0705.1978/ms.tex
1: % paper on Chandra and VLA observations of NGC 5408
2: %\documentclass{aastex}
3: %\usepackage{emulateapj5,apjfonts}
4: 
5: % format needed for submission
6: \documentclass[12pt,preprint]{aastex}
7: % preprint2 produces a double-column, single-spaced document:
8: %\documentclass{aastex}
9: %\usepackage{psfig}
10: 
11: % You can insert a short comment on the title page using the command below.
12: \slugcomment{submitted to ApJ}
13: 
14: \shorttitle{ULX in NGC 5408}
15: \shortauthors{Lang et al.}
16: 
17: \begin{document}
18: 
19: \title{A Radio Nebula Surrounding the Ultraluminous X-Ray Source in NGC 5408}
20: 
21: \author{Cornelia C.\ Lang, Philip Kaaret} 
22: \affil{Department of Physics and Astronomy}
23: \affil{University of Iowa,  Van Allen Hall, Iowa City, IA 52242, USA}
24: \email{cornelia-lang@uiowa.edu}
25: \author{St\'ephane Corbel} 
26: \affil{AIM$-$Unit\'e Mixte de Recherche CEA$-$CNRS}
27: \affil{Universit\'e Paris VII$-$UMR 7158, CEA Saclay, Service d'Astrophysique, F$-$491191 Gif sur Yvette, France.}
28: \and
29: \author{Allison Mercer}
30: \affil{Department of Physics and Astronomy}
31: \affil{University of Iowa,  Van Allen Hall, Iowa City, IA 52242, USA}
32: 
33: \begin{abstract}
34: 
35: New radio observations of the counterpart of the ultraluminous X-ray
36: source in NGC 5408 show for the first time that the radio emission is 
37: resolved with an angular size of $1.5\arcsec$ to $2.0\arcsec$. This
38: corresponds to a physical size of 35--46~pc, and rules out interpretation 
39: of the radio emission as beamed emission from a relativistic jet. In addition, the
40: radio spectral index of the counterpart is well determined from three
41: frequencies and found to be $\alpha$=$-$0.8 $\pm$0.2. The radio emission 
42: is likely to be optically-thin synchrotron emission
43: from a nebula surrounding the X-ray source.  The radio luminosity of
44: the counterpart is $3.8 \times 10^{34} \rm \, erg \, s^{-1}$ and the minimum energy
45: required to power the nebula is $\sim 1 \times 10^{49} \rm \, erg$.  These
46: values are two orders of magnitude larger than in any Galactic nebula powered
47: by an accreting compact object.
48: 
49: \end{abstract}
50: 
51: \keywords{black hole physics -- galaxies: individual: NGC 5408
52: galaxies: stellar content -- X-rays: galaxies -- X-rays: black holes}
53: 
54: 
55: \section{Introduction}
56: 
57: Ultraluminous x-ray sources (ULXs) are non-nuclear x-ray sources in
58: external galaxies with apparent luminosities above the Eddington limit
59: for a $20 M_{\odot}$ black hole, the maximum mass
60: of any dynamically-measured black hole mass in the Galaxy \citep{remillard06}; 
61: for a review, see \citet{fabbiano06}. 
62: ULXs with strong variability are likely accreting objects and may either be
63: ``intermediate mass'' black holes
64: \citep{colbert99,makishima00,kaaret01} or normal (stellar-mass) black
65: holes with beamed or super Eddington radiation
66: \citep{king01,kording02,begelman02}.  The ULX in the dwarf irregular
67: galaxy NGC 5408 (NGC 5408 X-1) is variable and is one of the few ULXs
68: with a known radio counterpart \citep{kaaret03}.  The radio emission
69: could arise directly from a relativistic jet beamed toward our line of
70: sight, in which case a stellar-mass black hole would suffice to produce
71: the radio and X-ray emission, or from a nebula surrounding the compact
72: object. Recent radio observations presented by \citet{soria06} suggest
73: that the source has a steep spectrum ($\alpha$ $<$ $-$1), but show no 
74: indication of radio flux density variability. 
75: 
76: In order to better understand the nature of the radio emission from NGC
77: 5408 X-1,  we obtained new joint observations in the radio using the Very
78: Large Array (VLA) of the National Radio Astronomy
79: Observatory\footnotemark\footnotetext{The National Radio Astronomy 
80: Observatory is a facility of the National Science Foundation operated
81: under cooperative agreement by Associated Universities, Inc.} and in 
82: the X-ray using the {\it Chandra X-Ray Observatory}. In addition, because
83: the target is a low-declination source ($\delta$ $\sim$ $-$41\arcdeg), we
84: present new and archival radio observations from the Australia Telescope Compact
85: Array (ATCA). Finally, we also reanalyzed archival optical
86: observations made using the {\it Hubble Space Telescope} (HST).  We describe
87: the observations and data reduction in \S~2,  present our results in
88: \S~3, and discuss the interpretation in \S~4.
89: 
90: 
91: \section{Observations and Analysis}
92: 
93: \subsection{VLA Observations}
94: 
95: VLA observations of the ULX source in NGC 5408 were made at 4.9~GHz
96: (6~cm) with four different array configurations between December 2003 and January 2005
97: : A, BnA, B, and CnB (see Table~\ref{obstable}). Each observation was approximately
98: 3.5~hours long and corresponds to a joint {\it Chandra} X-ray 
99: observation. Standard procedures were carried out for flux and phase
100: calibration using the Astronomical Imaging Processing System (AIPS) 
101: of the NRAO. We used the flux density calibrator, J1331+305, and the phase
102: calibrator, J1353-441, a source which is located within $3\arcdeg$ of the target.  
103: For all observations, fast switching between the target source and
104: calibrator was used with a cycle time for the calibrator-source pair of
105: 2.5 to 3 minutes (with 45 seconds on the calibrator and 90
106: to 120 seconds on the source).
