1: %\documentstyle[epsfig,mncite,times]{mn}
2: %\documentstyle[epsfig,mncite]{mn}
3: \documentclass[useAMS]{mn2e}
4: \usepackage{rotating}
5: \usepackage{mncite}
6:
7: %\renewcommand{\baselinestretch}{2}
8: \newcommand\msun {M$_{\odot}$}
9: \def\approxgt{\ifmmode \rlap{$>$}{}_{{}_{{}_{\textstyle\sim}}} \else%
10: $\rlap{$>$}{}_{{}_{{}_{\textstyle\sim}}}$\fi}
11: \def\approxlt{\ifmmode \rlap{$<$}{}_{{}_{{}_{\textstyle\sim}}} \else%
12: $\rlap{$<$}{}_{{}_{{}_{\textstyle\sim}}}$\fi}
13:
14: \LARGE \normalsize \title[]{Radio and X--ray observations during the
15: outburst decay of the Black Hole Candidate XTE~J1908+094 }
16:
17: \author[P.G. Jonker et al.] {P.G. Jonker$^{1,2,3}$\thanks{email :
18: pjonker@cfa.harvard.edu}, E. Gallo$^4$, V. Dhawan$^5$, M. Rupen$^5$, R.P. Fender$^4$, G. Dubus$^7$ \newauthor \\
19: $^1$Institute of Astronomy, Madingley Road, CB3 0HA, Cambridge, UK\\
20: $^2$Present address, Harvard--Smithsonian Center for Astrophysics, 60 Garden Street, MS83, MA 02138, Cambridge, U.S.A.\\
21: $^3${\it Chandra} Fellow\\
22: $^4$Astronomical Institute ``Anton Pannekoek'',
23: University of Amsterdam, Kruislaan 403, 1098 SJ Amsterdam, The Netherlands\\
24: $^5$National Radio Astronomy Observatory, Socorro, NM 87801, USA\\
25: $^6$Laboratoire Leprince--Ringuet, Ecole Polytechnique 91128
26: Palaiseau, France}
27:
28: \begin{document}
29:
30: \maketitle
31:
32: \begin{abstract}
33: \noindent
34: Obtaining simultaneous radio and X--ray data during the outburst decay
35: of soft X--ray transients is a potentially important tool to study the
36: disc -- jet connection. Here we report results of the analysis of
37: (nearly) simultaneous radio (VLA or WSRT) and {\it Chandra} X--ray
38: observations of XTE~J1908+094 during the last part of the decay of the
39: source after an outburst. The limit on the index of a radio -- X--ray
40: correlation we find is consistent with the value of $\sim0.7$ which
41: was found for other black hole candidates in the low/hard
42: state. Interestingly, the limit we find seems more consistent with a
43: value of 1.4 which was recently shown to be typical for radiatively
44: efficient accretion flow models. We further show that when the
45: correlation--index is the same for two sources one can use the
46: differences in normalisation in the radio -- X--ray flux correlation
47: to estimate the distance towards the sources if the distance of one of
48: them is accurately known (assuming black hole spin and mass and jet
49: Lorentz factor differences are unimportant or minimal). Finally, we
50: observed a strong increase in the rate of decay of the X--ray flux.
51: Between March 23, 2003 and April 19, 2003 the X--ray flux decayed with
52: a factor $\sim$5 whereas between April 19, 2003 and May 13, 2003, the
53: X--ray flux decreased by a factor $\sim$750. The source (0.5--10 keV)
54: luminosity at the last {\it Chandra} observation was ${\rm L\approx
55: 3\times10^{32} (\frac{d}{8.5 kpc})^2 erg\,s^{-1}}$.
56:
57:
58: \end{abstract}
59:
60: \begin{keywords} stars: individual (XTE~J1908+094) --- stars: black holes
61: --- X-rays: stars
62: \end{keywords}
63:
64: \section{Introduction}
65: \label{intro}
66:
67: Low--mass X--ray binaries (LMXBs) are binary systems in which a
68: $\approxlt 1\,M_{\odot}$ star transfers matter to a neutron star or a
69: black hole. These systems form one of our main windows on the physical
70: processes taking place around black holes and hence they can provide
71: us with information about the fundamental properties of spacetime. One
72: reason for this is that the great majority of Galactic black hole
73: candidates (BHCs) are found in transient LMXB systems.
74:
75: Over the last few years it has become apparent that jets are an
76: integral and energetically important part of these BHC systems
77: (especially) when these systems are in the so called low/hard state
78: (\pcite{2001MNRAS.322...31F}; \pcite{2001MNRAS.327.1273S}). Recently,
79: it was found that there exists a correlation between the radio and
80: X--ray flux in the low/hard state of several BHCs over 3--4 decades in
81: X--ray flux showing that there must be some form of disc--jet coupling
82: (\pcite{2003A&A...400.1007C}; \pcite{2003MNRAS.344...60G}).
83: \scite{2003MNRAS.343L..59H}, \scite{falcke2003}, and
84: \scite{merheinzdim2003} review the disc--jet connection in terms of
85: different accretion disc and jet models. \scite{merheinzdim2003},
86: building on previous work of \scite{2003MNRAS.343L..59H}, showed that
87: inefficient accretion flow models can reproduce the observed radio --
88: X--ray correlation index for the initial parameter space they covered.
89: \scite{2003A&A...397..645M} showed that the jet--model explaining the
90: observed X--rays in terms of synchrotron emission from the jet of
91: \scite{2001A&A...372L..25M} can reproduce both the observed
92: correlation index as well as the normalisation.
93: \scite{2003MNRAS.343L..99F} used the observed radio -- X--ray
94: correlation for BHCs to argue that there is no need to advect energy
95: across a black hole event horizon in order to explain the observed
96: difference in quiescent luminosity between the neutron star and BHC
97: transient systems as was proposed by e.g.~\scite{2001ApJ...553L..47G}.
98: An important assumption in the work of \scite{2003MNRAS.343L..99F} is
99: that the observed radio -- X--ray correlation holds down to X--ray
100: luminosities as low as L$_X \sim10^{30-32}$ erg s$^{-1}$.
101:
102: XTE~J1908+094 was discovered serendipitously during {\it RXTE}
103: observations of the soft gamma--ray repeater SGR~1900+14 by
104: \scite{2002IAUC.7856....1W}. The source flux is absorbed (${\rm
105: N_H\sim2.3\times10^{22} cm^{-2}}$), the spectrum is well--fit with a
106: hard power--law with a photon index of 1.55. Subsequent {\it
107: BeppoSAX} observations (\pcite{2002IAUC.7873....1I};
108: \pcite{2002A&A...394..553I}) confirmed both the hard spectrum (the
109: source was detected up to 250 keV) and the high Galactic absorption. A
110: broadened iron emission line was present in spectra extracted from
111: both the RXTE and the BeppoSAX observations. In 't Zand et al.
112: (2002) presented strong evidence for a low/hard -- high/soft state
113: change. The fact that the source displayed both a low/hard and a
114: high/soft state during the outburst is confirmed by the timing and
115: spectral analysis of the RXTE/PCA observations by
116: \scite{2002xrb..confE..11G} (see also \pcite{gogus2004}). A radio
117: counterpart was discovered by \scite{2002IAUC.7874....1R} whereas a
118: near--infrared counterpart was found by \scite{2002MNRAS.337L..23C}.
119: These authors also found that the optical upper limits
120: (\pcite{2002ATel...86....1W} and \pcite{2002IAUC.7877....4G}) are
121: fully consistent with the near--infrared colours of and the high
122: extinction towards the source.
123:
124: In this paper we report the findings of our (nearly) simultaneous Very
125: Large Array (VLA)\footnote{The National Radio Astronomy Observatory is
126: a facility of the National Science Foundation operated under
127: cooperative agreement by Associated Universities, Inc.} and
128: Westerbork Synthesis Radio Telescopes (WSRT)\footnote{The Westerbork
129: Synthesis Radio Telescope is operated by the ASTRON (Netherlands
130: Foundation for Research in Astronomy) with support from the
131: Netherlands Foundation for Scientific Research NWO} radio and {\it
132: Chandra} X--ray observations of XTE~J1908+094 during the last part
133: of the decay of the source after an outburst.
134:
135: \section{Observations and analysis}
136: We have observed the BHC soft X--ray transient (SXT) XTE~J1908+094
137: using the ACIS detector in its Timed Exposure mode on--board the {\it
138: Chandra} satellite (\pcite{1988SSRv...47...47W}) on three occasions.
139: Within a few days of the X--ray observations radio observations using
140: either the VLA or the WSRT were performed. A log of the observations
141: can be found in Table~\ref{log}. All observing times have been
142: converted to UTC.
143:
144: \subsection{The {\it Chandra} X--ray data}
145:
146: The X--ray data were processed by the {\it Chandra} X--ray Center;
147: events with ASCA grades of 1, 5, 7, cosmic rays, hot pixels, and
148: events close to CCD node boundaries were rejected. We used the
149: standard {\it CIAO} software to reduce the data (version 2.3 and CALDB
150: version 2.21). A streak caused by the arrival of photons during the
151: CCD readout period (which lasts $\sim$41~ms in total) was present during
152: the first observation.
153: %We removed this streak before processing the data further
154: %using the {\it CIAO} tool {\sc acisreadcorr} while taking into account
155: %that we had windowed the CCD to 1/8 to reduce effects of pile--up.
156:
157:
158: In all the observations we detect only one source. After applying the
159: {\it CIAO} web--based tool {\sc fix offsets} to correct for known
160: aspect offsets to each of the observations separately, we derive the
161: following coordinates for the source: R.A.=19h08m53.07s,
162: Decl.=+09$^\circ$23'05.0" (typical error 0.6", equinox 2000.0). The
163: radio coordinates of this transient (R.A.=19h08m53.077s
164: Decl.=+09$^\circ$23'04.9", with an error of 0.1", equinox 2000.0;
165: \pcite{2002IAUC.7874....1R}) are fully consistent with this.
166:
167: The source is detected at a count rate of 1.84$\pm$0.02 counts
168: s$^{-1}$ and 0.54$\pm$0.01 counts s$^{-1}$ for the March 23 and the
169: April 19 observation, respectively. These count rates have not been
170: corrected for effects of pile--up (see below). During the third
171: observation we barely detected the source; we detected 9.1$\pm$3.2
172: source counts (0.3--8 keV) spread over the entire length of the
173: observation (10.7 ksec, i.e.~a count rate of $\sim8.5\times10^{-4}$
174: counts s$^{-1}$). In Fig.~\ref{xdecay} we plot the three unabsorbed
175: X--ray fluxes (0.5--10 keV) as determined from spectral model fits to
176: the {\it Chandra} data (see below). The X--ray flux on April 12 was
177: estimated from an interpolation of the flux decay between March 23 and
178: April 19. We arbitrarily added a flux error of 50 per cent to this
179: point. Between March 23 and April 19 the flux decayed by a factor of
180: $\sim$5, whereas between April 19 and May 13 the flux decayed by a
181: factor of $\sim$750. During the last observation the unabsorbed
182: 0.5--10 keV source flux was $(3.6^{+0.8}_{-0.2})\times10^{-14}$ erg
183: cm$^{-2}$ s$^{-1}$. In this estimate we used a power--law index of
184: 2.0$\pm0.5$, which is often found for quiescent BHCs
185: (\pcite{2002ApJ...570..277K}). This flux translates to a source
186: luminosity in the 0.5--10 keV band of ${\rm L\approx 3\times10^{32}
187: (\frac{d}{8.5 kpc})^2 erg\,s^{-1}}$.
188:
189: \begin{figure*}
190: \includegraphics[width=12cm,height=9cm]{xdecay.ps}
191: \caption{The decay of the unabsorbed X--ray flux (0.5--10 keV) as a
192: function of time. Time zero is March 23, 23:51 UTC, 2003 (MJD
193: 52721.9937). The diamond is at the estimated X--ray flux of the
194: source on April 12 assuming an exponential decay between March 23
195: and April 19 (see text).}
196: \label{xdecay}
197: \end{figure*}
198:
199:
200: The source spectra of the first two observations are extracted with 20
201: counts per bin. We only include energies above 0.3 and below 8 keV in
202: our spectral analysis since the ACIS Timed Exposure mode spectral
203: response is not well calibrated below 0.3 keV and above 8 keV. We fit
204: the spectra using the {\sc ISIS} package version 1.1.3
205: (\pcite{2000adass...9..591H}). Since the source is heavily absorbed we
206: hardly detected photons with energies below 1 keV. For this reason we
207: did not have to worry about the additional absorption due to
208: contamination by the optical blocking filters in our spectral fits
209: \footnote{see http://asc.harvard.edu/cal/Acis/Cal\_prods/qeDeg/}
210: (which primarily affects photons below 1 keV). However, we still used
211: the tool {\sc corrarf} to alter the auxiliary response file such that
212: this excess absorption is taken into account. Due to the high count
213: rate during the first observation the pile--up fraction is 25--30 per
214: cent for our frame time of 0.44~s (i.e.~a frame time of 0.4~s
215: since we had windowed the CCD to 1/8th of the nominal size plus a
216: deadtime of 41~ms for reading out the CCD). The pile--up fraction
217: during the second observation is approximately 10 per cent for the
218: same frame time. In both cases we used the pile--up model of
219: \scite{2001ApJ...562..575D} to fit the spectra to model the effect of
220: pile--up on the spectrum. Furthermore, in order to test the results
221: from the pile--up model we extracted photons from the region covered
222: by the read--out streak which is unaffected by pile--up for the first
223: observation; two boxes, one $\sim$10x45 arcseconds$^2$ and the other
224: $\sim$10x15 arcseconds$^2$, were placed $\sim$5 arcseconds East and
225: West of the best--fit source position, respectively. The background
226: was determined from a region $\sim$10x60 arcseconds$^2$ placed
227: $\sim$16 arcseconds North of the best--fit source position. We fitted
228: the spectrum obtained from the read--out streak using {\sc XSPEC}
229: (\pcite{ar1996}) version 11.3.0.
230:
231: The effective exposure time for the readout streak data is much lower
232: than the actual exposure time since each pixel along the read--out
233: direction is illuminated by the source only for 40 $\mu$seconds (the
234: time necessary to transfer and read--out one row). Since the
235: streak pixels are illuminated at the approximately the same position
236: (with respect to on--axis) as the pixels during the frame integration
237: time we do not have to worry about differences in the effective area
238: and hence Auxiliary Response File for the streak and full frame
239: pixels. The effective streak exposure time for the streak boxes,
240: spanning $\sim130$ streak pixels as described above, is $\sim5.3\times
241: 10^{-3}$~s per frame. For a total exposure time of 5.2~ksec a total of
242: approximately 5.2$\times 10^{3}$/0.44 frames are obtained. Hence, the
243: effective streak exposure time is $\sim$63~s.
244:
245: Due to the pile--up the power--law index, normalisation, N$_H$, and
246: pile--up parameter $\alpha$ are degenerate. Indeed, a fit to the data
247: using an absorbed power law including the pile--up model gave an
248: unrealistically low value for N$_H$ of 1.6$\pm$0.5$\times10^{21}$
249: cm$^{-2}$ for the March 23 observation whereas N$_H$ was
250: 4.0$\pm$0.3$\times10^{22}$ cm$^{-2}$ for the April 19 observation. The
251: N$_H$ is most likely close to 2.5$\times10^{22}$ cm$^{-2}$ found by
252: \scite{2002A&A...394..553I} using spectral modelling of BeppoSAX
253: outburst data of XTE~J1908+094. This is close to 1.8$\times10^{22}$
254: cm$^{-2}$ estimated using the model of \scite{1990ARA&A..28..215D}.
255: Therefore, during the fit we kept the value of the interstellar
256: absorption fixed at a value of 2.5$\times10^{22}$ cm$^{-2}$. An
257: absorbed blackbody model also represents the data well statistically
258: although the residuals show systematic trends and the pile--up
259: fraction needed to fit the second observation was unrealistically high
260: ($\alpha\sim1$ whereas a value of 0.5 as found using the power--law
261: model is closer to the value expected based upon the pile--up
262: estimates). For that reason and since spectra of BHCs in the low/hard
263: state are usually well--fit with a power--law spectrum we only
264: included the parameters of the best--fit power--law model in
265: Table~\ref{specpars}. The fit result for the power--law index derived
266: using the read--out streak and pile--up model are consistent. To
267: estimate the uncertainty in the source flux we used the extrema of the
268: power--law indexes and normalisations encompassing the 90 per cent
269: confidence regions. The best--fit power--law spectrum of the first
270: observation extracted from the read--out streak is plotted in
271: Figure~\ref{xspec}.
272:
273: \begin{figure*}
274: \includegraphics[width=9cm,height=12cm,angle=-90]{streakxspecfit.ps}
275: \caption{{\it Top panel:} The 0.3--8 keV X--ray spectrum of
276: XTE~J1908+094 obtained with the {\it Chandra} observatory extracted
277: from the read--out streak (see text) on March 23, 2003. The solid line
278: is the best--fit absorbed power--law model. {\it Bottom panel:} The
279: residuals (data minus model) of the best--fit power--law model. }
280: \label{xspec}
281: \end{figure*}
282:
283: \begin{table*}
284: \caption{Log of the radio and X--ray observations.}
285: \label{log}
286: \begin{center}
287:
288: \begin{tabular}{lcccc}
289: \hline
290: Observation & Calender Date & Date MJD & Exposure time$^a$ & Band \\
291: type & UTC & UTC & ksec. & \\
292: \hline
293: \hline
294: X--ray & March 23, 2003 at 23:51:00 & 52721.9937 & 5.2 & \\
295: Radio & March 25, 2003 at 13:00 & 52723.5416& 14.4 & 8.3 GHz\\
296: Radio & April 12, 2003 at 12:50 & 52741.5347& 14.4& 8.3 GHz\\
297: X--ray & April 19, 2003 at 14:57:16 & 52748.6231& 5.15 & \\
298: Radio & April 19, 2003 at 03:46 &52748.1569 & 43.2 & 5 GHz \\
299: Radio & May 13, 2003 at 13:17 &52772.5534 & 9 & 8.3 GHz\\
300: Radio & May 14, 2003 at 05:44 &52773.2388 & 25.2 & 8.3 GHz\\
301: X--ray & May 13, 2003 at 23:31:32 & 52772.9802& 10.7 & \\
302: \end{tabular}
303: \end{center}
304:
305: {\footnotesize $^a$ For the radio observations the effective exposure
306: times are 60--70 per cent of the quoted times since calibrations are
307: performed intermittent the science exposures. For the X--ray
308: observations the effective exposure time is $\sim$90 per cent of the
309: quoted times since after a frame time of 0.4~s a deadtime for readout
310: of the frame of 41~ms is present.}\newline
311: \end{table*}
312:
313: \begin{table*}
314: \caption{Best fit parameters of the spectra of XTE~J1908+094. All
315: quoted errors are at the 90 per cent confidence level. The local
316: absorption due to the {\it Chandra} optical blocking filters was
317: accounted for. The first line gives the best--fit results using the
318: spectrum extracted from the read--out streak. The X-ray flux is
319: corrected for interstellar absorption. }
320: \label{specpars}
321: \begin{center}
322:
323: \begin{tabular}{lccccccc}
324: \hline
325: Observation & N$_H$ & PL$^a$ & PL normalisation& Pile--up parameter& Flux (0.5--10 keV) & Reduced \\MJD (UTC)
326: & ($\times10^{22}$ cm$^{-2}$) & Index & photons keV$^{-1}$ cm$^{-2}$ s$^{-1,b}$ & $\alpha$& ergs cm$^{-2}$ s$^{-1}$ & $\chi^2/$d.o.f. \\
327: \hline
328: \hline
329: 52721.9937 & 2.5$^c$ & 1.9$\pm$0.3 & (63$\pm$20)$\times$10$^{-3,d}$ & ... & 2.6$\times
330: 10^{-10,d}$ & 1.17/36 \\
331: 52721.9937 & 2.5$^c$ & 2.1$\pm$0.1 & (33$\pm$5)$\times$10$^{-3}$ &
332: 0.8$\pm$0.2 & 1.3$^{+1.2}_{-0.3} \times 10^{-10}$ & 2.46/320 \\
333: 52748.6231 & 2.5$^c$ & 1.3$\pm$0.2 & (2.5$\pm$0.3)$\times$10$^{-3}$ &
334: 0.5$\pm$0.2 & $(2.7\pm1.0)\times 10^{-11}$ & 0.95/106 \\
335:
336:
337: \end{tabular}
338: \end{center}
339:
340: {\footnotesize $^a$ PL = power law}\newline {\footnotesize $^b$ Power
341: law normalisation at 1 keV.}\newline {\footnotesize $^c$ Parameter
342: fixed at this value during the fit.}\newline {\footnotesize $^d$ The
343: flux in the read--out streak and the PL normalisation are corrected
344: for the effective readout--streak exposure time (see text).}\newline
345: \end{table*}
346:
347:
348: \subsection{The VLA $+$ WSRT radio data}
349:
350: We observed XTE~J1908+094 on four occasions at radio wavelengths.
351: Three VLA observations with the VLA in the D configuration at 8.3 GHz
352: and one WSRT observation at 5 GHz were obtained (see Table~\ref{log}).
353: The VLA radio data were reduced using the Astronomical Image
354: Processing System (AIPS), whereas the WSRT data was reduced using the
355: Multichannel Image Reconstruction, Image Analysis and Display package
356: (MIRIAD; \pcite{1995adass...4..433S}). A source was detected on two of
357: the four occasions at the position of the radio counterpart. The radio
358: flux was 0.355$\pm$0.018 mJy on March 25, 0.10$\pm$0.02 mJy on April
359: 12, and 3 $\sigma$ upper limits of 0.1 mJy/beam on April 19 (WSRT
360: data, 5 GHz), and 57 $\mu$Jy/beam on May 14, 2003 on the source flux
361: at the position of the radio counterpart were obtained. As mentioned
362: above the WSRT observations were made at 5 GHz whereas the VLA
363: observations were made at 8.3 GHz. However, the radio spectrum of BHC
364: SXTs in the low--hard state is usually flat $\alpha\approxgt0$
365: (\pcite{2001MNRAS.322...31F}; S$_\nu\propto\nu^{\alpha}$). Hence,
366: assuming a flat radio spectrum the flux at 8.3 GHz will be the same as
367: that at 5 GHz.
368:
369: \subsection{The X--ray -- radio correlation}
370:
371: We have one radio observation close in time (within two days) to the
372: X--ray observation on March 23, two radio observations which are
373: simultaneous with an X--ray observation (on April 19, and May 13,
374: 2003), and one radio observation (on April 12, 2003) which was
375: obtained a week before an X--ray observation (see Table~\ref{log}). We
376: estimated the X--ray flux on April 12 by assuming an exponential decay
377: in X--ray flux with time between March 23 and April 19. However, if we
378: extrapolate the steep part of the decay (that between April 19 and May
379: 13) backwards in time we would derive a much higher X--ray flux for
380: April 12. Furthermore, for several SXTs only the upper--envelope of
381: the decay follows the exponential decay profile, the flux on a
382: specific date can be lower than that predicted by an exponential decay
383: (\pcite{1997ApJ...491..312C}); hence the flux on April 12 could also
384: have been lower than our estimate. With these caviats in mind we
385: plotted the radio -- X--ray points in Figure~\ref{xrrel}. If we use
386: these points to put a limit on a radio -- X--ray correlation it seems
387: that the correlation is steeper than the previously observed
388: 0.7$\pm$0.1 power--law index slope (denoted by the thick dashed line;
389: Gallo et al.~2003; \pcite{2003MNRAS.345L..19M}). If we assume that the
390: radio flux on April 19, 2003 was at the WSRT 1~$\sigma$ upper limit
391: and also take the March 23, 2003 radio/X--ray point, the power--law
392: index would be 1.5$^{+0.45}_{-0.3}$ (1~$\sigma$ errors, taking into
393: account the asymmetric errors in the X--ray flux).
394:
395: Gallo, Fender, Pooley (2002, 2003) discuss reasons for the observed
396: spread in the normalisation constant of the radio -- X--ray
397: correlation. They show that the observed spread in the relation can be
398: described if the various sources have low Lorentz
399: factors. \scite{merheinzdim2003} showed that the normalisation depends
400: also on black hole mass and spin but for comparisons between X--ray
401: binaries this is likely to be less important. The main assumption in
402: the work cited above is that the physics responsible for the disc--jet
403: coupling is the same for each source. Below, we show that under this
404: assumption the observed spread in the correlation normalisation can
405: also be ascribed to the different distances towards the sources.
406: \begin{equation}
407: {\rm F_{Radio} \propto L_{Radio} \times d^{-2}}
408: \end{equation}
409: \begin{equation}
410: {\rm F_{Radio} \propto L_X^{0.7} \times d^{-2}}
411: \end{equation}
412: where we have used the ${\rm L_{Radio}\propto L_X^{0.7}}$ found by
413: \scite{2003A&A...400.1007C} and \scite{2003MNRAS.344...60G}. This
414: leads to
415: \begin{equation}
416: {\rm F_{Radio} \propto F_X^{0.7} \times d^{-0.6}}
417: \end{equation}
418: for the relation between the fluxes. Alternatively, if the radio--
419: X--ray correlation index is 1.4 instead of 0.7 then
420: %\begin{equation}
421: %{\rm F_{Radio} \propto L_X^{1.4} \times d^{-2}}
422: %\end{equation}
423: \begin{equation}
424: {\rm F_{Radio} \propto F_X^{1.4} \times d^{+0.8}}.
425: \end{equation}
426:
427: From these considerations it follows that a nearby source will be more
428: radio loud than a source at the same X--ray flux that is further away
429: if the relation between the radio and X--ray flux has an index smaller
430: than 1 whereas the reverse holds for systems with an index larger than
431: 1. If the normalisation constant and distance are well known for one
432: source, we can in principle estimate the distance towards another
433: source which has the same index in the radio -- X--ray correlation
434: from the difference in the normalisations.
435:
436: Assume we have measured the radio -- X--ray flux correlation for two
437: sources (both sources must have the same index). If the measured
438: normalisations of both relations are K$_1$ and K$_2$, the distance
439: d$_1=(\frac{K_1}{K_2})^{2-2b}\times d_2$, where b is the index of the
440: radio -- X--ray correlation. Obviously, if there are differences in
441: the jet formation for the different sources the dependence of the
442: normalisation on source distance would be diluted.
443:
444: \begin{figure*}
445: \includegraphics[width=12cm]{relation.radio.xray.ps}
446: \caption{Nearly simultaneous radio and X--ray observations of the BHC
447: XTE~J1908+094. The power--law X--ray -- radio correlation with
448: index 0.7 index as found before in other sources is indicated with
449: the dashed line. The diamond is the point for which we estimated
450: the X--ray flux (see text). The arrows indicate 3 $\sigma$ upper
451: limits on the radio flux (VLA and WSRT). }
452: \label{xrrel}
453: \end{figure*}
454:
455: \section{Discussion}
456:
457: We have obtained (nearly) simultaneous VLA and WSRT radio and {\it
458: Chandra} X--ray observations of the BHC XTE~J1908+094 during the decay
459: after an outburst. We find that: \newline {\it (i)} Limits on a
460: power--law correlation between radio and X--ray flux suggest that the
461: power--law index may be larger than that found before in other
462: BHCs.\newline {\it (ii)} The rate of decay increases from a factor of
463: $\sim$5 in $\sim$25 days to a factor of $\sim$750 in $\sim$25
464: days. \newline {\it (iii)} The source spectrum hardens during the
465: decay. \newline Below we will discuss these findings in more detail.
466:
467: Previously, \scite{2003A&A...400.1007C} and
468: \scite{2003MNRAS.344...60G} found that there is a correlation between
469: the radio and X--ray flux over more than 4 orders of magnitude. The
470: index of the power--law relation they fitted was consistent with being
471: the same for several sources at a value of 0.7$\pm$0.1 (Gallo et
472: al.~2003). If we assume that the radio flux of the source on April 19,
473: 2003 was at the 1~$\sigma$ level of the WSRT upper limit the index was
474: 1.5$^{+0.45}_{-0.3}$ for a two--point power--law decay for
475: XTE~J1908+094. This is consistent at the $\sim3\sigma$ level with the
476: value of 0.7 found before. However, there are some caveats. This index
477: was determined over a limited range in X--ray flux and using very few
478: measurements only. Furthermore, we assumed that the radio spectral
479: index is flat with $\alpha=0$ (S$_\nu\propto\nu^{\alpha}$). Finally,
480: the radio flux on Qpril 19, 2003 may have been much lower than what we
481: have assumed; this would make the correlation index steeper.
482:
483: Previously, \scite{2003MNRAS.342L..67M} found an index of $\sim$1.4
484: for the radio -- X--ray correlation over a small range in X--ray flux
485: in the neutron star system 4U~1728--34. Possibly, XTE~J1908+094 is a
486: neutron star as well. However, \scite{2002A&A...394..553I} and
487: \scite{2002xrb..confE..11G} argue strongly in favour of a BHC nature
488: for XTE~J1908+094 on the basis of the observed outburst X--ray
489: spectral and timing properties (see also Gogos et al.~2004,
490: submitted), but obviously a dynamical mass estimate showing the mass
491: of the compact object to be more than 3 M$_\odot$ would settle the
492: issue.
493:
494: If we follow the line of reasoning laid--out in
495: \scite{2003MNRAS.343L..99F} but taking ${\rm L_{Radio}\propto
496: L_X^{1.4}}$ instead of ${\rm L_{Radio}\propto L_X^{0.7}}$ we find
497: that ${\rm L_{Jet}\propto L_X}$. Hence, the ratio between jet and
498: accretion power (which is assumed to be tracked by the X--ray
499: luminosity) remains the same as the source flux decays. We note that
500: the fact that the index was found to be 1.4 for a neutron star system
501: does not affect the conclusion of \scite{2003MNRAS.343L..99F} that the
502: difference in quiescent X--ray luminosity between BHCs and neutron
503: star soft X--ray transients can be explained without the need of
504: advection of energy across the event horizon as long as the index for
505: the BHCs is 0.7.
506:
507: Recently, \scite{2003MNRAS.343L..59H} showed that the index of 0.7
508: follows naturally for several jet--models if one assumes that those
509: jet models are scale invariant (i.e.~the Schwarzschild radius, r$_s$,
510: is the only relevant length scale for jet formation). Building on the
511: work of \scite{2003MNRAS.343L..59H}, \scite{merheinzdim2003} also
512: showed that other indexes for the radio -- X--ray correlations could
513: be found, i.e.~for both a gas and a radiation pressure dominated disc
514: the index would be $\sim$1.4, whereas it would be close to 0.7 for
515: radiatively inefficient accretion flows. \scite{2003A&A...397..645M}
516: show that the model explaining part of the X--ray emission as jet
517: synchrotron emission (\pcite{2001A&A...372L..25M}) can explain the 0.7
518: index as well as the normalisation of the radio -- X--ray correlation.
519:
520: Hence, it seems that in XTE~J1908+094, during the part of the decay
521: that we covered with our radio and X--ray observations, a standard
522: geometrically thin optically thick disc plus a corona could have been
523: present. Why accretion would proceed via a geometrically thin disc in
524: XTE~J1908+094 whereas in other sources it is thought that the standard
525: disc is not present in the low/hard state is unclear. Perhaps it has
526: something to do with the luminosity levels at which the various
527: sources are observed so far, but since the distance to XTE~J1908+094
528: is ill--constrained (\pcite{2002A&A...394..553I} argue that the
529: distance must be larger than 3 kpc) the source luminosity is not well
530: known.
531:
532: From Fig.~\ref{xdecay} it is clear that the rate of decay increased
533: enormously after April 12, 2003. Such a steep decrease has been
534: observed before for several BHCs and neutron star soft X--ray
535: transients (e.g.~\pcite{1997ApJ...491..312C}). In neutron star systems
536: this has been interpreted as evidence for the onset of the propeller
537: effect (\pcite{1998ApJ...499L..65C}), however, since such a drop in
538: luminosity seems to be common for BHCs as well, this interpretation
539: may need to be revised (\pcite{2003MNRAS.341..823J}).
540:
541: Finally, we find that the X--ray spectrum hardens between the first
542: and second X--ray observation. Spectral hardening is often observed
543: during/just after a transition to the low/hard state
544: (cf.~\pcite{2001ApJ...563..229T}). However, the findings of Gogos et
545: al.~(2004) show that XTE~J1908+094 was already in the low/hard state
546: several months before the first {\it Chandra} observation was made.
547: Perhaps the source changed from the low/hard state back to a soft
548: state in between the {\it RXTE} and the {\it Chandra} observations.
549: Such a change would be consistent with the fact that the limit on the
550: radio -- X--ray correlation index is close to 1.4. We conclude that
551: more observations at these low flux levels (and likely low
552: luminosities) are necessary to determine the behaviour of these
553: sources when they return back to quiescence.
554:
555: \section*{Acknowledgments}
556: \noindent
557: We would like to thank the {\it Chandra} Director, Harvey Tananbaum,
558: for approving these DDT observations, Yousaf Butt for help with
559: selecting the best observation modes, and the referee for useful
560: comments which helped improve the manuscript. PGJ was supported by EC
561: Marie Curie Fellowship HPMF--CT--2001--01308 and is supported by NASA
562: through Chandra Postdoctoral Fellowship grant number PF3--40027
563: awarded by the Chandra X--ray Center, which is operated by the
564: Smithsonian Astrophysical Observatory for NASA under contract
565: NAS8--39073. This research made use of results provided by the
566: ASM/RXTE teams at MIT and at the RXTE SOF and GOF at NASA's GSFC.
567:
568: \begin{thebibliography}{{{Rupen}, {Dhawan} \& {Mioduszewski}}{2002}}
569:
570: \bibitem[\protect\citefmt{{Arnaud}}{1996}]{ar1996}
571: {Arnaud}~K.~A., 1996, in ASP Conf. Ser. 101: Astronomical Data Analysis
572: Software and Systems V.
573: \newblock p.~17
574:
575: \bibitem[\protect\citefmt{{Campana} {\rm et~al.}}{1998}]{1998ApJ...499L..65C}
576: {Campana}~S., {Stella}~L., {Mereghetti}~S., {Colpi}~M., {Tavani}~M.,
577: {Ricci}~D., {Fiume}~D.~D., {Belloni}~T., 1998, \apjl, 499, L65
578:
579: \bibitem[\protect\citefmt{{Chaty}, {Mignani} \&
580: {Israel}}{2002}]{2002MNRAS.337L..23C}
581: {Chaty}~S., {Mignani}~R.~P., {Israel}~G.~L., 2002, \mnras, 337, L23
582:
583: \bibitem[\protect\citefmt{{Chen}, {Shrader} \&
584: {Livio}}{1997}]{1997ApJ...491..312C}
585: {Chen}~W., {Shrader}~C.~R., {Livio}~M., 1997, \apj, 491, 312
586:
587: \bibitem[\protect\citefmt{{Corbel} {\rm et~al.}}{2003}]{2003A&A...400.1007C}
588: {Corbel}~S., {Nowak}~M.~A., {Fender}~R.~P., {Tzioumis}~A.~K., {Markoff}~S.,
589: 2003, \aap, 400, 1007
590:
591: \bibitem[\protect\citefmt{{Davis}}{2001}]{2001ApJ...562..575D}
592: {Davis}~J.~E., 2001, \apj, 562, 575
593:
594: \bibitem[\protect\citefmt{{Dickey} \& {Lockman}}{1990}]{1990ARA&A..28..215D}
595: {Dickey}~J.~M., {Lockman}~F.~J., 1990, \araa, 28, 215
596:
597: \bibitem[\protect\citefmt{{Falcke}, {Kording} \& {Markoff}}{2003}]{falcke2003}
598: {Falcke}~H., {Kording}~E., {Markoff}~S., 2003, \aap, , submitted
599:
600: \bibitem[\protect\citefmt{{Fender}, {Gallo} \&
601: {Jonker}}{2003}]{2003MNRAS.343L..99F}
602: {Fender}~R.~P., {Gallo}~E., {Jonker}~P.~G., 2003, \mnras, 343, L99
603:
604: \bibitem[\protect\citefmt{{Fender}}{2001}]{2001MNRAS.322...31F}
605: {Fender}~R.~P., 2001, \mnras, 322, 31
606:
607: \bibitem[\protect\citefmt{{Gallo}, {Fender} \&
608: {Pooley}}{2003}]{2003MNRAS.344...60G}
609: {Gallo}~E., {Fender}~R.~P., {Pooley}~G.~G., 2003, \mnras, 344, 60
610:
611: \bibitem[\protect\citefmt{{Garcia} {\rm et~al.}}{2001}]{2001ApJ...553L..47G}
612: {Garcia}~M.~R., {McClintock}~J.~E., {Narayan}~R., {Callanan}~P., {Barret}~D.,
613: {Murray}~S.~S., 2001, \apjl, 553, L47
614:
615: \bibitem[\protect\citefmt{{Garnavich}, {Quinn} \&
616: {Callanan}}{2002}]{2002IAUC.7877....4G}
617: {Garnavich}~P., {Quinn}~J., {Callanan}~P., 2002, in International Astronomical
618: Union Circular.
619: \newblock p.~4
620:
621: \bibitem[\protect\citefmt{{Gogus} {\rm et~al.}}{2002}]{2002xrb..confE..11G}
622: {Gogus}~E. {\rm et~al.}, 2002, in X-ray Binaries in the Chandra and XMM-Newton
623: Era (with an emphasis on Targets of Opportunity), Abstracts of the conference
624: held 14-15 November, 2002 at MIT Draper Labs Extension, Cambridge, MA. Hosted
625: by Chandra X-ray Center, 2002.
626:
627: \bibitem[\protect\citefmt{{Gogus} {\rm et~al.}}{2004}]{gogus2004}
628: {Gogus}~E. {\rm et~al.}, 2004, accepted, \apj
629:
630: \bibitem[\protect\citefmt{{Heinz} \& {Sunyaev}}{2003}]{2003MNRAS.343L..59H}
631: {Heinz}~S., {Sunyaev}~R.~A., 2003, \mnras, 343, L59
632:
633: \bibitem[\protect\citefmt{{Houck} \& {Denicola}}{2000}]{2000adass...9..591H}
634: {Houck}~J.~C., {Denicola}~L.~A., 2000, in ASP Conf. Ser. 216: Astronomical Data
635: Analysis Software and Systems IX.
636: \newblock p.~591
637:
638: \bibitem[\protect\citefmt{{in 't Zand}, {Capalbi} \&
639: {Perri}}{2002}]{2002IAUC.7873....1I}
640: {in 't Zand}~J.~J.~M., {Capalbi}~M., {Perri}~M., 2002, in International
641: Astronomical Union Circular.
642: \newblock p.~1
643:
644: \bibitem[\protect\citefmt{{in't Zand} {\rm et~al.}}{2002}]{2002A&A...394..553I}
645: {in't Zand}~J.~J.~M., {Miller}~J.~M., {Oosterbroek}~T., {Parmar}~A.~N., 2002,
646: \aap, 394, 553
647:
648: \bibitem[\protect\citefmt{{Jonker} {\rm et~al.}}{2003}]{2003MNRAS.341..823J}
649: {Jonker}~P.~G., {M{\' e}ndez}~M., {Nelemans}~G., {Wijnands}~R., {van der
650: Klis}~M., 2003, \mnras, 341, 823
651:
652: \bibitem[\protect\citefmt{{Kong} {\rm et~al.}}{2002}]{2002ApJ...570..277K}
653: {Kong}~A.~K.~H., {McClintock}~J.~E., {Garcia}~M.~R., {Murray}~S.~S.,
654: {Barret}~D., 2002, \apj, 570, 277
655:
656: \bibitem[\protect\citefmt{{Maccarone}, {Gallo} \&
657: {Fender}}{2003}]{2003MNRAS.345L..19M}
658: {Maccarone}~T.~J., {Gallo}~E., {Fender}~R., 2003, \mnras, 345, L19
659:
660: \bibitem[\protect\citefmt{{Markoff}, {Falcke} \&
661: {Fender}}{2001}]{2001A&A...372L..25M}
662: {Markoff}~S., {Falcke}~H., {Fender}~R., 2001, \aap, 372, L25
663:
664: \bibitem[\protect\citefmt{{Markoff} {\rm et~al.}}{2003}]{2003A&A...397..645M}
665: {Markoff}~S., {Nowak}~M., {Corbel}~S., {Fender}~R., {Falcke}~H., 2003, \aap,
666: 397, 645
667:
668: \bibitem[\protect\citefmt{{Merloni}, {Heinz} \& {Di
669: Matteo}}{2003}]{merheinzdim2003}
670: {Merloni}~A., {Heinz}~S., {Di Matteo}~T., 2003, \mnras, , submitted
671:
672: \bibitem[\protect\citefmt{{Migliari} {\rm et~al.}}{2003}]{2003MNRAS.342L..67M}
673: {Migliari}~S., {Fender}~R.~P., {Rupen}~M., {Jonker}~P.~G., {Klein-Wolt}~M.,
674: {Hjellming}~R.~M., {van der Klis}~M., 2003, \mnras, 342, L67
675:
676: \bibitem[\protect\citefmt{{Rupen}, {Dhawan} \&
677: {Mioduszewski}}{2002}]{2002IAUC.7874....1R}
678: {Rupen}~M.~P., {Dhawan}~V., {Mioduszewski}~A.~J., 2002, \iaucirc, 7874, 1
679:
680: \bibitem[\protect\citefmt{{Sault}, {Teuben} \&
681: {Wright}}{1995}]{1995adass...4..433S}
682: {Sault}~R.~J., {Teuben}~P.~J., {Wright}~M.~C.~H., 1995, in ASP Conf. Ser. 77:
683: Astronomical Data Analysis Software and Systems IV.
684: \newblock p.~433
685:
686: \bibitem[\protect\citefmt{{Stirling} {\rm et~al.}}{2001}]{2001MNRAS.327.1273S}
687: {Stirling}~A.~M., {Spencer}~R.~E., {de la Force}~C.~J., {Garrett}~M.~A.,
688: {Fender}~R.~P., {Ogley}~R.~N., 2001, \mnras, 327, 1273
689:
690: \bibitem[\protect\citefmt{{Tomsick}, {Corbel} \&
691: {Kaaret}}{2001}]{2001ApJ...563..229T}
692: {Tomsick}~J.~A., {Corbel}~S., {Kaaret}~P., 2001, \apj, 563, 229
693:
694: \bibitem[\protect\citefmt{{Wagner} \& {Starrfield}}{2002}]{2002ATel...86....1W}
695: {Wagner}~R.~M., {Starrfield}~S., 2002, The Astronomer's Telegram, 86, 1
696:
697: \bibitem[\protect\citefmt{{Weisskopf}}{1988}]{1988SSRv...47...47W}
698: {Weisskopf}~M.~C., 1988, Space Science Reviews, 47, 47
699:
700: \bibitem[\protect\citefmt{{Woods} {\rm et~al.}}{2002}]{2002IAUC.7856....1W}
701: {Woods}~P.~M., {Kouveliotou}~C., {Finger}~M.~H., {Gogus}~E., {Swank}~J.,
702: {Markwardt}~C., {Strohmayer}~T., 2002, in International Astronomical Union
703: Circular.
704: \newblock p.~1
705:
706: \end{thebibliography}
707: \end{document}
708:
709:
710:
711:
712:
713:
714:
715: