprojected pressures should match those in the plateau, as measured using the spectral fit and profile in Figure 8. To test this we take rectangular sections at the end of each tendril modeled as cylinders of radius 21 arcsec and lengths 126 and 162 arcsec centered at distances from the center of the gas distribution of 380 and 530 arcsec for the SW and NW tendrils, respectively. If the tendrils are in the plane of the sky, the external pressure in the SW and NW is 5.0 × 10⁻¹³ and 3.6 × 10⁻¹³ Pa, respectively. For minimum energy the two tendrils then agree in being under-pressured by a factor of 5.5 (3.9), where again the two values correspond to a lepton-only plasma or one where the lepton energy density is matched by that in protons. If in the plane of the sky with no significant entrainment, pressure balance can be restored if the magnetic field is 29 (35)% of the minimum-energy value. These departures from minimum energy are remarkably similar to those required in the cavity, and by the region in the NW tendril the magnetic field strength so estimated would have dropped to about half that in the cavity. However, while there may be reasons for dynamic structures to depart from minimum energy, it is more appealing for these buoyant flows to have reached a state of minimum energy. The tendrils extend into the region where the X-ray gas pressure is falling with radius, r, as roughly r^−1.7, and they would be in pressure balance with this gas if lying at about θ = 25(29)° to the line of sight. The actual value of θ is likely to be set by the unknown direction of the velocity vector of NGC 7018, such that the buoyant tendrils trail, and such values of θ are plausible. Some increase in internal pressure in the tendrils through gas entrainment is also likely, allowing θ to increase from these estimates. Not being in the plane of the sky has the advantage of explaining the rather dramatic fading of the tendrils which occurs at larger r for smaller θ. Strong fading is expected sufficiently far out in the cluster outer atmosphere that the tendrils are no longer supported, and the relativistic plasma expands adiabatically. This is easily accommodated by a further steepening in the radial profiles beyond the values of r for which we are able to make measurements in the current data.

The broad tendril to the SE is the extension of the brighter (approaching) jet of NGC 7016. The two jet bends, first to the south and then to the east (Figure 1) and presumably the result of shocks, are sufficiently large in projection to suggest that this jet is at a relatively small angle to the line of sight. We test a cylinder of radius 33 arcsec and length 98 arcsec (position angle 70°) at 340 arcsec from the center of the gas distribution where the external pressure (in the cooler gas here) is 2.8 × 10⁻¹³ Pa. As for NGC 7018's tendrils, there is an underpressure by a factor of a few that can be restored if the radio emission is lying at 24 (31)° to the line of sight.

X-ray cavities are sufficiently common that they are regarded as an important heat source, with enough power to balance radiative cooling in dense cluster cores (e.g., Dunn et al. 2005; Rafferty et al. 2006). In the case of the A3744's cavity, we can estimate the change in enthalpy as γPV/(γ − 1), where P is the pressure of the X-ray gas, V is the volume of the cavity, and γ = 4/3 is the ratio of specific heats for relativistic gas. The resulting value is roughly 2 × 10⁵³ J. In Section 3.1 we pointed out that in its integrated properties the cluster is too hot by kΔT ≈ 1.5 keV for its luminosity, based on scaling relations. This is equivalent to an excess enthalpy of γM_gaskΔT/μm_H(γ − 1), where M_gas is the total gas mass, μm_H is the particle mass, and γ = 5/3 is the ratio of specific heats for the X-ray-emitting gas. The resulting value is roughly 1.7 × 10⁵⁵ J. A staggering 85 cavities like those attributed to NGC 7018 would be needed to explain the excess heat in A3744 that causes it to deviate from scaling relations.

The most likely source of the excess heat is therefore a merger. The temperature structure seen in Figure 10 points to such a merger being roughly along a NW–SE direction, such that gas heated in collisions escapes to the sides. It is also consistent with the velocity data discussed in Section 6 and the two centers marked as crosses in Figure 5. If we take the merging group to have about 5% of the mass of A3744, based on the number of galaxies within known velocities, and use the radial-velocity offset of roughly 1700 km s⁻¹ as the total velocity difference, then the conversion of 20% of the kinetic energy of the group would be sufficient to produce the required excess heating. That is, the heating requires only a minor merger event. The absence of a shock feature in the X-ray image is not unexpected if this merger is indeed responsible for the heating, since such shocks are only seen when the merger is in the plane of the sky, and the radial velocity difference would be excessive if the merger axis is far from the line of sight in the present case.

Numerical simulations have shown that mergers lead to persistent (gigayears) relative motion of gas streams which are in pressure equilibrium (sloshing), and this can explain X-ray features seen rather commonly in relaxed cool-core galaxy clusters and known as cold fronts, where a large density discontinuity is in pressure balance due to the dense gas being cooler (Ascasibar & Markevitch 2006; Markevitch & Vikhlinin 2007). A3744 is far from relaxed, and shows no obvious evidence of sloshing. We clearly see density substructure, but the temperature and density features lie in different parts of the gas, the former having a closer relationship to the location of the radio tendrils. The most likely explanation is that we are witnessing a relatively recent merger encounter, one which may also have helped trigger the radio galaxies into their current phase of activity.