107:   
108: The VLA image resolutions vary between $1.01\arcsec \times 0.28\arcsec$ (A-array)
109: and $6.01\arcsec \times 3.30\arcsec$ (CnB-array).  Because of the
110: proximity of the bright starburst region in NGC5408 and the weak radio emission
111: associated with the ULX, obtaining an accurate flux density for this
112: source is difficult.  However, with the high resolution of the VLA A, BnA and
113: B-arrays, it is possible to distinguish the radio emission associated
114: with the ULX from the extended emission from the starburst region in
115: most of our images.  
116: 
117: In fact, in attempting to determine the accurate flux density and
118: source structure, the extended emission from this starburst 
119: region turns out to be a key issue. One way to examine the contribution of the extended
120: emission is to adjust the imaging weighting function. The  Briggs'
121: ROBUST parameter controls the data weights in the {\it (u,v)} plane. 
122: Positive values of the ROBUST parameter (1 to 5) bring out the more
123: extended structure by weighting the inner part of the {\it (u,v)} plane
124: more heavily (``natural weighting''). Such values will increase the signal 
125: to noise on extended features but lower the spatial resolution. 
126: Negative values of the ROBUST parameter ($-$1 to $-$5) give more equal
127: weight to all points in the {\it (u,v)} plane (``uniform weighting'') which tends
128: to maximize spatial resolution but at the cost of signal to noise. 
129: A ROBUST=0 is balanced between these extremes. As many four images were made 
130: from data for an individual array using varying ROBUST parameters. 
131: 
132: In addition, a more quantitative way to remove extended flux is 
133: to limit the range of {\it (u,v)} baselines used for imaging. In
134: cases where the emission from the ULX is contaminated by extended
135: flux from the starburst, a {\it (u,v)} cutoff of 5 k$\lambda$ was 
136: applied. This cutoff limits flux on angular size scales larger
137: than $\sim$6\arcsec. For the VLA-B and VLA-CnB array images, where extended
138: flux is present, such cutoffs have been applied. 
139: 
140: 
141: \subsection{ATCA Observations}
142: 
143: We also made continuum radio observations of NGC 5408 using the
144: Australia Telescope Compact Array (ATCA) located in Narrabri, New South
145: Wales, Australia.  The ATCA consists of five 22-m antennas positioned
146: along an E-W track with a short N-S spur and a sixth
147: antenna at a fixed location. We obtained three new ATCA observations 
148: of 12 hours each on consecutive days with the 6~km configuration 6.0A (6A). 
149: This provides the best angular resolution possible with ATCA ($\sim$1\arcsec~
150: at 4.8 GHz).  For the first two observations, we switched the feed-horns at roughly one hour
151: intervals observing in either the 6~cm band (4.8 GHz) or the
152: 13 and 20~cm bands (1.4 and 2.4 GHz). The duty cycle of the
153: switching was arranged so the integration at 6~cm was four times as
154: long as those at  13 and 20~cm.  On the last day, we observed with the 
155: 5 cm band (6.1 GHz) instead of the 6 cm band (4.8 GHz). 
156: 
157: J1934-638 and J1349-439 were used for flux and phase calibration.
158: Editing, calibration, imaging and analysis were performed using 
159: standard routines in the MIRIAD software package
160: \citep{sault98}. A radio counterpart to the ULX was detected at
161: all frequencies.  Because of the low declination of NGC 5408, the
162: ATCA forms a more symmetric synthesized beam than the VLA. The
163: extended emission associated with the starburst in NGC 5408 is present,
164: but in the ATCA images, the radio counterpart to the ULX is 
165: clearly distinguished, especially in the 6A configuration
166: data at 4.8, 6.2 GHz. However, at the lower frequencies (1.4 and 2.3 GHz) 
167: there is still contamination from the starburst. During imaging, a variety 
168: of ROBUST parameters were also used to
169: downweight the extended emission, similar to the procedure in the 
170: VLA imaging. In addition, we analyzed 
171: archival ATCA observations 
172: of NGC 5408 at a range of frequencies, observed during 2000-2004 
173: (listed in Table~\ref{atca_obs}). 
174: In the several of the ATCA datasets, (e.g., the 6D-array 4.9 GHz
175: data (Kaaret et al. (2003), and the 2.3 and 1.4 GHz archival datasets)  
176: a {\it (u,v)} cutoff range of 3 k$\lambda$ was used to remove flux on 
177: extended scales larger than 11\arcsec. 
178: 
179: \subsection{X-ray Observations}
180: 
181: NGC 5408 was observed with the {\it Chandra X-Ray Observatory}
182: (Weisskopf 1988) five times using the ACIS spectroscopic array (ACIS-S;
183: Bautz et al.\ 1998) in imaging mode and the High-Resolution Mirror
184: Assembly (HRMA; van Speybroeck et al.\ 1997). The observations began in
185: 2002 and ended in 2005, see Table~\ref{obstable}.  Because a high X-ray
186: count rate was expected,  only the S3 chip was operated with a 1/4
187: sub-array mode for the first observation and a 1/8 sub-array for the
188: other observations.  This gave an exposure time of 0.8~s for the first
189: observation and 0.4~s for the others.  For each observation, we constructed 
190: an image using all valid events on
191: the S3 chip and used the {\it wavdetect} tool which is part of the {\it
192: CIAO} data analysis package to search for X-ray sources.  Typically
193: only the ULX was detected in each observation.   The position of the
194: ULX was within $0.23\arcsec$ of RA = 14h03m19.63s $\pm$ 0.01s, DEC
195: -41d22'58.65" $\pm 0.2\arcsec$ (J2000) in all observations.
196: 
197: We fit the X-ray spectrum of the ULX for each observation using the
198: {\it XSPEC} spectral fitting tool which is part of the {\it LHEASOFT}
199: X-ray data analysis package and response matrices calculated using {\it
200: CIAO}.  As previously reported by \citet{kaaret03}, we found that
201: absorbed single power-law models did not provide an adequate fits
202: (except for the last observation), while models consisting of either an
203: absorbed broken power-law or the absorbed sum of a multicolor disk
204: blackbody plus a power-law did provide adequate fits.  There were no
205: pronounced changes in spectra shape between the different
206: observations.  The fit parameters were similar to those quoted in
207: \citet{kaaret03}.  The flux for each observation calculated using the
208: fitted multicolor disk blackbody plus power-law model are given in
209: Table~\ref{obstable}. We also did not observe any significant changes
210: in the X-ray flux level between observations. 
211: 
212: 
213: \subsection{HST Observations}
214: 
215: Two WFPC2 observations of NGC 5408 are present in the {\it Hubble Space
216: Telescope} (HST) archive.  The observations were obtained as part of a
217: snapshot survey of nearby dwarf galaxy candidates (GO-8601, PI Seitzer)
218: and consist of 600~s WFPC2 exposures with the F606W and F814W filters
219: obtained on 4 July 2000.  After extracting the data from the HST
220: archive, we used IRAF to mosaic and clean the images.  We corrected the
221: absolute astrometry of the HST image using stars from the 2MASS catalog
222: \citep{2mass} using the Graphical Astronomy and Image Analysis Tool
223: (GAIA).  We used only 2MASS sources on the WF3 chips where the ULX is
224: located in order to preclude issues regarding the relative positioning
225: of the WFPC2 chips.  We estimate that the astrometric uncertainty is
226: $0.2\arcsec$.  We used the HSTphot stellar photometry package to obtain
227: photometry \citep{dolphin00}.  We removed bad pixels, cosmic rays, and
228: hot pixels and then obtained simultaneous photometry for the F606W and
229: F814W images.  
230: 
231: 
232: \section{Results}
233: 
234: \subsection{Radio Counterpart}
235: 
236: The flux density of the radio emission associated with the ULX in
237: the VLA and ATCA images was measured by fitting a 2-D Gaussian
238: point source to the peak radio emission in each of the various
239: images. In all cases, the source appears to be point-like with
240: slightly extended emission, but nothing which we can confidently 
241: resolve.  The radio source position 
242: from our VLA B-array image (resolution of
243: $3.17\arcsec \times 0.92\arcsec$) is RA, DEC (J2000) = 14h03m19.63s 
244: $\pm$ 0.01s, $-$41d22\arcmin~58.7\arcsec~ $\pm 0.2\arcsec$, which
245: is within $0.1\arcsec$ of the ULX position
246: calculated from the {\it Chandra} observations \citep{kaaret03}.  This is
247: well within the {\it Chandra} error circle. Using the ATCA 4.8~GHz and
248: 6.1~GHz data, the position of the radio source is RA, DEC (J2000) = 14h03m19.61s
249: $\pm$ 0.02s, $-$41d22'58.5" $\pm$0.2".
250: 
251: \subsubsection{4.9 GHz Emission}
252: 
253: Table 3 lists the measured 4.9 GHz flux density and corresponding 
254: ROBUST weighting, array configuration, geometrical beam size 
255: ($\sqrt{\theta_{maj}\theta_{min}}$) and an indication of any
256: {\it (u,v)} cutoff used in the imaging.
257: Radio emission was detected above 3$\sigma$ in 20 of the 21 images
258: made at 4.9 GHz from the VLA and ATCA data. 
259: The source was not detected in the highest resolution 4.9 GHz VLA image 
260: (1.01\arcsec~$\times$ 0.28\arcsec; A-array, Robust=$-$1) with an upper 
261: limit of 0.1 mJy. The largest angular size detectable with the VLA in
262: its A-array configuration is 10\arcsec. The radio source is detected in 
263: the other two VLA A-array images, but with modest detections 
264: ($\sim 7\sigma$ in both cases). Figure~\ref{vla_image} shows the BnA-array image of
265: NGC5408 (Robust=1). The resolution of this image is $1.94\arcsec \times
266: 1.20\arcsec$ and the source is clearly detected at the $10\sigma$ level. 
267: 
268: Figure~\ref{atca_unif} shows the highest resolution 4.9 GHz ATCA image,
269: which has a beamsize very similar to the VLA BnA-array image shown in 
270:  Figure~\ref{vla_image}. In the majority of the VLA and ATCA images, the 
271: radio emission associated with NGC 5408 X-1 is obvious and clearly separated from
272: the extended emission associated with the NGC 5408 starburst. However, in cases 
273: such as the VLA B-array naturally-weighted image, where the resolution is lower and
274: favors the extended structures, a significant amount of extended emission is 
275: present (see Fig.~\ref{vla_barray}, left). Removing the shortest 
276: ({\it u,v}) baselines ($<$ 5 k$\lambda$) for these lower-resolution images 
277: produces an image, shown in Fig.~\ref{vla_barray}, right, where the the radio 
278: emission associated with NGC 5408 X-1 is clearly
279: separated from the extended emission associated with the starburst.
280: 
281: Figure~\ref{flux_vs_beam} shows the flux density and associated errors
282: versus the geometric beamsize for measurements from 16 images of
283: VLA (diamonds) and ATCA (square; new and archival) data taken from 
284: Table~\ref{ulx_radioflux}. The filled 
285: symbols indicate that a {\it (u,v)} cutoff was applied to the data in order
286: to remove extended emission which may contaminate the flux density. 
287: However, not all measurements were included in Figure~\ref{flux_vs_beam}. 
288: Images made with the VLA CnB-array at all robust weightings and with
289: the ATCA 6D configuration have low enough resolution that it is difficult
290: to separate the background emission from NGC 5408 from the radio emission
291: at the position of the ULX. In fact, the flux densities for the VLA CnB-array 
292: and ATCA 6D-array measurements are in the range of 0.32 to 0.51 mJy, up to twice as 
293: high as measurements made
294: with more extended arrays, indicating that the background contributes
295: significantly to the measurement. For this reason, only the measurements from
296: images where we were confident we were separating out the radio emission
297: associated with NGC 5408 X-1 were used to make Figure~\ref{flux_vs_beam}. 
298: 
299: \subsubsection{Source Size and Structure}
300: 
301: Previously, \citet{soria06} had made some of the highest resolution
302: observations of the ULX in NGC 5408 using the ATCA. They measured 
303: the flux density in images with beamsizes of $1.5\arcsec$ 
304: to $3.5\arcsec$ and found an essentially flat distribution of 
305: flux density with beamsize. However, their observations do not resolve
306: the source and therefore are only consistent with emission from 
307: a point source (Soria et al. 2006). The VLA data presented here 
308: for the first time probe even higher angular resolutions (0.5\arcsec~to 1.5\arcsec). 
309: \ref{flux_vs_beam} shows that the 
310: VLA data are crucial in studying the flux density versus
311: geometric beam size for scales $<$ 1.5\arcsec. There is a clear 
312: trend of decreasing flux density at smaller
313: beamsizes below a geometric beamsize of 1.5\arcsec.  
314: The linear rise of flux density with beamsize shown in
315: Figure~\ref{flux_vs_beam} indicates that that there is extended
316: emission associated with this source on beamsize scales between $\sim
317: 0.5\arcsec$ and $\sim1.5\arcsec$. For beamsizes greater than $\sim$ 
318: $1.5\arcsec$, the flux density measurements are relatively constant
319: (within the errors), suggesting the source has an angular extent in
320: the range of $1.5\arcsec$ to $2.0\arcsec$. 
321: 
322: \subsubsection{Variability}
323: 
324: Because the radio source associated with the ULX is likely to be somewhat
325: extended, different configurations of the VLA radio telescope will be 
326: sensitive to emission on differing angular size scales. Therefore, it is 
327: not possible to look for flux variability in our high-resolution VLA observations (where the 
328: configuration ranges from A-array to CnB-array). However, ATCA observations 
329: over a number of epochs (e.g., \citet{kaaret03}, \citet{soria06}, and this paper) have
330: shown that the flux density at 4.9 GHz does not appear to vary significantly
331: and has an average, overall value near 0.20 mJy.
332: 
333: \subsubsection{Multifrequency Data and Spectral Index}
334: 
335: Multi-frequency observations at 1.4, 2.3, 4.9 and 6.1 GHz 
336: were made as part of the new ATCA data presented here. At each frequency, 
337: all existing ATCA data were combined (see
338: Table 2) and a point source was fit to the radio emission associated
339: with the ULX. We find flux densities of $0.62 \pm$ 0.10 mJy at 1.4 GHz 
340: (geometric beam size of 6.7\arcsec; ROBUST=$-5$; ({\it u,v}) cutoff 
341: $<$ 5 k$\lambda$), $0.37 \pm 0.08$ mJy at 2.4~GHz (geometric beam size 
342: of 3.3\arcsec; ROBUST=$-5$), $0.20 \pm 0.03$ mJy at 4.9~GHz (geometric 
343: beam size=1.5\arcsec, ROBUST=$-$5) and $0.17 \pm 0.03$ at 6.1~GHz.  
344: At 1.4 GHz, the contribution from the extended background
345: may be present even though we used uniform weighting and
346: limited the shortest {\it (u,v)} data; the beamsize is large. 
347: At 2.3 GHz, the image is made with uniform weighting and the
348: contribution from the extended starburst region is not apparent. 
349:  Therefore, we determine the spectral index based on the 
350: 2.4, 4.9 and 6.1 GHz measurements only and
351: obtain $\alpha$=$-$0.8 $\pm$0.2, where S$_{\nu}$ $\propto$ $\nu^{\alpha}$. 
352: Figure~\ref{spectrum_3pts} shows the flux density versus frequency and includes the
353: 1.4 GHz measurement as well as the 8.5 GHz ATCA upper limit from Kaaret et al. (2003). 
354: Although not included in the fit, the 1.4 and 8.5 GHz 
355: measurements are consistent with the fitted spectral index. 
356: 
357: \subsection{Optical counterpart}
358: 
359: Figure~\ref{ulxclose} shows the HST/WFPC2 image in the F606W filter for
360: the area near the ULX.  The circle drawn on the figure is centered at
361: the VLA source position quoted above and has a radius of $0.28\arcsec$,
362: which represents the relative position uncertainty including both the
363: radio position uncertain and the optical astrometry uncertainty.  Only
364: one object lies within the error circle. It is located at a position of
365: RA, DEC (J2000) = 14h03m19.62s, $-$41d22'58.54 (J2000) which is $0.17\arcsec$
366: from the radio position and well inside the error circle. The next
367: closest HST source is $0.43\arcsec$ from the VLA position and is
368: outside the error circle.  We identify the HST source within the error
369: circle as the likely optical counterpart of the ULX.
370: 
371: The optical counterpart has magnitudes, in the WFPC2 flight photometric
372: system, of $22.387 \pm 0.021$ in the F606W filter and $22.396 \pm
373: 0.043$ in the F814W filter (Dolphin 2000).  The equivalent Johnson-Cousins 
374: magnitudes are $V = 22.4$ and $I = 22.4$.  We corrected for reddening using an extinction
375: $\rm E(B-V) = 0.068$ based on the dust maps of \citet{schlegel98} and
376: using an $\rm R_V = 3.1$ extinction curve.  The dereddened magnitudes
377: are $V_0 = 22.2$ and $I_0 = 22.3$ and the color is $V_0 - I_0 = -0.1
378: \pm 0.1$.
379: 
380: 
381: \section{Discussion}
382: 
383: The radio counterpart size of $1.5\arcsec$ to $2.0\arcsec$ corresponds
384: to a physical diameter of 35--46~pc at the distance of 4.8~Mpc to NGC
385: 5408, as determined by from the tip of the red giant branch
386: \citet{karachentsev02}.  Therefore, we can rule out the interpretation
387: of the radio emission associated with the NGC 5408 ULX as a
388: relativistically beamed jet \citep{kaaret03}.  
389: 
390: The radio surface brightness of NGC 5408 at 4.9 GHz is 
391: $\sim$4 $\times$ 10$^{-20}$ W m$^{-2}$ Hz$^{-1}$ sterrad$^{-1}$, 
392: assuming a source size of 1.5\arcsec. Taking a spectral index
393: of $-$0.8, we can calculate the surface brightness at 1.4 GHz
394: in order to compare to values for known supernova remnants
395: of similar sizes \citep{green04}. The value of 
396: $\sim$1 $\times$ 10$^{-19}$ W m$^{-2}$ Hz$^{-1}$ sterrad$^{-1}$
397: is significantly higher than that of known supernova
398: remnants of similar size. It is therefore more likely that 
399: the radio emission associated with NGC 5408 X-1 arises 
400: instead from an extended radio lobe. The radio spectrum 
401: ($\alpha$=$-$0.8) of this source is consistent with 
402: optically-thin synchrotron emission.
403: 
404: For comparison, the radio nebula W50 surrounds the relativistic jet source 
405: SS 433 \citep{margon84} and is thought to be powered by the relativistic
406: outflow.  SS 433 has been suggested as a possible Galactic analog to
407: the ULXs \citep{fabrika01,begelman06}. The radio
408: nebula in W50 is interesting because the nebular radio emission is unlikely to
409: be beamed.  The physical size of W50 is roughly 50~pc and is similar to
410: what we estimate for the radio nebula surrounding NGC 5408 X-1.  The
411: radio spectral indexes of the various components of W50 range from
412: $-0.5$ to $-0.8$, also consistent with the spectral index of the radio
413: emission in NGC 5408 X-1 ($\alpha$ $\sim$$-$0.8).  A comparison of the radio 
414: brightness can be made for the two sources. W50 has an integrated flux density 
415: at 1.4 GHz of $\sim$ 70 Jy and a distance of $\sim$5 kpc (Dubner
416: et al. 1998). Assuming a spectral index of $-$0.5, that translates
417: to a 4.9 GHz flux density of 40 Jy, and if it were at the distance of
418: NGC 5408 (4.8 Mpc), its flux density would be $\sim$ 40 $\mu$Jy. 
419: The integrated flux density of NGC 5408 X-1 at 4.9 GHz is $\sim$200 $\mu$Jy, 
420: so its radio brightness is more than a factor of 5 greater than that of W50. 
421: 
422: We investigate the energetics of the radio lobe assuming radiation via
423: synchrotron emission, equipartition between particles and fields, and
424: equal energy in electrons and baryons.  We use a spectral index $\alpha
425: = -0.8$ (see $\S$3.1.4), a lower frequency cutoff of 1.3~GHz,
426: and an upper frequency cutoff of 6.2~GHz.  The total radio luminosity
427: of the source is $3.8 \times 10^{34} \rm \, erg \, s^{-1}$. For a
428: source diameter of 46~pc and a filling factor of unity, we find that
429: the total energy required is $3.6 \times 10^{49} \rm \, erg$, the
430: magnetic field is $16 \mu$G, and the synchrotron lifetime is $\sim$20~Myr. 
431: For contrast, for a diameter of 35~pc and a filling factor of 0.1, the total energy
432: required is $9 \times 10^{48} \rm \, erg$, the magnetic field is $39
433: \mu$G, and the lifetime is $\sim$5~Myr.  We note that the estimates of
434: \citet{soria06} for the energy content of a synchrotron nebula
435: surrounding NGC 5408 X-1 are about an order of magnitude larger.  This
436: is primarily because they assume an electron energy
437: distribution that extends down to a Lorentz factor of 1, while we,
438: conservatively, assume that the electron energy distribution extends
439: only over the range needed to produce the observed radio emission (1.3 to 6.2 GHz).
440: 
441: The total energy content in relativistic electrons in W50 is in the range
442: $(0.5-7) \times 10^{46} \rm \, erg$ \citep{dubner98}.  The total energy
443: content in relativistic electrons in the radio nebula surrounding NGC
444: 5408 X-1 is at least two orders of magnitude larger.  Therefore, if two
445: nebula are similar, then the jet powering the nebula surrounding NGC
446: 5408 X-1 must be at least two orders of magnitude more powerful than
447: that from SS 433 powering W50.  The radio lobes powered by the
448: persistent accreting stellar-mass black hole GRS 1758-258 have a radio
449: luminosity of $3 \times 10^{30} \rm \, erg \, s^{-1}$ and require an
450: energy content of $2 \times 10^{45} \rm \, erg$ \citep{rodriguez92} and
451: are even less powerful than W50.  We note that the ULX Holmberg II X-1
452: has an associated radio nebula with a total energy similar to that of
453: the NGC 5408 X-1 nebula \citep{miller05}.
454: 
455: At the distance to NGC 5408, the absolute magnitude of the optical
456: counterpart would be $M_V = -6.2$.  If the light arises only from the
457: stellar companion, then the magnitude and color exclude main sequence
458: and giant stars and require a supergiant star.  The absolute magnitude
459: and color are consistent with classification as a B or early A
460: supergiant.  The luminosity of such a star would be in the range of 1$-$5 $\times
461: 10^{38} \rm \, erg \, s^{-1}$.  However, the stellar classification is
462: suspect since light may arise from reprocessing of X-rays from the
463: compact object and since the X-ray luminosity ($\sim 10^{40} \rm \, erg
464: \, s^{-1}$ if the emission is isotropic) exceeds the expected stellar
465: luminosity by a factor of at least 20 and may strongly affect the
466: physical state of the star.  
467: 
468: The X-ray to optical flux ratio, defined following \citet{vp95} as $\xi
469: = B_{0} + 2.5 \log F_{X}$ where $F_{X}$ is the X-ray flux density at
470: 2~keV in $\mu$Jy and we approximate $B_0 = V_0$ due to lack of a B-band
471: image, is $\xi = 20.0$.  If we use the flux density at 1~keV, this
472: rises to $\xi = 21.1$.  These values are higher than those of any
473: high-mass X-ray binary (HMXB), other than LMC X-3, and are in the range
474: typically found for low-mass X-ray binaries (LMXBs).  Thus, it is possible 
475: that the companion star contributes little to 
476: the observed optical emission, e.g., it is not supergiant, and that most of
477: the optical light arises from reprocessing of X-rays, as occurs in LMXBs 
478: \citep{kaaret05}.
479: 
480: 
481: \section*{Acknowledgments}
482: 
483: PK acknowledges partial support from {\it Chandra} grant CXC GO4-5035A and a
484: Faculty Scholar Award from the University of Iowa.  STSDAS and PyRAF
485: are products of the Space Telescope Science Institute, which is
486: operated by AURA for NASA. The authors also thank Michael Rupen for planning and 
487: carrying out the VLA observations. 
488: 
489: %--------------
490: 
491: \begin{thebibliography}{}
492: 
493: %\bibitem[et~al.\ 19]{484} et al. 19, ApJ,
494: 
495: \bibitem[Arnaud(1996)]{1996ASPC..101...17A} Arnaud, K.~A.\ 1996, 
496: Astronomical Data Analysis Software and Systems V, 101, 17 
497: 
498: \bibitem[Begelman(2002)]{begelman02} Begelman, M.C.\ 2002, ApJ, 568,
499: L97
500: 
501: \bibitem[Begelman(2006)]{begelman06} Begelman, M.C., King, A.R.,
502: Pringle, J. E.\ 2006, MNRAS, 370, 399	
503: 
504: \bibitem[Bautz et al.(1998)]{bautz98} Bautz, M.~W., et al.\ 1998, \procspie, 3444, 210 
505: 
506: \bibitem[Colbert \& Mushotzky(1999)]{colbert99} Colbert, E.J.M.\ \&
507: Mushotzky, R.F.\ 1999, ApJ, 519, 89
508: 
509: \bibitem[Dubner et al.(1998)]{dubner98} Dubner, G.M., Holdaway, M.,
510: Goss, W.M., Mirabel, I.F.\ 1998, AJ, 116, 1842
511: 
512: \bibitem[Dolphin(2000)]{dolphin00} Dolphin, A.E.\ 2000, PASP, 112, 1383
513: 
514: \bibitem[Fabbiano(2006)]{fabbiano06} Fabbiano, G.\ 2006, \araa, 
515: 44, 323 
516: 
517: \bibitem[Fabrika \& Mescheryakov(2001)]{fabrika01} Fabrika, S.\ \&
518: Mescheryakov, A.\ 2001, IAUS, 205, 268
519: 
520: \bibitem[Green(2004)]{green04} Green, D.A.\ 2004, Bull.\ Astron. Soc.\
521: India, 32, 335
522: 
523: \bibitem[Kaaret et al.(2001)]{kaaret01} Kaaret, P.\ et al.\ 2001,
524: MNRAS, 321, L29
525: 
526: \bibitem[Kaaret et al.(2003)]{kaaret03} Kaaret, P., Corbel, S.,
527: Prestwich, A.H., Zezas, A.\ 2003, Science,  299, 365.
528: 
529: \bibitem[Kaaret et al.(2005)]{kaaret05} Kaaret, P.\ 2005, ApJ, 629, 233
530: 
531: \bibitem[Karachentsev et al.(2002)]{karachentsev02} Karachentsev, I.D.\
532: et al.\ 2002, A\&A, 385, 21
533: 
534: \bibitem[King et al.(2001)]{king01} King, A.R.\ et al.\ 2001, ApJ, 552,
535: L109
536: 
537: \bibitem[K\"ording et al.(2002)]{kording02} K\"ording, E. Falcke, H.,
538: \& Markoff, S.\ 2002, A\&A, 382, L13
539: 
540: \bibitem[Makishima et al.(2000)]{makishima00} Makishima, K.\ et al.\
541: 2000, ApJ, 535, 632
542: 
543: \bibitem[Margon(1984)]{margon84} Margon, B.\ 1984, ARA\&A, 22, 507
544: 
545: \bibitem[Miller et al.(2005)]{miller05} Miller, N.A, Mushotzky, R.F.,
546: Neff, S.G.\ 2005, ApJ, 623, L109
547: 
548: \bibitem[Remillard \& McClintock(2006)]{remillard06} 
549: Remillard, R.~A., \& McClintock, J.~E.\ 2006, \araa, 44, 49 
550: 
551: \bibitem[Rodr\'{\i}guez, Mirabel, Mart\'{\i}(1992)]{rodriguez92} Rodr\'{\i}guez,
552: L.F., Mirabel, I.F., Mart\'{\i}, J.\ 1992, ApJ, 401, L15
553: 
554: \bibitem[Sault \& Killeen(1998)]{sault98} Sault R.J. \& Killeen N.E.B.
555: 1998, The Miriad User's Guide, Sydney: Australia Telescope National
556: Facility
557: 
558: \bibitem[Schlegel, Finkbeiner, \& Davis(1998)]{schlegel98} Schlegel,
559: D., Finkbeiner, D., \& Davis, M.\ 1998, ApJ, 500, 525
560: 
561: \bibitem[Soria et al.(2006)]{soria06} Soria, R., Fender, R.P., 
562: Hannikainen, D.C., Read, A.M., \& Stevens, I.R.\ 2006, MNRAS, 368, 1527
563: 
564: \bibitem[Skrutskie et al.(2006)]{2mass} Skrutskie, R.M.\ et al.\ 2006,
565: AJ, 131, 1163.
566: 
567: \bibitem[van Paradijs \& McClintock(1995)]{vp95} van Paradijs, J.\ \&
568: McClintock J.E.\ in X-Ray Binaries, eds.\ W.H.G.\ Lewin,  J.\ van
569: Paradijs, \& E.P.J. van den Heuvel, (Cambridge Univ.\ Press, 1995),
570: pp.\ 58-125.
571: 
572: \bibitem[van Speybroeck et al.(1997)]{vanspey97} van Speybroeck, 
573: L.~P., Jerius, D., Edgar, R.~J., Gaetz, T.~J., Zhao, P., \& Reid, P.~B.\ 
574: 1997, \procspie, 3113, 89 
575: 
576: \bibitem[Weisskopf(1988)]{weisskopf88} Weisskopf, M.~C.\ 1988, 
577: Space Science Reviews, 47, 47 
578: 
579: \end{thebibliography}
580: 
581: %\label{lastpage}
582: \clearpage
583: 
584: \begin{deluxetable}{ccl} \tabletypesize{\scriptsize}
585: \tablecaption{Chandra and VLA Observations of NGC 5408 \label{obstable}}
586: \tablewidth{0pt} 
587: \tablehead{\colhead{Date/Time} & \colhead{X-Ray Flux} & \colhead{VLA array}}
588: \startdata 
589: 2002-05-07 13:45:47 & 1.42 & $-$\\ 
590: 2003-12-20 13:06:33 & 1.49 & B \\
591: 2004-02-09 10:29:47 & 1.58 & CnB \\ 
592: 2004-12-20 13:14:36 & 1.47 & A \\ 
593: 2005-01-29 11:02:49 & 1.50 & BnA \\ 
594: \enddata 
595: \vspace{-12pt}\tablecomments{The table lists the Chandra/VLA
596: observations of NGC 5408 and includes the date and UT time of the start
597: of the Chandra observation (the VLA observations were roughly
598: simultaneous),  the ULX X-ray flux in the 0.3-8~keV band in units of
599: $10^{-12} \rm \, erg \, cm^{-2} \, s^{-1}$ during the observation, and
600: the VLA array configuration.} 
601: \end{deluxetable}
602: 
603: \clearpage
604: \begin{deluxetable}{lccl} \tabletypesize{\scriptsize}
605: \tablecaption{ATCA Observations of NGC 5408 \label{atca_obs}}
606: \tablewidth{0pt} 
607: \tablehead{\colhead{Date/Time} & \colhead{Array} & \colhead{Frequency (GHz)} & \colhead{Reference}}
608: \startdata 
609: 2000-04-01 02:09&6D & 4.8, 8.6 & Kaaret et. al (2003)\\
610: 2003-03-03 05:31&6A& 4.8, 8.6 & Archival 1 \\
611: 2003-05-20 22:05&1.5C& 1.4 & Archival 2 \\	 	 
612: 2003-07-27 17:45&6D& 1.4 & Archival 3\\	
613: 2003-12-07 10:05&6A& 2.3, 4.8 & this paper\\
614: 2003-12-08 11:30&6A& 2.3, 4.8 & this paper\\
615: 2003-12-09 11:02&6A& 2.3, 6.1 & this paper\\
616: 2004-12-09 to 12-11&6D& 1.4, 2.3 & Soria et al. (2006)\\
617: 2004-12-09 to 12-11&6D& 4.8, 6.1 & Soria et al. (2006)\\
618: \enddata 
619: \vspace{-12pt}\tablecomments{The table lists the ATCA observations of
620: NGC 5408 and includes the observation date, the frequency
621: or frequencies of observation, and the measured flux density.}
622: \end{deluxetable}
623: 
624: \clearpage
625: \begin{deluxetable}{lrccc} \tabletypesize{\scriptsize}
626: \tablecaption{4.9 GHz Flux Density of NGC 5408 \label{ulx_radioflux}}
627: \tablewidth{0pt} 
628: \tablehead{\colhead{Telescope} & \colhead{Robust} & \colhead{Flux Density} & \colhead{Geometric} &\colhead{(u,v)}\\
629: \colhead{Array} & \colhead{Weight}& \colhead{(mJy)} & \colhead{Beamsize} &\colhead{cutoff}}
630: \startdata 
631: VLA-A&$-$1&$<$0.10$\pm$0.03&0.53&\\
632: VLA-A&1&0.13$\pm$0.02&0.75&\\
633: VLA-A&3&0.14$\pm$0.02&0.83&\\
634: VLA-BnA&$-$1&0.15$\pm$0.03&1.17&\\
635: VLA-BnA&1&0.20$\pm$0.02&1.53&\\
636: VLA-BnA&3&0.23$\pm$0.02&1.70&\\
637: VLA-B&$-$1&0.21$\pm$0.03&1.71&\\
638: VLA-B&0&0.23$\pm$0.03&1.93&\\
639: VLA-B&$-$5&0.20$\pm$0.03&1.62\\
640: VLA-B&5&0.23$\pm$0.02&2.66&$<$ 5 k$\lambda$\\
641: ATCA-6A&$-$5&0.21$\pm$0.03&1.57&\\
642: ATCA-6A&0&0.20$\pm$0.03&1.89\\
643: ATCA-6A&2&0.19$\pm$0.02&3.25\\
644: ATCA-6A&5&0.19$\pm$0.02&3.24\\
645: ATCA-6D&$-$5&0.25$\pm$0.06&1.59&$<$ 3 k$\lambda$\\
646: ATCA-6D&5&0.18$\pm$0.04&3.18&$<$ 3 k$\lambda$\\
647: \hline
648: VLA-CnB&0&0.51$\pm$0.03&4.45&\\
649: VLA-CnB&$-$5&0.46$\pm$0.04&3.77\\
650: VLA-CnB&$-$5&0.32$\pm$0.04&3.71&$<$ 5 k$\lambda$\\
651: VLA-CnB&$-$1&0.47$\pm$0.03&4.08\\
652: ATCA-6D&5&0.43$\pm$0.04&3.46\\
653: 
654: \enddata 
655: \vspace{-12pt}\tablecomments{The table lists the 4.9 GHz flux
656: densities for the radio counterpart of the ULX in NGC5408. 
657: All fits are for an unresolved point source. It is not possible
658: to distinguish between the source and background emission for
659: the measurements listed below the horizontal line.}
660: \end{deluxetable}
661: 
662: \clearpage
663: \begin{figure}
664: \centerline{\includegraphics[angle=270,width=6in]{f1.ps}}
665: \caption{VLA 4.9 GHz BnA-array image of the radio emission in NGC5408.
666: The compact radio source associated with the ULX is located in the
667: lower left (SE) of the plot. The cross represents the position of the
668: ULX. The spatial resolution of this image
669: is 1.94\arcsec~$\times$ 1.20\arcsec, PA=8\arcdeg, and the image has been made with
670: ROBUST=1 weighting (slightly naturally weighted). The contour levels represent 
671: radio emission at -3, 3, 5, 7, 9,  10, 15, 20, 25, 50, 75, 100, 150, and 200 times the rms
672: level of 20 $\mu$Jy beam$^{-1}$.} 
673: \label{vla_image} 
674: \end{figure}
675: 
676: \clearpage
677: \begin{figure}
678: \centerline{\includegraphics[angle=270,width=6in]{f2.ps}}
679: \caption{ATCA 4.8 GHz image of the radio emission in NGC 5408. The 
680: compact radio source associated with the ULX is located in the
681: lower left (SE) of the plot. The crosses represents the position of the 
682: ULX. The spatial resolution of this image
683: is 1.90\arcsec~$\times$ 1.26\arcsec, PA=$-$55\arcdeg, and contours
684: represent radio emission at $-$3, 3, 4, 5, 7, 9, 11, 13, 18, 21, 25, 30, 52, 84, 
685: and 136 the rms level of 32 $\mu$Jy beam$^{-1}$. The image was made
686: with uniform weighting (ROBUST=-5). 
687: \label{atca_unif}} \end{figure}
688: 
689: \clearpage
690: \begin{figure}
691: \centerline{\includegraphics[angle=0,width=6in]{f3.eps}}
692: \caption{VLA 4.9 GHz B-array images of the radio emission in NGC 5408
693: made with ROBUST=5 (natural weighting). The left panel shows an image with 
694: a spatial resolution and the right panel shows an image made with data
695: which have had a ({\it u,v}) cutoff $<$ 5 k$\lambda$ applied to them.  
696: In both cases, the spatial resolution is 5.0\arcsec~$\times$1.5\arcsec, 
697: PA=$-$3\arcdeg, and contour levels reprsent $-$3, 3, 5, 7, 10, 15, 20, 25, 
698: 50, 75, 100, 125 and 150 times the rms level of 20 $\mu$Jy beam$^{-1}$. 
699: \label{vla_barray}} \end{figure}
700: 
701: \clearpage
702: \begin{figure}
703: \centerline{\includegraphics[angle=90,width=6in]{f4.ps}}
704: \caption{Flux Density vs. Geometric Beamsize for ATCA (squares) and VLA 
705: (diamond) radio observations at 4.9 GHz from this paper and from
706: archival observations of Soria et al. (2006). Filled symbols represent 
707: measurements from images where the shortest ({\it u,v}) baselines
708: have been removed to reduce contamination from the background galaxy. 
709: Error bars represent the error in fitting 2-D Gaussians to the
710: unresolved point source. 
711: \label{flux_vs_beam}} 
712: \end{figure}
713: 
714: \clearpage
715: \begin{figure}
716: \centerline{\includegraphics[angle=0]{f5.eps}}
717: \caption{Flux density vs. frequency for the radio emission associated 
718: with NGC 5408 X-1. These measurements are from ATCA data, and the 
719: spectral index of $\alpha$=$-$0.82$\pm$24 is based on the 2.3, 4.9 and 
720: 6.1 GHz points. The 1.4 GHz flux density and 8.5 GHz upper limit 
721: are also included on this plot. 
722: \label{spectrum_3pts}} 
723: \end{figure}
724: 
725: \clearpage
726: \begin{figure}
727: \centerline{\includegraphics[angle=0,width=4in]{f6.ps}}
728: \caption{Image of the field near NGC5408 X-1 in the F606W band.  The
729: circle is centered at the position of the VLA counterpart of the ULX
730: and has a radius of $0.28\arcsec$, which represents the relative
731: position uncertainty including both the radio position uncertain and
732: the optical astrometry uncertainty. Only one source lies within the
733: error circle and we identify it as the likely optical counterpart of
734: the ULX.  The arrow has a length of $1\arcsec$ and points north.
735: \label{ulxclose}} \end{figure}
736: 
737: 
738: 
739: \end{document}
740: 
741: