FRONTIER FIELDS CLUSTERS: CHANDRA AND JVLA VIEW OF THE PRE-MERGING CLUSTER MACS J0416.1-2403

Galaxy clusters grow by merging with other clusters and by accreting smaller mass structures from the intergalactic medium. Signs of these interactions are imprinted in the intracluster medium (ICM) and detected in X-ray observations as cold fronts, shock fronts, turbulence, and substructure in the ICM (e.g., Markevitch & Vikhlinin 2007; Randall et al. 2008a; Zhuravleva et al. 2015). Other footprints of cluster interactions can be seen in the radio band as halos and relics (e.g., Feretti et al. 2012). If the merger is not in the plane of the sky, mergers can be detected in optical observations based on multiple peaks in the radial velocity distribution and in the spatial galaxy distribution. Furthermore, comparison of X-ray and optical/lensing data also reveals signatures of merging events, most notably as offsets between the dark matter (DM) and the gas components of the colliding clusters (e.g., Markevitch et al. 2004; Clowe et al. 2006; Randall et al. 2008b; Merten et al. 2011; Dawson et al. 2012).

Radio halos, which have been detected in some mergers, are diffuse synchrotron-emitting sources with low surface brightness, steep spectral indices (α < −1, ${S}_{\nu }\propto {\nu }^{\alpha }$), and typical sizes of ∼1 Mpc. Because of their steep radio spectra, low-frequency radio observations play an important role in characterizing their properties. Two main models have been proposed to explain the origin of such halos:

• 1.  
  Reacceleration by turbulence: Large-scale turbulence generated during the merger event supplies the energy required to reaccelerate fossil cosmic rays (CRs) back to relativistic energies, at which time they become synchrotron-bright (e.g., Brunetti et al. 2001; Petrosian 2001).
• 2.  
  Acceleration by hadronic collisions: Inelastic collisions between CR protons trapped in the gravitational potential of a cluster (e.g., originating from supernova explosions, active galactic nuclei (AGNs) outbursts, previous merger events, etc.) and thermal ICM protons give rise to a secondary population of CR electrons, which consequently are visible at radio frequencies (e.g., Blasi & Colafrancesco 1999; Dolag & Enßlin 2000).

However, hybrid models have also been postulated, in which radio halos may be produced by turbulent reacceleration of secondary particles resulting from hadronic proton–proton collisions (Brunetti & Blasi 2005; Brunetti & Lazarian 2011). The radio halos predicted by hybrid models are expected to be found in more relaxed clusters and to be underluminous for the masses of the hosting clusters (see also Brunetti & Jones 2014, for a review).

Hadronic radio halo models predict that magnetic fields should either be different in clusters with and without halos, or that gamma-ray emission should be detected in clusters hosting halos (e.g., Jeltema & Profumo 2011). However, there is no evidence that magnetic fields are stronger in clusters with radio halos (e.g., Bonafede 2010), and there has been no conclusive gamma-ray detection with the Fermi telescope (Ackermann et al. 2010). Therefore, current observational evidence disfavors hadronic models.

A textbook example of a merging cluster with a bright radio halo is the famous Bullet cluster, 1E 0657-56 (Elvis et al. 1992; Liang et al. 2000; Markevitch et al. 2002; Shimwell et al. 2014). Chandra has revealed that this system has a bullet-like subcluster core moving through the disturbed ICM of the main cluster (Markevitch et al. 2002). Immediately ahead of the core there is a density discontinuity associated with a cold front, while further out in the same direction there is a density discontinuity associated with a shock front (Markevitch et al. 2002; Owers et al. 2009). Recently, another shock front has been found on the opposite side of the Bullet cluster, behind the bullet-like core (Shimwell et al. 2015). Chandra observations of the Bullet cluster were followed up with optical/lensing data acquired, most notably, with the Very Large Telescope and the Hubble Space Telescope (HST). Chung et al. (2010) have shown that the redshift distribution of the Bullet cluster is bimodal, with two redshift peaks at z ∼ 0.21 and z ∼ 0.35, which imply a velocity difference ∼3000 km s⁻¹ between the two merging subclusters. The gravitational lensing analysis has shown that the Bullet cluster is a dissociative merger, in which the essentially collisionless DM (σ_DM < 0.7 g cm⁻², Randall et al. 2008b) has decoupled from the collisional ICM (Markevitch et al. 2004; Randall et al. 2008b). Lage & Farrar (2014) have combined the X-ray and lensing data with numerical simulations to constrain the merger scenario of the Bullet cluster.

Here, we present results from Chandra, Jansky Very Large Array (JVLA), and Giant Metrewave Radio Telescope (GMRT) observations of another merging galaxy cluster: the HST Frontier Fields²² cluster MACS J0416.1-2403 (z = 0.396, Ebeling et al. 2001; Mann & Ebeling 2012). Optical and lensing studies of this cluster have been presented recently by, e.g., Zitrin et al. (2013, 2015), Schirmer et al. (2015), Jauzac et al. (2014), Grillo et al. (2015). Most recently, Jauzac et al. (2015) mapped the DM distribution of this merging system using strong- and weak-lensing data. Their analysis revealed two mass concentrations associated with the two main subclusters involved in the merger, plus two smaller X-ray-dark mass structures NE and SW of the cluster center. Jauzac et al. (2015) combined their lensing data with archival Chandra observations to study the offsets between the DM and the gas components. They report good DM-gas alignment for the NE subcluster, but a significant offset for the SW one. Based on the spectroscopic, lensing, and X-ray results, Jauzac et al. (2015) proposed two possible scenarios for the merger event in MACS J0416.1-2403—one pre-merging and one post-merging—but were unable to distinguish between the two.

The analysis presented herein combines significantly deeper Chandra observations with recently acquired JVLA and GMRT data and with the optical/lensing results reported by Jauzac et al. (2015) to improve our understanding of the merger event in MACS J0416.1-2403. The Chandra JVLA, and GMRT observations and data reduction procedures are described in Section 2. In Section 3 we discuss the background analysis of the Chandra data, and in Sections 4–7 we present the X-ray results. The radio results are presented in Section 8. In Section 9 we use the deeper Chandra data to revise the DM-gas offsets reported by Jauzac et al. (2015). In Section 10 we discuss the implications of our findings for the merger scenario of MACS J0416.1-2403. In Section 11 we examine the origin of the radio halo. A summary of our results is provided in Section 12.

Throughout the paper we assume a ΛCDM cosmology with H₀ = 70 km s⁻¹ Mpc⁻¹, Ω_M = 0.3, and Ω_Λ = 0.7. At the redshift of the cluster, 1' corresponds to approximately 320 kpc. Unless stated otherwise, the errors are quoted at the 90% confidence level.

2.1. Chandra

The HST Frontier Fields cluster MACS J0416.1-2403 was observed with Chandra for 324 ks between 2009 June and 2014 December. A summary of the observations is given in Table 1.

Table 1.  Chandra Observations

ObsID	Observing Mode	CCDs On	Starting Date	Total Time	Clean Time
				(ks)	(ks)
10446	VFAINT	0, 1, 2, 3, 6	2009 Jun 07	15.8	15.8
16236	VFAINT	0, 1, 2, 3, 6	2014 Aug 31	39.9	38.6
16237	FAINT	0, 1, 2, 3, 6	2013 Nov 20	36.6	36.3
16304	VFAINT	0, 1, 2, 3, 6	2014 Jun 10	97.8	95.2
17313	VFAINT	0, 1, 2, 3, 6	2014 Nov 28	62.8	59.9
16523	VFAINT	0, 1, 2, 3, 6	2014 Dec 19	71.1	69.1

Download table as:  ASCIITypeset image

The data were reduced using CIAO v4.7 with the calibration files in CALDB v4.6.5. The particle background level of the VFAINT observations was lowered by filtering out CR events associated with significant flux in the 16 border pixels of the 5 × 5 event islands. Soft proton flares were screened out from all observations using the deflare routine, which is based on the lc_clean code written by M. Markevitch. The flare screening was done using an off-cluster region at the edges of the field of view (FOV), which was selected to include only pixels located at distances greater than 2.5 Mpc from the cluster center and to exclude any point sources detectable by eye. The clean exposure times following this cleaning are listed in Table 1.

The inspection of an ObsID 10446 spectrum extracted from an off-cluster region showed a significant high-energy tail, which indicates that this observation is contaminated by flares that were not detected by the filtering routine. Given the relatively short exposure time of this observation, we decided to exclude ObsID 10446 from our analysis rather than attempt to model the flare component.

We reprojected the other five observations to a common reference frame, and created merged images in the energy bands 0.5–2, 0.5–3, 0.5–4, 0.5–7, and 2–7 keV. The exposure-corrected, vignetting-corrected 0.5–4 keV mosaic image is shown in Figure 1. To detect point sources, we also created merged exposure map-weighted point-spread function images (ECF = 90%) in each of the five energy bands. Point sources were detected in the individual bands with the CIAO task wavdetect, using the merged maps, wavelet scales of 1, 2, 4, 8, 16, and 32 pixels, a sigma threshold of 5 × 10⁻⁷, and ellipses with 5σ axes. A few additional point sources that were missed by wavdetect were selected by eye and excluded from the data. All point sources detected by wavdetect were excluded from the data very conservatively, using the elliptical region that covered the largest area among the five elliptical regions identified for the different energy bands.

Zoom In Zoom Out Reset image size

2.2. JVLA

2.3. GMRT

2.4. VLITE

We analyzed the Chandra data using the Group F stowed background event files, which are appropriate for observations taken after 2009 September 21. The particle background was subtracted from the data, while the foreground and background sky components were modeled from off-cluster regions. For each of the cluster ObsIDs, we created corresponding stowed background event files, applied the VFAINT cleaning to those associated with cluster observations taken in this observing mode, and renormalized the stowed background observations such that the count rates in the energy band 10–12 keV matched those of the corresponding observation in the same energy band. We modeled the sky components as the sum of unabsorbed thermal emission from the Local Hot Bubble (LHB; APEC component with T ≈ 0.1 keV; Snowden 1998), absorbed thermal emission from the Galactic Halo (GH; APEC component with T ≈ 0.2 keV; Henley et al. 2010), and absorbed non-thermal emission from undetected point sources in the FOV (power-law component with index 1.41 ± 0.06; De Luca & Molendi 2004). The redshifts of the foreground components were set to 0, while the abundances were set to solar values assuming the abundance table of Anders & Grevesse (1989). The hydrogen column density was fixed to 2.89 × 10²⁰ atoms cm⁻² (Kalberla et al. 2005). The Chandra spectra were fit in the energy band 0.5–7 keV using Xspec v12.8.2—the lower limit was chosen to avoid calibration uncertainties at low energies, and the upper limit to increase the signal-to-noise ratio (S/N).

When fitting the Chandra spectra in the energy band 0.5–7 keV, the ∼0.1 keV LHB component cannot be constrained. Instead, we constrain this component using a ROentgen SATellite (ROSAT) All-Sky Survey (RASS) spectrum²⁴ corresponding to an annulus with radii 015 and 10 around the cluster. The ROSAT spectrum was fit in the energy band 0.1–2.4 keV, with the normalization of the power-law background component fixed to 8.85 × 10⁻⁷ photons keV⁻¹ cm⁻² s⁻¹ arcmin⁻² (Moretti et al. 2003). The temperatures and normalizations of the LHB and GH components were free in the fit. The results of the fit to the ROSAT spectrum are presented in Table 4.

Table 4.  Foreground and Background Spectral Models

	Chandra		ROSAT
Component	Ta	${\mathcal{N}}$ b	Ta	${\mathcal{N}}$ b
LHB	0.10 ± 0.01c	${9.32}_{-0.83}^{+0.62}\times {10}^{-7}$ c	0.10 ± 0.01	${9.32}_{-0.83}^{+0.62}\times {10}^{-7}$
GH	${0.25}_{-0.08}^{+0.23}$	${2.35}_{-1.66}^{+4.20}\times {10}^{-7}$	${0.22}_{-0.04}^{+0.05}$	${6.74}_{-2.06}^{+3.44}\times {10}^{-7}$
Power-law	⋯	${4.20}_{-0.73}^{+0.86}\times {10}^{-7}$	⋯	⋯
ObsID 16236
Power-law	⋯	${2.17}_{-0.86}^{+0.76}\times {10}^{-7}$	⋯	⋯
ObsID 16237
Power-law	⋯	${2.45}_{-0.59}^{+0.60}\times {10}^{-7}$	⋯	⋯
ObsID 16304
Power-law	⋯	${3.70}_{-0.84}^{+0.86}\times {10}^{-7}$	⋯	⋯
ObsID 17313
Power-law	⋯	${4.33}_{-0.69}^{+0.71}\times {10}^{-7}$	⋯	⋯
ObsID 16523

Notes.

^aTemperature, in units of keV. ^bNormalization, in units of cm⁻⁵ arcmin⁻² for the thermal components, and in units of photons keV⁻¹ cm⁻² s⁻¹ arcmin⁻² for the non-thermal components. ^cFixed parameter.

Download table as:  ASCIITypeset image

Based on the ROSAT results, the parameters of the LHB components were fixed for the Chandra spectra. The five Chandra spectra were then fitted in parallel, assuming that they are described by the same foreground model. The power-law normalizations, on the other hand, were left to vary independently, in order to account for the varying exposure time across the merged observation that was used for point source detection. The energy sub-band 0.5–0.8 keV was ignored for ObsID 16237 due to negative spectral residuals caused by a change in the spectral shape of the background component that is not filtered by the VFAINT cleaning, which occurred between 2009 (the date of the Group F background files) and 2013 (the date of the observation; A. Vikhlinin 2015, private communication). The subtraction of the instrumental background resulted in small line residuals near the spectral positions of the Al Kα (E ≈ 1.49 keV), Si Kα (E ≈ 1.75 keV), and Au Mα, β (E ≈ 2.1 keV) fluorescent instrumental lines. Therefore, we excluded from the spectra very narrow bands (ΔE = 0.10 keV) surrounding these lines.

The best-fitting foreground and background parameters are summarized in Table 4. The spectra were grouped to have at least 1 count per bin, and the fit was done using the extended C-statistic²⁵ (Cash 1979; Wachter et al. 1979).

MACS J0416.1-2403 has been previously identified as a merging galaxy cluster at z = 0.396 (Mann & Ebeling 2012). The brightness of the NE subcluster core is significantly more peaked than that of the SW subcluster, which instead has a rather flat brightness distribution (Figure 1).

We determined the average cluster temperature by extracting spectra from a circular region of radius 1.2 Mpc (approximately R₅₀₀, Sayers et al. 2013) centered at α = 4^h 16^m 088 and δ = −24° 04' 140 (J2000). The spectra extracted from the five Chandra ObsIDs were fit in parallel, assuming the cluster emission is described by a single-temperature thermal component.²⁶ For this global fit, the sky background parameters were fixed to the values listed in Table 4, and the redshift was fixed to 0.396. We found $T={10.06}_{-0.49}^{+0.50}$ keV and ${L}_{{\rm{X}},\;0.1-2.4\;\mathrm{keV}}=(7.43\pm 0.08)\times {10}^{44}$ erg s⁻¹. The average cluster parameters are shown in Table 5.

Table 5.  Cluster Properties in R₅₀₀

T (keV)	${10.06}_{-0.49}^{+0.50}$
Z (${{\rm{Z}}}_{\odot }$)	${0.24}_{-0.04}^{+0.05}$
${L}_{{\rm{X}},0.1-2.4\ {\rm{keV}}}\;(\mathrm{erg}\;{{\rm{s}}}^{-1})$	(7.43 ± 0.08) × 1044

Download table as:  ASCIITypeset image

5.1. Mapping of the ICM Temperature

We mapped the cluster properties using contbin (Sanders 2006).²⁷ The merged Chandra image was adaptively binned in regions of ∼3600 source plus sky background counts in the energy band 0.5–7 keV. The bins follow the surface brightness contours of the cluster, thus minimizing possible gas mixing in bins located near density discontinuities. Instrumental background and total spectra were extracted from regions corresponding to each bin, and grouped to have at least one count per bin. The net spectra were modeled as the sum of sky background components plus a single temperature absorbed thermal model (phabs×APEC) with free temperature and normalization. The metallicity was fixed to $Z=0.24\;{Z}_{\odot }$ (see Section 4). The sky background emission was fixed to the model summarized in Table 4. The spectra extracted from the five Chandra data sets were fitted in parallel, with the temperatures and normalizations linked between the different ObsIDs. The fits were done using the extended C-statistic. The resulting temperature map is shown in Figure 2.

Zoom In Zoom Out Reset image size

The temperature is high throughout the cluster, which causes the uncertainties on the measurements to be high. The temperature map does not show significant structure. Within the uncertainties, the temperatures out to ∼400 kpc from the cluster center are consistent with ∼10 keV.

5.2. Mapping of the ICM Pressure and Entropy

The electron number density can be derived either from the model normalization of the fitted spectrum, or from the surface brightness in the regions of interest. Extracting the number density from either of these quantities requires assumptions about the cluster geometry. Here, we estimated the electron number density from the surface brightness; we note that our results are unchanged when deriving the density from the spectral normalization.

The surface brightness, ζ, is essentially proportional to the emission measure,

$$\begin{eqnarray}&&\zeta \propto \int {n}_{{\rm{e}}}^{2}dl\end{eqnarray} \tag{ 1 }$$

where the integration is done along the line of sight and n_e is the electron number density. More accurately, the surface brightness is also temperature-dependent. This dependence introduces an uncertainty of ≲10% in the pressure and entropy maps, which is less than the uncertainties on the temperatures and does not affect our results.

We approximated the density as ζ^1/2, and defined the pseudo-pressure and pseudo-entropy as:

$$\begin{eqnarray}&&P=T\;{\zeta }_{0.5-2\ {\rm{keV}}}^{1/2},\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&K=T\;{\zeta }_{0.5-2\ {\rm{keV}}}^{-1/3}.\end{eqnarray} \tag{ 3 }$$

The pseudo-pressure and pseudo-entropy maps are included in Figure 2.

Strong jumps in pressure and entropy would indicate the presence of strong density discontinuities in the ICM. However, neither the pressure map, nor the entropy map of MACS J0416.1-2403 shows evidence of such strong jumps between the regions. We note that while a pressure and entropy discontinuity can be seen between regions #3 and #0, the size of region #0 is too large for this result to truly indicate that there is a density discontinuity at the boundary of the two regions.

In Table 6 we list the best-fitting temperatures, spectral normalizations, and 0.5–2 keV count rates for the regions in Figure 2. The bin numbers necessary to relate Table 6 and Figure 2 are shown in Figure 3. An interactive version of the temperature map, which combines Figures 2, 3 and Table 6 is available at http://hea-www.cfa.harvard.edu/~gogrean/interactive/MACSJ0416_Tmap.html.

Zoom In Zoom Out Reset image size

Table 6.  Best-fitting Temperatures, Normalizations, and 0.5–2 keV Count Rates for the Regions in Figure 3

Bin Number	Ta	${\mathcal{N}}$ b	${S}_{{\rm{X}}}$ c
0	${7.95}_{-0.92}^{+1.32}$	(2.24 ± 0.06) × 10−5	(7.39 ± 0.07) × 10−6
1	${10.74}_{-1.37}^{+2.10}$	(1.70 ± 0.05) × 10−4	(3.38 ± 0.06) × 10−5
2	${13.25}_{-2.32}^{+3.07}$	(2.31 ± 0.07) × 10−4	(4.34 ± 0.08) × 10−5
3	${10.32}_{-1.35}^{+1.91}$	(2.77 ± 0.08) × 10−4	(5.30 ± 0.09) × 10−5
4	${11.69}_{-1.46}^{+2.10}$	(1.63 ± 0.05) × 10−3	(2.96 ± 0.06) × 10−4
5	${9.73}_{-1.24}^{+1.52}$	(9.80 ± 0.29) × 10−4	(1.84 ± 0.04) × 10−4
6	${15.39}_{-2.87}^{+3.42}$	(9.03 ± 0.23) × 10−4	(1.62 ± 0.03) × 10−4
7	${12.90}_{-2.15}^{+3.64}$	(7.13 ± 0.22) × 10−4	(1.27 ± 0.03) × 10−4
8	${12.18}_{-1.90}^{+2.41}$	${7.10}_{-0.21}^{+0.22}\times {10}^{-4}$	(1.31 ± 0.03) × 10−4
9	${11.43}_{-1.65}^{+2.29}$	(5.51 ± 0.17) × 10−4	(1.02 ± 0.02) × 10−4
10	${8.41}_{-0.90}^{+1.37}$	(4.12 ± 0.13) × 10−4	(8.02 ± 0.15) × 10−5

Notes.

^aTemperature, in units of keV. ^bSpectral normalization, in units of ${\mathrm{cm}}^{-5}\;{\mathrm{arcmin}}^{-2}.$ ^c0.5–2 keV count rate, in units of $\mathrm{photons}\;{\mathrm{cm}}^{-2}\;{{\rm{s}}}^{-1}\;{\mathrm{arcmin}}^{-2}.$

Download table as:  ASCIITypeset image

The merging history of the individual subclusters is closely related to their degree of disturbance. In the following two sections, we perform the imaging analysis of the NE and SW subclusters in order to characterize their merging states. For both subclusters, we created surface brightness profiles in sectors centered on the respective X-ray peaks, and attempted to model the profiles with isothermal β-models (Cavaliere & Fusco-Femiano 1976, 1978):

$$\begin{eqnarray}&&S(r)={S}_{0}{\left[1+{\left(\displaystyle \frac{r}{{r}_{{\rm{c}}}}\right)}^{2}\right]}^{-3\beta +0.5}\end{eqnarray} \tag{ 4 }$$

where S₀ is the central surface brightness, r_c is the core radius, β is the beta parameter, and r is the radius from the cluster center.

Simple β-models provide good descriptions of the X-ray profiles of galaxy clusters that do not have strong ICM temperature gradients (e.g., Ettori 2000a, 2000b). Based on Figure 2, there are indeed no strong temperature gradients in the NE and the SW subclusters of MACS J0416.1-2403.

Before fitting, all the profiles were binned to a minimum of 1 count/bin. The fits used Cash statistics. Fitting was done using a modified version of the PROFFIT package²⁸ (Eckert et al. 2011). All the surface brightness profiles presented in the following subsections are in the energy band 0.5–4 keV. The profiles were instrumental background-subtracted, and exposure- and vignetting-corrected.

6.1. NE Subcluster

The sector used to create the surface brightness profile of the NE subcluster and the fits to this profile are shown in Figure 4. The sky background surface brightness level was determined by fitting a constant to the outer bins (radii 50–107) of the profile. The sky background level was kept fixed in the following fits. The very central part of the profile is significantly underestimated by the best-fitting β-model. Instead, the profile is described better by a double β-model:

$$\begin{eqnarray}&&S(r)=\displaystyle \sum _{i=1,2}\;{S}_{0,i}{\left[1+{\left(\displaystyle \frac{r}{{r}_{{\rm{c}},i}}\right)}^{2}\right]}^{-3\beta +0.5}.\end{eqnarray} \tag{ 5 }$$

Double β-models provide good representations of cooling-core clusters (e.g., Jones & Forman 1984; Ota et al. 2013). However, the core of the NE subcluster in MACS J0416.1-2403 is far from cool, having a temperature >10 keV based on the temperature map in Figure 2. To constrain a possible cooler component, we examined the possibility that the projected temperature of the gas in bin #4 (see Figure 3) is increased due to the projection of hot gas onto the subcluster core. We modeled the spectrum of bin #4 with a two-temperature APEC model: one describing possible cool gas in the cluster center, and another describing hotter gas (T > 10 keV) in the cluster outskirts. The normalizations of the two components were free in the fit. The count statistics do not allow us to leave both temperatures free in the fit. Therefore, we left the temperature of the hot plasma free, and fixed the temperature of the core to 5 keV.²⁹ In the best-fitting model, the normalization of the cool component is about 5 times lower than that of the hot component, which implies that the gas density in the core is lower than the density right outside the core radius; such a density model is not physical. The two-temperature APEC model also provides no improvement in the statistics of the fit. If the core has a temperature <5 keV, then the normalization of the cool component would be even lower relative to the normalization of the hot component, so the core density would be even lower.

Zoom In Zoom Out Reset image size

Cooler gas in the NE core could be masked by inverse Compton (IC) emission from the AGN hosted by the NE subcluster brightest cluster galaxy (BCG). To examine this possibility, we measured the temperature in an annulus with radii 32 and 64 kpc around the AGN in the NE BCG. The best-fitting temperature in this annulus is very high, ${15.65}_{-3.22}^{+5.03}$ keV. In a smaller circle with a radius of 35 kpc around the NE core, the best-fitting temperature is ${10.54}_{-2.11}^{+3.00}$ keV, and consistent with the temperature calculated in an annulus around the core. Therefore, we find no evidence that the NE core is cool.

From the best-fitting double β-model, we calculated the central density of the NE core to be (1.4 ± 0.3) × 10⁻² cm⁻³. The density and temperature of the NE core hence imply a cooling time of ${3.5}_{-0.9}^{+1.0}$ Gyr, which further supports our conclusion that the NE subcluster cannot be classified as a cool core cluster based on currently available X-ray data.

To investigate if the double β-model shape of the NE profile is caused by substructure in a particular direction, we divided the NE sector shown in Figure 4 into three subsectors, and modeled the profile of each subsector with β- and double β-models. The fits are shown in Figure 5. For each of the subsectors, a double β-model describes the profile better than a single β-model at a confidence level >99.99%.

Zoom In Zoom Out Reset image size

6.2. SW Subcluster

The surface brightness profile of the SW subcluster and the sector used in the extraction of this profile are shown in Figure 6. As for the NE profiles, the sky background profile was modeled by fitting a constant to the outer bins of the profile, in the range r = 45–107. Unlike the NE profiles, the SW profile is modeled well by a simple β-model. The fit is shown in Figure 6.

Zoom In Zoom Out Reset image size

A weak edge in the profile can be seen at r ∼ 15. We divided the profile into three subsectors with equal opening angles (60°) to check whether the edge is seen along all three directions, and found that it is present only in the central subsector (position angles 280°–340°, measured from the W in a counterclockwise direction). To describe it, we fitted a projected broken power-law elliptical density model to the surface brightness profile between r = 05–40. The density model is defined as:

$$\begin{eqnarray}n(r)=\left\{\begin{array}{ll}C\;{n}_{0}{\left(\displaystyle \frac{r}{{r}_{{\rm{d}}}}\right)}^{-\alpha }, & {\rm{if}}\;{\rm{}}r\leqslant {r}_{{\rm{d}}}\\ {n}_{0}{\left(\displaystyle \frac{r}{{r}_{{\rm{d}}}}\right)}^{-\beta }, & {\rm{if}}\;{\rm{}}r\gt {r}_{{\rm{d}}}\end{array}\right.,\end{eqnarray} \tag{ 6 }$$

where n is the electron number density, n₀ is the density immediately ahead of the density jump, C is the density compression, r is the distance from the center of the sector, r_d is the radius at which the density jump is located, and α and β are the indices of the two power-laws. The fit is shown in Figure 7. If the density discontinuity is a shock front, then its magnitude corresponds to a shock with Mach number ${\mathcal{M}}={1.40}_{-0.12}^{+0.14}.$ Unfortunately, the count statistics are too poor to allow us to distinguish between a cold front and a shock front based on the temperature jump, and thus we cannot determine the nature of the surface brightness edge.

Zoom In Zoom Out Reset image size

The search for substructure in the ICM is motivated by the identification in the lensing maps of two less massive structures in addition to the main NE and SW subclusters (Jauzac et al. 2015). The positions of these mass structures are shown in Figure 8 and denoted by S1 and S2 for consistency with the notation of Jauzac et al. (2015). In the analysis of Jauzac et al. (2015), S1 and S2 were found to be X-ray-dark; however, the X-ray data presented here are ∼6 times deeper, which would make it easier to observe ICM substructure.

Zoom In Zoom Out Reset image size

We searched for substructure using the unsharp-masked image of the cluster, which was created by dividing the difference of two 0.5–4 keV fluxed images convolved with Gaussians of widths 4'' and 10'' by their sum. The resulting image, shown in Figure 8, highlights substructure on scales of ∼20–50 kpc.³⁰ There is no excess X-ray emission at the positions of S1 and S2. However, the emission is elongated in the direction of both mass structures. In the south, the ICM appears elongated in the direction of S1, while in the north the emission is elongated along the line connecting the NE and SW subclusters and then appears to curve in the direction of S2. We note that if S1 and S2 have already merged with the NE and SW subclusters, the DM would have decoupled from the gas, and therefore we do not necessarily expect a spatial overlap between the DM and gas components.

Interestingly, the unsharp-masked image enchances a small cavity with a diameter ∼50 kpc NW of the core of the NE subcluster. We examined the significance of the cavity from the azimuthal surface brightness profile in an annulus around the cluster center. The annulus was divided in 14 partial annuli with equal opening angles. The partial annuli and the azimuthal surface brightness profile are shown in Figure 9. The azimuthal surface brightness profile is lowest in four partial annuli located NW of the NE core. The largest dip is in the partial annulus that crosses the middle of the X-ray cavity seen in the unsharp-masked image; the opening angles of this partial annulus are 51°–77°. No radio emission fills the X-ray cavity, but the cavity was likely inflated by the AGN hosted by the NE BCG (see Section 8).

Zoom In Zoom Out Reset image size

The JVLA 1–2 GHz images reveal several compact sources in the cluster region (Figure 10). Two of these sources are associated with cD galaxies in the NE and SW subclusters. These sources are also detected in the GMRT 610 MHz image (Figure 11). These two point-like AGN have 1.5 GHz integrated flux densities³¹ of 1.47 ± 0.08 (NE) and 0.27 ± 0.03 (SW) mJy. At 610 MHz we measure flux densities of 3.24 ± 0.34 (NE) and 0.33 ± 0.07 (SW) mJy for these sources. Using these fluxes, we compute spectral indices of α = –0.88 ± 0.13 (NE) and α = –0.22 ± 0.27 (SW), which are typical for AGN (e.g., Prandoni et al. 2009).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We also find diffuse extended emission in the cluster. The diffuse emission reveals itself in the JVLA image as an increase in the "noise" in the general cluster area. This diffuse emission is better visible in our low-resolution tapered image with emission from compact sources subtracted (Figure 10). The diffuse emission has a elongated shape measuring about 120'' × 45'' (0.65 by 0.24 Mpc) and is oriented along a NE-SW axis, following the overall distribution of the X-ray emission. We also find evidence for this diffuse emission in the GMRT 610 MHz image, although less clearly than in the JVLA 1.5 GHz image (Figure 11). In the low-resolution tapered GMRT image there is again evidence for diffuse emission, but the peak flux is only at a level of 3σ_rms.

For the integrated flux of the diffuse emission, we measure 1.60 ± 0.14 mJy (at 1.5 GHz) in an ellipse with radii of 70'' by 40'' oriented with a position angle of 45° (following the overall brightness distribution). From the GMRT tapered image, we estimate an integrated 610 MHz flux of 6.8 ± 3.0 mJy for the diffuse emission in the same area. We evaluated the accuracy of the compact source subtraction and conclude that this does not add significantly to the uncertainty on the radio halo flux. Based on the residuals visible for the brightest source in the field, we find that the compact source subtraction was performed at a level better than 0.5%.

Taking the two flux measurements at 1.5 and 0.61 GHz, we obtain a spectral index of ${\alpha }_{1500}^{610}=-1.6\pm 0.5$ for the diffuse emission. Based on the above results, we compute a radio power of ${P}_{1.4\;{\rm{GHz}}}=(1.3\pm 0.3)\times {10}^{24}$ W Hz⁻¹, scaling with the spectral index of α = –1.6 ± 0.5.³²

VLITE shows emission co-incident with the NE subcluster where the higher resolution 1–2 GHz images (Figure 10) reveal the compact sources and the brightest portion of the diffuse emission. We fit the 320–360 MHz VLITE emission with a single Gaussian component to measure the integrated flux. We measure a total flux of 20.5 ± 7.0 mJy, where we have included a 13% uncertainty in the source flux, as determined for a low signal-to-noise source in VLITE (Clarke et al. 2015, in preparation). The VLITE emission contains both the diffuse component and the compact emission from the two point sources associated with the NE and SW clusters. We use the VLA and GMRT flux measurements to determine the spectral index of the two compact sources, assume that spectral index extends to the central VLITE frequency, and estimate the contribution to the total VLITE emission from the compact sources. We subtract that from the VLITE flux and get an estimate of the diffuse component detected by VLITE of 14.6 ± 7.0 mJy. Comparing the VLITE flux to the JVLA flux, we calculate a spectral index of ${\alpha }_{1500}^{340}=-1.5\pm 0.8.$ The results are consistent with the spectral index estimate from the GMRT data. Deeper low-frequency data are required to better constrain the spectral properties of the diffuse emission.

In the JVLA 1–2 GHz data we did not detect diffuse polarized emission in the cluster in Stokes Q and U images we made. Halos are generally unpolarized at the few percent level or less so this result is not surprising (e.g., Feretti et al. 2012). We note however, that a proper search for diffuse emission, taking full advantage of the large bandwidth, would require Faraday Rotation Measure Synthesis.

The combined X-ray and lensing analysis of Jauzac et al. (2015) determined that the gas component of the SW subcluster has decoupled from the DM component and lags behind the subcluster's galaxies as it travels toward the NE subcluster. In the NE subcluster, on the other hand, the DM and gas components were determined by Jauzac et al. (2015) to spatially overlap.³³ Offsets between the DM and gas components of merging galaxy clusters set crucial constraints on the merger state. In particular, significant offsets indicate a post-merging system, while a lack of offsets supports a pre-merging scenario. Below, we discuss the DM-gas offsets in light of the deeper Chandra data.

The centers of the DM halos of the two subclusters (M. Jauzac 2015, private communication) are marked on Figure 12, which shows an HST ACS/F814W image of the cluster, with overlaid Chandra and JVLA contours. The centers of the X-ray cores were defined as the peaks in the X-ray emission, and were determined from the 0.5 to 4 keV Chandra surface brightness map of the cluster, convolved with a Gaussian of σ = 4''. We also considered the uncertainties on the surface brightness distribution, and defined the uncertainties on the X-ray peaks as the regions around the cores in which the X-ray brightness is consistent (within the propagated Poisson errors on the counts image) with the brightness of the corresponding peak. We note that our approach only provides a lower limit on the extent of the peak regions, because the background was included in the surface brightness map used for this analysis. The uncertainties on the DM centers are <1'', significantly smaller than the uncertainties on the centers of the X-ray cores. In Figure 13, we compare the position of the DM centers with that of the X-ray peak regions. Rather than confirming the previously-reported offset between the gas and DM components of the SW subcluster, Figure 13 shows that both X-ray peaks are, within the uncertainties, co-located with the DM centers. We speculate that the discrepancy between the results presented herein and those reported by Jauzac et al. (2015) is caused by a higher noise level in the significantly shallower X-ray data used in previous studies; our data is ≈6 times deeper, which is equivalent to a reduction in noise by a factor of ≈2.5.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Jauzac et al. (2015) showed that the galaxy redshift distribution in MACS J0416.1-2403 is bimodal relative to the mean redshift, with the SW subcluster (mean redshift 0.3966) moving toward the observer and the NE subcluster (mean redshift 0.3990) moving away from the observer. Based on these observations, Jauzac et al. (2015) suggested two possible merger scenarios:

• 1.  
  MACS J0416.1-2403 is a pre-merging system, in which the SW subcluster comes from behind the NE subcluster, and is now seen near first core passage;
• 2.  
  MACS J0416.1-2403 is a post-merging system, in which the SW cluster approached the NE subcluster from above, and is now seen near its second core passage.

We refer the reader to Figure 13 of Jauzac et al. (2015) for a sketch of the two geometries. In this section, we try to distinguish between the two scenariosusing our X-ray and radio findings.

10.1. Cooler Gas in the NE?

10.2. No Gas-DM offset

10.3. Subcluster Morphology

The effects of ram pressure on the morphologies of two merging clusters also provide clues on the merger scenario. When two clusters merge, tails that trail the cluster cores may form. In a post-merging system seen before turnaround, these plasma tails are elongated toward the core of the opposite subcluster core. In a pre-merging system, the tails are elongated away from the opposite subcluster core. The two morphologies are sketched in Figure 14. In MACS J0416.1-2403, tails are not immediately seen in the surface brightness map. However, the unsharp-masked image in Figure 8 reveals that both the NE and the SW subclusters are trailed by tails whose directions are consistent with a pre-merger scenario.

Zoom In Zoom Out Reset image size

10.4. Substructure in the NE Core

10.5. The AGN and the X-Ray Cavity in the NE Core

10.6. SW Density Discontinuity

We classify the diffuse radio emission in the cluster as a halo, since the emission has low-surface brightness, large extent, and is centrally located. With a largest linear size (LLS) of 0.65 Mpc, the halo has a somewhat smaller size than other radio halos in merging clusters, which have LSSs of 1–1.5 Mpc. (e.g., Feretti et al. 2012). However, the LLS is larger than that of the radio mini-halos found in cool core clusters (e.g., Burns et al. 1992; Gitti et al. 2007; Giacintucci et al. 2014). Additionally, an important difference from mini-halos is that neither the NE nor the SW subclusters contains a cool core.Furthermore, as shown in Figure 12, the halo seems to be associated with both subclusters. The halo in MACS J0416.1-2403 is also rather elongated compared to most known radio halos. However, there are other clusters with similarly elongated radio halos (e.g., Bonafede et al. 2012; van Weeren et al. 2012a, 2012b; Shimwell et al. 2014).

The radio emission around the northern subcluster contains about a factor of 2 more flux than the emission around the southern one. An interesting question is whether we observe a single radio halo, caused by the merger between the NE and the SW subclusters, or two individual halos belonging to the two subclusters. In the latter case, these halos must have originated from previous merger events within two subclusters themselves. This would make MACS J0416.1-2403 similar to the case of the radio halo pair in Abell 399 and Abell 401 (Murgia et al. 2010), with the difference that Abell 399/401 seems to be at a much earlier pre-merger stage.

For radio halos, a correlation is observed between the cluster's X-ray luminosity and the radio halo power (the ${L}_{{\rm{X}}}-P$ correlation), with the most X-ray luminous clusters corresponding to the most powerful radio halos (e.g., Liang et al. 2000; Enßlin & Röttgering 2002; Cassano et al. 2006). The X-ray luminosity is used as a proxy of cluster mass. The distribution in the ${L}_{{\rm{X}}}-P$ plane is bimodal when clusters without halos are included. Merging clusters with radio halos follow the ${L}_{{\rm{X}}}-P$ correlation, while the upper limits on the radio power of other clusters are well below the correlation (Brunetti et al. 2007).

A correlation is also observed between the cluster's integrated Sunyaev–Zel'dovich (SZ) effect (i.e., the integrated Compton Y_SZ parameter) and the radio halo power (e.g., Basu 2012, 2013). Y_SZ is proportional to the volume integral of the pressure. The SZ signal should be less affected by the cluster dynamical state and therefore is expected to be a better indicator of the cluster's mass. Basu (2012, 2013) and Cassano et al. (2013) showed that there is good agreement between the Y_SZ – P scaling relation and the scaling relations based on X-ray data.

In Figure 15, we place MACS J0416.1-2403 on the ${L}_{{\rm{X}}}-P$ and ${Y}_{{\rm{SZ}}}-P$ diagrams, with the values for the other clusters taken from Cassano et al. (2013). The ${Y}_{\mathrm{SZ},500}$ value for the cluster was recently calculated by Planck Collaboration et al. (2015), which found ${Y}_{\mathrm{SZ},500}=(0.71\pm 0.24)\times {10}^{-4}$ Mpc². Specifically, this is the value of ${Y}_{\mathrm{SZ},500}$ obtained from the union catalog based on the MMF1³⁴ detection algorithm. We also computed the value of ${Y}_{\mathrm{SZ},500}$ using the gNFW model fit to Bolocam presented by Czakon et al. (2015) and obtained ${Y}_{\mathrm{SZ},500}=(1.49\pm 0.28)\times {10}^{-4}$ Mpc². The difference between the Planck and Bolocam measurements is not understood, but is likely due to some combination of random noise fluctuations, deviations from the assumed pressure profile used to extract ${Y}_{\mathrm{SZ},500}$ (particularly at large radii), contamination from dust or other astronomical signals, and calibration uncertainties. Furthermore, we note that this cluster is not detected by PowellSnakes—one of the three Planck algorithms—and the values of ${Y}_{\mathrm{SZ},500}$ obtained from the other two algorithms (MMF1 and MMF3³⁵ ) are significantly different, perhaps indicating that measurements of ${Y}_{\mathrm{SZ},500}$ toward this cluster contain larger than typical systematic uncertainties. In Figure 15, we plot both the Planck and Bolocam values of ${Y}_{\mathrm{SZ},500}.$

Zoom In Zoom Out Reset image size

Interestingly, the radio halo power falls on the low end of the ${L}_{{\rm{X}}}-P$ correlationfor radio halos without ultra-steep spectra. However, using the Planck measurement of ${Y}_{\mathrm{SZ},500},$ we find that the halo power is consistent with the expected ${Y}_{{\rm{SZ}}}-P$ correlation, indicating the halo is not underluminous for the cluster mass. While the Bolocam value of ${Y}_{\mathrm{SZ},500}$ does imply lower than expected halo power, the nominal ${Y}_{{\rm{SZ}}}-P$ relation used for comparison is derived solely from Planck SZ measurements. We therefore consider our interpretation based on the Planck value of ${Y}_{\mathrm{SZ},500}$ to be more robust. Given this interpretation, MACS J0416.1-2403 could be just an outlier in the ${L}_{{\rm{X}}}-P$ diagram, based on the fact that there is significant scatter in the correlations.

It is also possible that the X-ray luminosity of MACS J0416.1-2403 is not a good indicator of the cluster's mass, and the luminosity is boosted by the merger. A boost in X-ray luminosity is predicted for approximately one sound-crossing time, which for MACS J0416.1-2403 is ≈1 Gyr (Ricker & Sarazin 2001). This boost would affect other clusters on the correlation as well, but it could affect MACS J0416.1-2403 more since we are seeing a cluster close to first core passage. The mass of the cluster in R₅₀₀ is (7.0 ± 1.3) × 10¹⁴ M_⊙ (Umetsu et al. 2014). Based on the scaling relations of Pratt et al. (2009), the mass corresponds to a 0.1–2.4 keV cluster luminosity of (1.1 ± 0.4) × 10⁴⁵ erg s⁻¹, which is in agreement with the luminosity calculated from the Chandra data. Therefore, based on the ${M}_{{\rm{500}}}-{L}_{{\rm{500}}}$ scaling relation, we have no indication that the cluster luminosity is higher than expected for a cluster of this mass.

Another possibility is that we are seeing an ultra-steep spectrum radio halo (α ≲ −1.5, Brunetti et al. 2008). In this case, the radio halo is only underluminous at higher frequencies (i.e., above a GHz), while the integrated flux density rapidly increases at lower frequencies. Ultra-steep spectrum radio halos are predicted by the turbulent re-acceleration model to form in less energetic mergers (Brunetti et al. 2008). The mass of MACS J0416.1-2403 is indeed near the lower end of the mass range of clusters with radio halos (see, e.g., Cassano et al. 2013, who list masses between 4.4 × 10¹⁴ M_⊙ and 1.7 × 10¹⁵ M_⊙ for clusters hosting radio halos). Some, but not all clusters with masses similar to that of MACS J0416.1-2403 host ultra-steep spectrum radio halos. The steepness of the radio spectrum is affected not only by the cluster mass though, but also by the magnetic fields of the clusters and by the particle acceleration efficiency; at the moment, these are not known.

An ultra-steep spectrum radio halo is also expected in the later phases of a merger, when the radio spectrum had time to steepen due to synchrotron and IC losses (e.g., Donnert et al. 2013). If the diffuse radio emission in MACS J0416.1-2403 is ultra-steep and the steepness was caused by ageing, then the emission in the cluster is in fact consisting of two individual radio halos that were formed in the NE and SW subclusters during previous mergers—possibly the same mergers that heated the cluster cores.

Unfortuntely, the 320–360 MHz VLITE and 610 MHz GMRT data presented in Section 8 are not sufficient to confirm, or rule out, the presence of an ultra-steep spectrum radio halo. Low-frequency GMRT (ν < 610 MHz) and/or deep P-band JVLA observations are required to settle this case.

A third possibility is that we could be witnessing the first stages of the formation of this radio halo. We do not think that the radio halo is fading, because the NE and SW subclusters are still in the process of merging and most likely they have not yet undergone core passage (see Section 10). Therefore, when the merger progresses further, the radio halo could brighten, increase in size, and move onto the correlation.

The HST Frontier Field cluster MACS J0416.1-2403 (z = 0.396) is a merging system that consists of two main subclusters and two additional smaller mass structures (Jauzac et al. 2015). Here, we presented results from deep Chandra JVLA, and GMRT observations of the cluster. The main aims of our analysis were to identify signatures of merger activity in the ICM, constrain the merger scenario, and detect and characterize diffuse radio sources. Below is a summary of our main findings:

• 1.  
  The cluster has an average temperature of ${10.06}_{-0.49}^{+0.50}$ keV, an average metallicity of ${0.24}_{-0.04}^{+0.05}$ Z_⊙, and a 0.1–2.4 keV luminosity of ${L}_{{\rm{X}}}=(7.43\pm 0.08)\times {10}^{44}$ erg s⁻¹.
• 2.  
  The NE subcluster has a very compact core and a nearby X-ray cavity, but its temperature is very high (T ∼ 10 keV), we find no evidence of multi-phase gas, and its cooling time is ≳3 Gyr. The presence of a compact core and of an X-ray cavity indicates that the NE subcluster was not disrupted by a recent major merger event.
• 3.  
  S-SW of the SW subcluster, there is a weak density discontinuity that is best-fitted by a density compression of ${1.56}_{-0.29}^{+0.38}.$ The discontinuity is located along the line connecting the NE subcluster, the SW subcluster, and the mass structure S1 reported by Jauzac et al. (2015). Most likely, the discontinuity was caused by a collision between the SW subcluster and S1, rather than by a collision between the NE and SW subclusters.
• 4.  
  The DM and gas components of the NE and SW subclusters are well-aligned, which favors a scenario in which the two subclusters are seen before first core passage.
• 5.  
  MACS J0416.1-2403 has a radio halo with a 1.4 GHz power of $(1.3\pm 0.3)\times {10}^{24}$ W Hz⁻¹, which is somewhat less luminous than predicted by the ${L}_{{\rm{X}}}-P$ correlationfor radio halos without ultra-steep spectra. However, the halo aligns well on the ${Y}_{\mathrm{SZ},500}-P$ correlation, indicating that it is not underluminous for the cluster mass.We could be observing the birth of a giant halo, or a very rare ultra-steep halo.

We thank the referee for constructive comments. We also thank Mathilde Jauzac for providing the coordinates of the centers of the DM halos, and Becky Canning and Aurora Simionescu for constructive discussions.

G.A.O. acknowledges support by NASA through a Hubble Fellowship grant HST-HF2-51345.001-A awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Incorporated, under NASA contract NAS5-26555. R.J.vW. is supported by NASA through the Einstein Postdoctoral grant number PF2-130104 awarded by the Chandra X-ray Center, which is operated by the Smithsonian Astrophysical Observatory for NASA, under contract NAS8-03060. A.Z. acknowledges support by NASA through a Hubble Fellowship grant HST-HF2-51334.001-A awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Incorporated, under NASA contract NAS5-26555. This research was performed while T.M. held a National Research Council Research Associateship Award at the Naval Research Laboratory (NRL). Basic research in radio astronomy at NRL by T.M. and T.E.C. is supported by 6.1 Base funding. F.A.S. acknowledges support from Chandra grant G03-14131X.

The scientific results reported in this article are based on observations made by the Chandra X-ray Observatory. Part of the reported results are based on observations made with the NASA/ESA Hubble Space Telescope, obtained from the Data Archive at the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-26555. The HST observations are associated with program #13496. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. We thank the staff of the GMRT that made these observations possible. The GMRT is run by the National Centre for Radio Astrophysics of the Tata Institute of Fundamental Research.

This research has made use of NASA's Astrophysics Data System, and 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. Some of the cosmological parameters in this paper were calculated using Ned Wright's cosmology calculator (Wright 2006). This research made use of APLpy, an open-source plotting package for Python hosted at http://aplpy.github.com.

Facilities: CXO (ACIS) - , HST (ACS) - .