The excess X-ray emission in the region we name the "plateau" can be understood since the centroid of the distribution of galaxies with reported velocities (biased toward the more massive ones) lies in this neighborhood. Its temperature is unremarkable compared with that in the X-ray brightest region, including the temperatures in the arm and surrounding the cavity. What is remarkable is that the radio tendrils of NGC 7018 appear to border the hottest gas (Figure 10), and envelop the roughly cylindrical gas region seen in Figure 5 that forms the bright X-ray bridge between the arm and the plateau. It is tempting to invoke cause and effect, such that the radio plasma of NGC 7018 helps to reduce thermal transport between the warmer and cooler gas, and to reduce momentum transport and hence viscous drag, so that X-ray gases of different temperatures can slide by one another with little heat exchange or mixing. The regions in the temperature map were defined automatically from the smoothed X-ray data, with no reference to the radio structures. It is also notable that the southern tendril of NGC 7016 lies along a temperature boundary.

The morphology of the radio and X-ray extensions suggest that the radio extensions of NGC 7018's lobes were caused by the relative motion of the host galaxy and the gas extending into the plateau. The attachment of radio plasma to this infalling gas would cause the extension of the radio lobes into the tendrils and cause the gas bridge to become at least partially enveloped by the radio plasma. Because the radio plasma is moderately strongly magnetized, it will act as an effective barrier to transport between the inner arm and the general cluster environment. Not only are the particle gyroradii smaller (by a factor of about 10) in the radio plasma than in the general cluster environment, but the field will also tend to be ordered by the stretching along the radio tendrils. This will reduce the (perpendicular) thermal conductivity between the two gas regions.

The degree of thermal protection offered by the radio features depends both on the reduction in transport through the radio plasma and the fraction of the interface between regions of different temperature covered by the radio plasma. Taking Spitzer conductivity

$$\begin{eqnarray*} &&\kappa \approx 5 \times 10^{-12} (T/{\rm K})^{5/2}\ {\rm W\ m}^{-1}\ {\rm K}^{-1} \end{eqnarray*}$$

(Spitzer 1962) appropriate for intracluster gas, and modeling the gas as a cylinder of radius R ≈ 50 kpc, the timescale for a temperature difference ΔT to be erased is of the order of

$$\begin{eqnarray*} &&\tau \approx {n_o R^2 k \over \kappa } \left({T \over \Delta T}\right) \end{eqnarray*}$$

or about 100 Myr, taking the proton density, n_o, to be roughly 940 m⁻³ (Section 3.1). This is significantly shorter than the time (about 300 Myr) taken for the gas to travel from NGC 7018 to the plateau at the sound speed. However, it appears that at least half the surface area is covered by radio plasma. If this radio plasma cuts κ by a factor of about 10, then effective heat transport occurs over only about half the surface of the cylinder and its thermal lifetime is raised by a factor of order three. The associated drop in viscosity may also help the gas to slide in toward the plateau.

There is good evidence of X-ray emission from both the east and west components of the northern double hotspot of NGC 7018, each with similar ratios of X-ray to radio flux density. The X-ray emission is about two orders of magnitude higher than expected due to inverse-Compton scattering if the electrons and magnetic field are at minimum energy. This result is in common with many other Chandra-observed hotspots, particularly those in FR II radio galaxies at the lower end of the spectrum of total radio power (Hardcastle et al. 2004), The large reduction in magnetic field (about a factor of 24 for NGC 7018) and the subsequent high increase in source energy needed to explain the X-ray emission by the inverse-Compton mechanism led Hardcastle et al. (2004) to suggest that the electron spectrum extends to high enough energies in such hotspots for synchrotron radiation to dominate the X-ray output. In NGC 7018's hotspots, if the electron spectrum breaks from a power law slope of p = 2.44 (giving the observed α_r = 0.72) by Δα = 1 at a Lorentz factor of γ = 6 × 10⁴, the synchrotron radiation from 100 TeV electrons could be responsible for the X-rays in both components in minimum-energy magnetic fields of about 10 nT. While this is an attractively simple explanation, we note that Werner et al. (2012) have used Spitzer data to claim that a single-component broken-power-law synchrotron spectrum can be excluded for 80% of 24 hotspots they study. They point to the work of De Young (2002) as possible support for magnetic fields in hotspots not having increased to minimum-energy levels.

An alternative reason that hotspots on the jet side of a source (as in NGC 7018) may be unusually X-ray bright is if relativistic beaming is a factor. An applicable mechanism suggested by Georganopoulos & Kazanas (2003) is that for a jet approaching a hotspot with high bulk Lorentz factor the available photon fields for Compton scattering will be boosted by the strongly directional (particularly in the jet frame) radio synchrotron emission from the terminal hotspot, resulting in inverse Compton X-rays beamed in the forward direction of the jet and offset (upstream) from the peak of the radio emission. At high redshift the spatial separation of components necessary to test this mechanism is not possible, although a claim at low redshift has been made for PKS B2152-699 (Worrall et al. 2012). In NGC 7018 there is some evidence for an offset between radio and X-ray emission, although we cannot rule out residual uncertainties in our 5 GHz radio mapping 3.5 arcmin off axis as responsible. A non-detection of the southern hotspot region, at the termination of the jet pointed away from the observer, is consistent with beaming playing a part in the detection of the X-ray hotspots, although statistics are such that the formal upper limit on the X-ray flux is at a similar level to the detections in the north, and so without more sensitive X-ray data firm conclusions cannot be drawn.