1: %auto-ignore
2:
3:
4: \paragraphfont{\small}
5: \subsectionfont{\normalsize}
6:
7:
8:
9: \newcommand\be{\begin{equation}}
10: \newcommand\ee{\end{equation}}
11: \renewcommand{\vec}[1]{ {{\bf #1}} }
12:
13: \renewcommand{\textfraction}{0}
14: \renewcommand{\floatpagefraction}{1.0}
15: \renewcommand{\topfraction}{1.0}
16: \renewcommand{\bottomfraction}{1.0}
17:
18: \renewcommand\caption[1]{%
19: \myCaption{#1}
20: }
21:
22: \title{\vspace*{-1cm}{\Large A low-frequency radio halo associated with a
23: cluster of galaxies}\vspace*{0.2cm} \\ {\em \large Supplementary Information}\vspace*{0.3cm}}
24:
25: \author{\parbox{13.5cm}{\small\sffamily%
26: G. Brunetti$^1$,%
27: S. Giacintucci$^{1,2}$,%
28: R. Cassano$^{1}$,%
29: W. Lane$^{3}$, %
30: D. Dallacasa$^{4}$,%
31: T. Venturi$^{1}$, %
32: N. Kassim$^{3}$, %
33: G. Setti$^{1,4}$, %
34: W. D. Cotton$^{5}$,\& %
35: M. Markevitch$^{2}$}\vspace*{-0.5cm}}
36:
37:
38: \renewcommand\refname{\large References}
39:
40: \date{}
41:
42: \bibliographystyle{naturemag}
43:
44:
45: \baselineskip14pt % On-and-half-space the manuscript.
46: \setlength{\parskip}{4pt}
47: \setlength{\parindent}{18pt}%
48:
49: \twocolumn
50:
51: \setlength{\footskip}{25pt}
52: \setlength{\textheight}{670pt}
53: \setlength{\oddsidemargin}{-8pt}
54: \setlength{\topmargin}{-41pt}
55: \setlength{\headsep}{18pt}
56: \setlength{\textwidth}{469pt}
57: \setlength{\marginparwidth}{42pt}
58: \setlength{\marginparpush}{5pt}
59:
60: \addtolength{\topmargin}{-0.6cm}
61:
62:
63: \maketitle
64:
65:
66: \renewcommand{\thefootnote}{\arabic{footnote}}
67: \footnotetext[1]{\footnotesize INAF - Istituto di Radioastronomia, Via P. Gobetti 101, I-40129
68: Bologna, Italy}
69: \footnotetext[2]{\footnotesize Harvard-Smithsonian Center for Astrophysics, Cambridge,
70: MA 02138, USA}
71: \footnotetext[3]{\footnotesize Naval Research Laboratory, Code 7213, Washington, DC 20375-5320, USA}
72: \footnotetext[4]{\footnotesize Dipartimento di Astronomia, Universita' di Bologna, Via Ranzani 1,
73: I-40127 Bologna, Italy}
74: \footnotetext[5]{\footnotesize National Radio Astronomy Observatory, Charlottesville, Virginia
75: 22903-2475, USA}
76:
77:
78:
79: {\bf\small
80: To complement the information given in the main body of the Letter,
81: these Sections provide supplementary discussion on the
82: methods adopted to derive upper limits to the flux density
83: of the halo reported
84: in Figure 2 (Sect.~1) and on the origin of the halo (Sect.~2 and 3). \vspace*{-0.3cm}}
85:
86: \renewcommand{\thefootnote}{\fnsymbol{footnote}}
87:
88: \small
89:
90:
91: \subsection*{Upper limits to the flux of the radio halo at 1.4 GHz}
92:
93: The diffuse emission in the center of Abell 521
94: discovered by the low frequency radio observations
95: presented in this Letter was not detected in VLA observations at 1.4
96: GHz$^{[17]}$.
97: We re-analyzed those BnC VLA data taken from the archive.
98: Given the sourthern declination of the cluster, they can
99: be considered equivalent to a C configuration observation. The quality
100: of the data is generally good, except for occasional intermittent
101: RFI in IF2 (1.44 GHz) and at the very end of the observation (both
102: 25-MHz IFs affected), when the source and its secondary calibrator were
103: observed at low elevation and data on a few short baselines had
104: corrupted amplitudes (very high circular polarization).
105: RFI excision was carried out with great care.
106: We also paid particular attention to the analysis of the secondary
107: calibrator data, whose visibility turned out to be affected
108: by substantial contribution
109: from sources in the field. Therefore we first followed the standard
110: calibration procedure, then we used the image of the secondary
111: calibrator obtained in this way to improve the a-priori amplitude (and
112: phase) calibration accuracy. The final dataset was obtained after a
113: number of phase-only self-calibrations.
114: Solutions had generally very high signal to noise ratio.
115:
116: The final images were obtained by considering a central square,
117: 0.8$^\circ$ wide, with additional 10 smaller (0.1$^\circ$) facets on
118: confusing sources.
119: In the full resolution image (12.7 arcsec $\times$6.9 arcsec in
120: p.a. 75$^\circ$) the r.m.s. noise level is 15 $\mu$Jy/beam.
121: We also produced images with lower resolution by applying natural
122: weighting to the data and also using a restricted uv-range.
123:
124: A low resolution 30 arcsec$\times$30 arcsec image of the region of Abell 521
125: is shown in Figure S1.
126: Suspicious, very faint, diffuse emission appears in the region
127: of the radio halo discovered at lower frequencies; however,
128: most of this emission is due to the convolution of point sources in
129: the same region (the image after subtraction of point sources is given
130: in Figure 1) and the detection of the radio halo in Abell 521 at 1.4 GHz
131: would not have been firmly established from these observations only.
132:
133: \begin{figure*}
134: \begin{center}
135: \resizebox{\hsize}{!}{\includegraphics{brunetti_fS1.ps}}
136: %\resizebox{180mm}{!}{\rotatebox{0}{\includegraphics*{brunetti_fS1.ps}}}
137: \end{center}
138: \caption[]{
139: {\bf Low resolution images of Abell 521 at 1.4 GHz.}
140: The upper panel shows the image from our analysis of the original data,
141: while lower panels show images from the analysis of the data containing
142: injected fake radio halos with $\Theta_{50\%} =$2.15 arcmin and $F$= 4 (left)
143: and 6 (right) mJy.
144: Crosses mark the position of the point sources identified in the high
145: resolution image in the central region of Abell 521 and in the region
146: where fake radio halos have been injected.
147: Radio contours are at 80, 160, 240, .. $\mu$Jy/beam level;
148: negative contours at -80$\mu$Jy/beam level are reported in dashed lines.
149: The r.m.s. noise level in the images is in the range 28-30 $\mu$Jy/beam
150: and the restoring beam is 30 arcsec $\times$ 30 arcsec.}
151: \end{figure*}
152:
153: We thus estimated the upper limit to the flux density of the radio halo
154: in these data at 1.4 GHz.
155: This should account also for the possibility that a fraction of the
156: flux density on the halo scale may be lost due to the
157: uv coverage at short baselines of the observations.
158: In order to derive a solid and conservative upper limit we
159: follow$^{[7,15]}$. We injected fake halos in
160: the uv--components of the dataset by means of the AIPS task UVMOD$^{[31]}$
161: and imaged the datasets
162: with the same set of parameters applied to the original data.
163: This procedure allows to evaluate the response of the 1.4 GHz VLA observation
164: to the presence of diffuse emission on the halo scale, as the uv
165: components of the fake radio emission are sampled by means of the real
166: uv coverage of the observation.
167:
168: In particular,
169: we model the brightness profile of the fake halos with sets of optically thin
170: spheres with different radius and flux density, and obtain {\it families}
171: of fake radio halos with total flux densities ranging from 3 to 15 mJy,
172: with sizescale enclosing 50\% of the flux
173: $\Theta_{50\%} =$1.9 -- 2.3 arcmin ($\approx 440$ and $530$ kpc, respectively)
174: and maximum
175: angular extension $\approx 5.2$ arcmin ($\approx 1.2$ Mpc).
176:
177: Figure S1 shows the results obtained for fake radio halos with
178: $\Theta_{50\%} =$2.15 arcmin and injected flux density $F$= 4 and 6 mJy.
179: In the case of the halo with $F$= 4 mJy,
180: about half of the injected flux is recovered
181: by the analysis of the dataset, and the resulting diffuse emission
182: in the image is still brighter than that at the center of Abell 521.
183:
184: The analysis of the {\it families} of datasets with fake radio halos
185: allowed us to derive a conservative upper limit to the flux density at
186: 1.4 GHz of the radio halo in Abell 521, $F < 5$ mJy.
187: This should be compared with the residual diffuse emission at
188: $F \approx 1-2$ mJy level measured in the region of the radio halo
189: after subtraction of discrete sources (Figure S1) in our images.
190: Because the flux density of these sources, $F=3-4$ mJy, is not negligible at
191: 1.4 GHz, they were fitted in the full resolution image (with beam =
192: 12.7 arcsec x 6.9 arcsec and r.m.s. =15 $\mu$Jy/beam) by using the AIPS
193: task JMFIT solving for a Gaussian component (which has then been subtracted
194: out) and a non-zero level (i.e. diffuse emission, which is not subtracted).
195:
196:
197:
198: \subsection*{Projection effects and the radio halo}
199:
200: Ghost radio plasma, confined within the intergalactic
201: medium, may be revived by the passage of a shock wave through adiabatic
202: compression$^{32}$; indeed this mechanism has been proposed
203: to explain radio relics as a scenario alternative to shock acceleration.
204:
205: In principle, a large scale shock crossing Abell 521 along the line of sight
206: may compress ghost radio plasma producing detectable synchrotron
207: radiation with steep spectrum projected on the cluster central region, that
208: may be classified as radio halo.
209: However, the steep spectrum of the halo in Abell 521,
210: its large ($\approx$ Mpc) extension and - in particular -
211: the morphological correlation
212: between radio halo and the X-ray emission from the cluster can be used
213: to conclude that this scenario is very unlikely.
214:
215: In this context we have to assume that
216: the bubble of ghost plasma extends on the Mpc scale.
217: This implies a dynamical age of this plasma
218: $\geq$ Gyr during which relativistic electrons lose energy via
219: radiative and adiabatic losses. This results in a cut-off in the synchrotron
220: spectrum emitted by the ghost plasma at very low frequencies, $\nu_{b,o}
221: \leq$MHz, before the shock--compression phase.
222: The passage of the shock
223: compresses the ghost plasma by a factor$^{32}$ $x = (P_{ds}/P_{us})^{3/4}$,
224: where $P_{ds}$ and $P_{us}$ are the pressures measured
225: in the thermal medium around the ghost plasma downstream
226: and upstream the shock, respectively.
227: The adiabatic compression boosts the frequency of the cut-off
228: of the synchrotron spectrum at$^{33}$
229: $\nu_c \approx \nu_{c,o} x^{4/3} \approx P_{ds}/P_{us}$.
230: In order to have $\nu_c \approx 200-300$ MHz as measured
231: in the spectrum of the radio halo in Abell 521, this
232: would require $P_{ds}/P_{us} > 300$
233: implying a Mach number of the shock (from the relation between pressure
234: jump and Mach number$^{34}$) :
235:
236: $$M = {1\over{\sqrt{5}}}
237: \left(4 {{P_{ds}}\over{P_{us} }} +1 \right)^{1/2} > 15
238: \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, (S.1)
239: $$
240:
241: Strong large-scale shocks with $M > 10$ are rare in galaxy
242: clusters$^{35,36,37}$ and may occur only outside the virialised region
243: of the cluster where the temperature of the gas is well below 1 keV.
244: Thus the compression of the ghost plasma should happen at $\geq 3$ Mpc
245: distance from the center of Abell 521 (provided that a ghost-Gyr old
246: plasma at such distances manages to survive from mixing with
247: the intergalactic medium).
248:
249: At the same time the radio halo is correlated with the cluster
250: X-ray emission and its radio brightness decreases by a factor $\approx$6-8
251: on the scale of the core of Abell 521.
252: In order to reproduce this morphological connection by chance
253: the ghost radio
254: bubble, at a distance of $\geq 3$ Mpc from the cluster center, should
255: be aligned with the center of Abell 521 within a projected distance
256: of a cluster core-radius, $\approx 300$ kpc.
257: Geometrically, this implies that the direction connecting Abell 521
258: and the ghost bubble should make an angle $\theta
259: \leq 7^o$ to the line of sight and the
260: chance probability for that, $\approx 1 - \cos(\theta)$, is $<$ 1\%.
261:
262: The {\it ad hoc} assumptions and this low probability
263: allow us to conclude that this scenario is very unlikely.
264:
265:
266:
267:
268: \subsection*{On the size of the radio relic in Abell 521}
269:
270: The radio halo in Abell 521 appears connected with the
271: radio relic (Figure 1).
272: This is similar to that observed in the case of
273: Abell 520$^{33}$ and Abell 2256$^{38}$; a bridge connecting the halo
274: and the relic is also observed in the Coma cluster$^{39}$.
275:
276: Radio relics are interpreted as the result of the acceleration
277: of relativistic particles due to large scale shocks that
278: form in the intergalactic medium during cluster mergers
279: or accretion of matter$^{40,41}$.
280: The relic in Abell 521 is found to coincide with a possible
281: shock front with Mach number $M \approx 2-4$, generated by the recent infall
282: of a subcluster along the North-West/South-East direction$^{[18]}$.
283:
284: Electrons accelerated at the shock flow away into the
285: intergalactic medium in a tail of emitting plasma,
286: and one might expect that part of the emission in the
287: region of the radio halo is due to this tail.
288: However in the following we show that this is unplausible : the extension
289: of the tail at a given frequency is constrained by the lifetime
290: of the electrons emitting at that frequency (that depends on inverse
291: Compton and synchrotron losses, and on adiabatic losses),
292: which turns out to be much smaller than the $\geq$Mpc projected
293: distance between the relic and halo region.
294:
295:
296: We adopt the most favorable geometry assuming that the
297: shock is moving on the plane of the sky along the
298: North-West/South-East direction,
299: in which case the distance that separates the relic and halo region
300: is not affected by projection effects.
301: We also assume that expansion in the shocked gas
302: plays a role on a timescale longer than the radiative timescale of the
303: electrons emitting at 240 MHz, that is $\approx 0.2$ Gyr (Eq.1), and
304: thus we neglect adiabatic losses of electrons.
305: This last assumption is justified since the gas in the
306: downstream region, over-pressured with respect to the ambient gas,
307: should expand in a timescale that is a fraction of
308: $\approx \phi_{r} / c_s \approx 0.5$ Gyr,
309: where $\phi_r \approx 500-700$ kpc is the longitudinal
310: size of the relic$^{18}$ and $c_s$ the sound velocity.
311:
312: Under these assumptions,
313: the size of the tail at a given frequency
314: can be estimated as $l_{tail} \approx v_{ds} \tau_r$, where $\tau_r$
315: is the radiative
316: lifetime of electrons emitting at that frequency, and $v_{ds}$ is
317: the downstream velocity of the gas (in the shock frame, from the jump
318: conditions$^{34}$) :
319:
320: $$
321: v_{ds} = c_s^u {{M^2 + 3}\over{4 M}}
322: %\simeq
323: %1700 \sqrt{ {{M^2+3}\over{5 M^2 -1}} }
324: \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, (S.2)
325: $$
326:
327: where $c_s^u$ is the sound velocity in the intergalactic medium
328: upstream of the shock.
329: For a given Mach number of the shock $c_s^u$
330: can be estimated from the temperature of Abell 521
331: (taken as downstream temperature, and from the relations
332: between temperature jump and Mach number$^{34}$) :
333:
334: $$
335: c_s^u \approx 1700 \sqrt{ {{16 M^2}\over{(5 M^2 -1)(M^2+3)}} }
336: \,\,\,\,\,\, {\rm (km/s)} \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, (S.3)
337: $$
338:
339: Combining S.2--S.3 and adopting $\tau_r \approx 0.2$ Gyr for the electrons
340: emitting at 240 MHz (Eq.~1), we obtain
341: $l_{tail} \approx 200$ kpc if the relic in Abell 521 in generated by a shock
342: with $M \approx 2-4$.
343: %On the other hand, if
344: %the shocked gas expands on a timescale shorter than
345: %the radiative timescale of electrons, the velocity of separation
346: %between the shock wave and the relativistic plasma that is accelerated at
347: %the shock is larger than that in
348: %Eq.~S.2 and, as soon as the flow of the shocked gas decelerates, it
349: %gradually approaches the shock velocity (upstream velocity),
350: %that is 2-3 times larger than the downstream
351: %velocity for shocks with $M = 2-4$.
352: %In this case however, adiabatic losses strongly
353: %suppress the synchrotron emissivity (and decrease the lifetime
354: %of electrons) so that the scale-length
355: %of the tail is expected to be comparable
356: %to (or even smaller than) that in the previous case.
357:
358: \noindent
359: Remarkably, $l_{tail} \approx 200$ kpc agrees with the
360: transverse size of the
361: relic at 240 and 330 MHz, and with the size of the synchrotron spectral
362: steepening in the relic as measured from radio observations$^{18}$.
363:
364: \bibliography{si}
365:
366:
367:
368: