0710.3351/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass{emulateapj}
3: 
4: \newcommand\spi{{\em SPITZER}}%
5: \newcommand\igrj{IGR J16318--4848}%
6: \newcommand\gx{GX 301--2}%
7: \newcommand\integ{{\em INTEGRAL}}%
8: \newcommand\msun{$M_{\odot}$}%
9: \newcommand\av{$A_{\rm V}$}%
10: \newcommand\nh{$N_{\rm H}$}%
11: \newcommand\hh{$\rm H_{\rm 2}$}%
12: \newcommand\necubic{$N_{\rm e}$}%
13: \newcommand\cmcubic{$\rm cm^{\rm -3}$}%
14: \newcommand{\niii}{[\ion{Ni}{2}]}
15: \newcommand{\neii}{[\ion{Ne}{2}]}
16: \newcommand{\neiii}{[\ion{Ne}{3}]}
17: \newcommand{\feii}{[\ion{Fe}{2}]}
18: \newcommand{\feiii}{[\ion{Fe}{3}]}
19: \newcommand{\siii}{[\ion{S}{3}]}
20: \newcommand{\silii}{[\ion{Si}{2}]}
21: 
22: \slugcomment{Submitted: \today}
23: 
24: \shorttitle{\spi\ Mid-Infrared Spectroscopy of Highly-Obscured X-ray Binaries}
25: \shortauthors{Moon et al.}
26: 
27: %\received{2003 December 30}
28: \begin{document}
29: 
30: \title{The Rich Mid-Infrared Environments of Two Highly-Obscured
31:   X-ray Binaries: \spi\ Observations of \igrj\ and \gx }
32: 
33: \author{Dae-Sik Moon\altaffilmark{1},
34: David L. Kaplan\altaffilmark{2},
35: William T. Reach\altaffilmark{3},
36: Fiona A. Harrison\altaffilmark{4},
37: Jeong-Eun Lee\altaffilmark{5},
38: Peter G. Martin\altaffilmark{6} }
39: \altaffiltext{1}{Department of Astronomy and Astrophysics, University of Toronto, Toronto, ON M5S 3H4, Canada; moon@astro.utoronto.ca}
40: \altaffiltext{2}{Pappalardo Fellow, Kavli Institute for Astrophysics
41:   \& Space Research and Department of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139; dlk@space.mit.edu}
42: \altaffiltext{3}{Infrared Processing Analysis Center, California Institute of Technology, MS 220-6, Pasadena, CA 91125; reach@ipac.caltech.edu}
43: \altaffiltext{4}{Space Radiation Laboratory, California Institute of Technology, MC 220-47, Pasadena, CA 91125; fiona@srl.caltech.edu}
44: %\altaffiltext{5}{Hubble Fellow, Department of Physics and Astronomy, University of California, Los Angeles, CA 90095}
45: \altaffiltext{5}{Department of Astronomy and Space Science, Sejong University, Seoul 143-747, Korea; jelee@sejong.ac.kr}
46: \altaffiltext{6}{Canadian Institute for Theoretical Astrophysics, University of Toronto, Toronto, ON M5S 3H8, Canada; pgmartin@cita.utoronto.ca}
47: 
48: \begin{abstract}
49: We present the results of \spi\ mid-infrared spectroscopic
50: observations of two highly-obscured massive X-ray binaries: \igrj\ and
51: \gx.  Our observations reveal for the first time the extremely rich
52: mid-infrared environments of this type of source, including multiple
53: continuum emission components (a hot component with $T$ $>$ 700~K and
54: a warm component with $T$ $\sim$ 180~K) with apparent silicate
55: absorption features, numerous \ion{H}{1} recombination lines, many
56: forbidden ionic lines of low ionization potentials, and pure
57: rotational \hh\ lines.  This indicates that both sources have hot and
58: warm circumstellar dust, ionized stellar winds, extended low-density
59: ionized regions, and photo-dissociated regions.  
60: It appears difficult to attribute the total optical extinction
61: of both sources to the hot and warm dust components, which suggests
62: that there could be an otherwise observable colder dust component responsible
63: for the most of the optical extinction and silicate absorption features.
64: The observed mid-infrared
65: spectra are similar to those from Luminous Blue Variables, indicating
66: that the highly-obscured massive X-ray binaries may
67: represent a previously unknown evolutionary phase of X-ray binaries with
68: early-type optical companions.  Our results highlight the importance and utility
69: of mid-infrared spectroscopy to investigate highly-obscured X-ray
70: binaries.
71: \end{abstract}
72: 
73: \keywords{circumstellar matter --- infrared: stars --- stars: emission-line, Be --- X-rays: binaries
74: --- X-rays: individual (\gx, \igrj)}
75: 
76: \section{Introduction} \label{sec_intro}
77: 
78: Recently, a large number of highly-obscured (e.g., \nh\ $\ge$
79: 10$^{23}$ cm$^{-2}$) massive X-ray binaries have been discovered by
80: the \integ\ hard X-ray ($\ge$ 15 keV) satellite \citep{wet03}.  The
81: prototypical case is \igrj\ which shows variable high obscuration in
82: the X-ray, sometimes reaching \nh\ $\simeq$ 2 $\times$ 10$^{24}$
83: \citep{coet03, walteret03}.  Its bright optical and near-infrared (IR)
84: counterpart is an early B-type supergiant star with numerous emission
85: lines \citep{fc04}.  Interestingly the obscuration toward \igrj\
86: obtained in the optical and near-IR wavebands (\av\ $\sim$ 18) is
87: almost two orders of magnitude smaller than that inferred from the
88: X-rays, which suggests that
89: the extreme obscuration seen in the X-ray is intrinsic only to the
90: X-ray source.  However, \av\ $\sim$ 18 is still  greater than the
91: interstellar obscuration, indicating the existence of substantial
92: circumstellar material around the supergiant companion.
93: 
94: In order to fully understand the implications of the \integ\
95: discoveries --- specifically if they imply existence of a separate
96: class of highly-obscured X-ray binaries --- we must investigate any
97: similarities between the new \integ\ sources and previously known
98: sources.  As already suggested by \citet{ret03}, the X-ray pulsar \gx\
99: appears similar to the new \integ\ highly-obscured massive X-ray
100: binaries: it has variable X-ray obscuration of \nh\ $\simeq$
101: 10$^{23}$--10$^{24}$ cm$^{-2}$, the compact source is a
102: neutron star, and the optical companion is an early B-type supergiant
103: \citep[or hypergiant;][]{ket95}, like \igrj.  Recently, \citet[][;
104: hereafter Paper~I]{kmr06}, using optical, near-, and mid-IR
105: ($\le$~20~$\mu$m) spectral energy distributions (SEDs), have found that
106: both sources have strong mid-IR excesses that they identify as
107: continuum emission from hot dust.  This is suggestive that they have
108: very similar circumstellar material which may be related to their
109: strong X-ray obscuration.  In this {\em Letter}, we present the
110: results of \spi\ mid-IR spectroscopic observations of \igrj\ and \gx,
111: showing that both sources indeed have very similar rich mid-IR
112: properties previously unknown for X-ray binaries.
113: 
114: \section{Observations and Data Reduction} \label{sec_obs}
115: 
116: We observed \igrj\ and \gx\ with {\em InfraRed Spectrogrpah} (IRS)
117: aboard \spi\ \citep{jhet04, wet04} on 2005 September 8 and 2005 July
118: 2, respectively.  The duration of the total observations was 1.32
119: (\igrj) and 0.82 (\gx) hours with the all six IRS modules in operation
120: in the standard staring mode, obtaining two nods of spectra separated
121: by 1/3 of the slit length.  For the basic data reduction, we used the
122: co-added Basic Calibration Data (BCD) produced through the standard
123: \spi\ Science Center data reduction pipeline (S13.2.0), and also the
124: \spi\ IRS Custom Extraction (i.e., SPICE) software (v1.3-beta1).  In
125: the low-resolution mode where the slit lengths are large (i.e.,
126: 57\arcsec\ for 5.2--14.5 $\mu$m and 168\arcsec\ for 14.0--38.0
127: $\mu$m), we subtracted out background emission of one nod using the
128: spectrum of the another nod, and combined the two
129: background-subtracted spectra for the final spectrum.  However, in the
130: high-resolution mode, the small slit length (i.e., 11\farcs3 for
131: 9.9--19.6 $\mu$m and 22\farcs3 for 18.7--37.2 $\mu$m), together with
132: the high brightnesses of the sources, made it impossible to subtract out
133: the background.  Therefore, we just combined the spectra from the two
134: nods for the final high-resolution spectrum without any background
135: subtraction. In addition, we used IRSCLEAN (v1.5) to mask and clean
136: rogue pixels from the co-added BCD images, and then interpolated over
137: neighboring normal pixels.
138: 
139: \section{Low-Resolution Spectrum: Continuum and Broad Features} \label{sec_lowres}
140: 
141: Some emission features,
142: such as ionic forbidden lines, \hh\ lines,
143: and Polycyclic Aromatic Hydrocarbon (PAH) emission,
144: present in the low-resolution spectra appear extended.
145: Here we consider just the emission from point sources, 
146: and defer discussion of the extended emission to a separate paper.
147: Figure~\ref{fig:igrj} presents the background-subtracted
148: low-resolution spectrum of \igrj, 
149: as well as the best-fit spectra of the continuum emission and the residual of the SED fit (see below).
150: Note that the low-resolution spectrum is extracted for the continuum source (i.e., \igrj)
151: based on the standard point source extraction procedure of SPICE.
152: Several features are prominent in Figure~\ref{fig:igrj}, 
153: including complicated multi-component continuum emission,
154: broad and strong absorption around 9.7 and 18 $\mu$m, and numerous
155: emission lines.  The spectrum of \gx\ (Figure~\ref{fig:gx}) is similar,
156: with strong  continuum emission,  silicate
157: absorptions feature around 9.7 $\mu$m, and several emission lines.
158: 
159: As in Figures~\ref{fig:igrj} and \ref{fig:gx},
160: the silicate absorption features and numerous emission lines 
161: make it difficult to carry out any meaningful independent SED fits
162: of the shorter wavelength (i.e., $\lambda$ $\le$ 8 $\mu$m) data of the IRS spectra.
163: Therefore, instead, we first examined if the IRS SEDs at the shorter wavelengths
164: are consistent with what have been found in Paper~I using only the line-free regions.  
165: According to Paper~I, the SEDs of \igrj\ and \gx\ at $\lambda$~$\le$~8~$\mu$m have 
166: contributions from hot (i.e.,T $\simeq$ 1040~K for \igrj\ and 720~K for \gx) dust emission and
167: stellar emission of a B0I star with temperature of 26000~K,
168: while the contribution of free-free emission is negligible.
169: Our IRS SEDs of the line-free regions in Figures~\ref{fig:igrj} and
170: \ref{fig:gx} at $\lambda$~$\le$~8$\mu$m were fitted very well with
171: those parameters, resulting in the reduced chi-square $\chi^2_\nu$
172: $\le$ 1.  We used the extinction \av\ = 18.5 and 7.1 for \igrj\ and
173: \gx, respectively, as in Paper~I.  However, in the longer wavelength
174: (i.e., $\lambda$~$\ge$~22~$\mu$m) range, we were unable to fit the IRS
175: SEDs with the given hot dust and stellar parameters because the best
176: fit resulted in unacceptably large chi-square values.  We therefore
177: performed SED fits with three components (i.e., hot dust, stellar, and
178: warm dust component) using the entire line-free SEDs in the
179: $\lambda$~$\le$~8~$\mu$m and $\ge$~22~$\mu$m range.  Over the fits the
180: parameters of the warm dust component were treated as free parameters,
181: while those of the hot dust and stellar component were fixed to be the
182: values in Paper~I.  As a result, we obtained the warm dust
183: temperature of 190~K (\igrj) and 170~K (\gx) with $\chi^2_\nu$
184: $\simeq$ 1.2 (\igrj) and 1.5 (\gx).  We used the extinction
185: coefficients compiled in \citet{m2000} excluding the 8--22 $\mu$m
186: range in order to avoid the uncertainties of the coefficients
187: associated with the silicate absorption features.  We instead
188: interpolated the extinction coefficients at $\lambda$ $\le$ 8 and
189: $\ge$ 22 $\mu$m to calculate the silicate absorption-free extinction
190: coefficients in the $\lambda$ = 8--22 $\mu$m range, and used them in the fits.
191: The main panel of Figures~\ref{fig:igrj} and \ref{fig:gx} presents the results of
192: our best SED fit obtained by excluding the silicate absorption features,
193: where we can confirm that the fit matches nicely 
194: the observed SED except the $\lambda$ = 8--22 $\mu$m range.
195: In the bottom panel, on the other hand,
196: we show the redisuals of the fit obtained by subtracting 
197: the fluxes computed by the best fit from the observed values.
198: For this residual construction, in order to see if our best SED fit 
199: is consistent with the observed silicate absorption features,
200: we combined the best-fit parameters obtained at $\lambda$ $\le$ 8 and $\ge$ 22 $\mu$m 
201: with the extinction coefficients in the $\lambda$ = 8--22 $\mu$m range \citep{m2000},
202: and calculated the silicate absorption feature-associated fluxes expected by the best SED fit
203: in the $\lambda$ = 8--22 $\mu$m range.
204: The residuals in Figures~\ref{fig:igrj} and \ref{fig:gx} show
205: that overall our best SED fit agrees with the silicate absorption features reasonably well,
206: although there appears to be some small discrepancies for \igrj.
207: 
208: The masses of the dust components were estimated under the
209: optically thin assumption \citep{het77}: $M_{\rm dust} =
210: F_{\lambda}d^2 / \kappa_{\lambda}B_{\lambda}(T_{\rm dust})$, where $d$
211: is the distance to the source, $\kappa_{\lambda}$ is the dust mass
212: absorption coefficient, and $B_\lambda(T)$ is the Planck function at
213: temperature $T$.  Using the interstellar mass absorption coefficients
214: of \citet{draine03} at 8 and 25 $\mu$m, the hot and warm dust masses
215: are estimated to be $M_{hot}$ $\simeq$ 4.5 $\times$ 10$^{-9}$ $d_5^2$ \msun\
216: and $M_{warm}$ $\simeq$ 3.1 $\times$ 10$^{-7}$ $d_5^2$ \msun\
217: for \igrj; $M_{hot}$ $\simeq$ 5.7 $\times$ 10$^{-9}$ $d_5^2$ \msun\ and
218: $M_{warm}$ $\simeq$ 1.4 $\times$ 10$^{-7}$ $d_5^2$ \msun\ for \gx, 
219: where $d_5$ is the distance to the sources normalized by 5 kpc. 
220: 
221: The silicate absorption features seen in Figures~\ref{fig:igrj} and
222: \ref{fig:gx} have a tight linear correlation with the optical
223: extinction in the diffuse interstellar medium (ISM),
224: although the correlation has been reported to break down 
225: in dense enviornment where the silicate feature does not 
226: show a monotonic increase with extinction 
227: at \av\ $\geq$ 12 mag \citep[e.g.,][and references therein]{cet07}.
228: The correlation in the diffuse ISM is \av/$\tau_{\rm 9.7}$ = 18.5 
229: \citep[][and references therein; $\tau_{\rm 9.7}$ is the optical depth of the silicate absorption at 9.7 $\mu$m]
230: {draine03}.  
231: If we adopt the above correlation given in the diffuse ISM,
232: the intensity ratio of the observed spectrum to the unabsorbed spectrum 
233: in Figures~\ref{fig:igrj} and \ref{fig:gx} at 9.7 $\mu$m 
234: correspond to \av\ $\sim$~18.5 and $\sim$~6 for \igrj\ and \gx, respectively, 
235: These are very similar to the values obtained in previous studies
236: for the total optical extinctions of the both sources.
237: This suggests that a substantial portion, if not all, 
238: of the silicate features seen in Figures~\ref{fig:igrj} 
239: and \ref{fig:gx} are associated with the optical extinction of \igrj\ and \gx.
240: 
241: \section{High-Resolution Spectrum: Line Emission} \label{sec_highres}
242: 
243: Figures~\ref{fig:igrjhigh} and \ref{fig:gxhigh} present high-resolution
244: spectrum of \igrj\ and \gx, revealing numerous emission lines.  
245: (Note that the background emission is {\it NOT} subtracted.)
246: The majority of them are \ion{H}{1} lines, while the
247: rest consists of metallic forbidden lines, pure rotational \hh\ lines,
248: and 11.3 $\mu$m PAH emission feature.
249: Table~\ref{tbl_HI} lists the intensities of some of the bright
250: \ion{H}{1} lines identified in both sources with high ($>$ 3)
251: signal-to-noise ratio.  The line intensities and the signal-to-noise
252: ratio were estimated with Gaussian profile fits.  The intensities of
253: the \ion{H}{1} lines are consistent with gas at $T$ $\simeq$ 10$^4$ K
254: and with \necubic\ $\simeq$ 10$^4$ \cmcubic\ in the Menzel Case~B
255: state, following \citet{fc04}, although the intensity ratios are not
256: very sensitive to those parameters \citep{hs87}.
257: 
258: Table~\ref{tbl_other} lists the parameters of forbidden ionic lines and \hh\ lines.
259: For the forbidden lines, \neii, \neiii, \siii, and \silii\ were detected
260: in both sources, while \niii\ and \feii\ were only in \igrj; \feiii\
261: is only in \gx. 
262: The \hh\ lines were detected in both sources.  The intensity ratio of
263: the \siii\ lines of 18.7 and 33.5 $\mu$m is diagnostic of the number
264: density in the temperature range of 5000--20000~K.  The
265: extinction-corrected \siii\ line surface brightness ratio of 18.7/33.5
266: $\mu$m is 1.21 for \igrj\ and 1.68 for \gx: according to the calculation
267: of \citet{het84} and \citet{aet99}, the ratios correspond to the
268: electron number density \necubic\ $\sim$ 1000 \cmcubic\ for both
269: sources.  On the other hand, the intensity ratio of the \neii\ line
270: (12.8 $\mu$m) and the \neiii\ line (15.6 $\mu$m) is sensitive to the
271: hardness of the radiation, or temperature of the environment.  If we
272: use the calculation of \citet{ket96}, the extinction-corrected
273: intensity ratio of \neii/\neiii\ is consistent with $\sim$
274: 4~$\times$~10$^4$~K for \igrj\ and $\sim$ 4.5~$\times$~10$^4$~K for
275: \gx.  The three \hh\ lines detected in the both sources are pure
276: rotational transition lines of \hh~(0,0)~S(0,1,2) which are usually
277: excited by collisions.  Assuming that the bottom two lines follow the
278: distribution of the local thermodynamic equilibrium and are optically
279: thin, the surface brightness ratios are consistent with the excitation
280: temperature $T_{\rm ex}$ $\sim$ 1000~K.
281: The estimated \hh\ column densities are $\sim$ 1 $\times$ 10$^{19}$
282: cm$^{-2}$ (\igrj) and 5 $\times$ 10$^{18}$ cm$^{-2}$ (\gx).  This
283: implies that the optical extinction associated with the warm molecular
284: gas is very small, \av\ $\ll$ 1, based on the relation \nh\ = 1.87
285: $\times$ 10$^{21}$ \av\ \citep{draine03}.
286: 
287: \section{Discussion and Conclusion} \label{sec_dis_sum}
288: 
289: Our \spi\ spectroscopic observations of \igrj\ and \gx\ have revealed
290: for the first time the rich mid-IR environment of highly-obscured X-ray binaries.
291: This includes two dust components with prominent silicate absorption,
292: numerous \ion{H}{1} recombination lines, many forbidden ionic  lines,
293: and pure rotational \hh\ lines.
294: Based on the observed spectra, we infer the following components for
295: \igrj\ and \gx: (1) hot ($T$ $>$ 700~K) and warm ($T$ $\sim$ 180~K)
296: circumstellar dust; (2) ionized stellar winds responsible for the
297: \ion{H}{1} lines; (3) extended low-density ionized regions for the
298: forbidden lines; and (4) photo-dissociated regions associated with the
299: PAH, \hh\ and possibly the \silii\ line emission.  For the forbidden
300: lines, all the detected lines have relatively low ionization potentials
301: like in starburst galaxies where the radiation is relatively soft (compared with active
302: galactic nuclei, for instance).  This may indicate that the
303: illumination of hard X-rays from the central compact X-ray source is
304: not primarily responsible for the forbidden line emission.  However,
305: the inferred temperature for the radiation exciting the \neii\ and
306: \neiii\ lines is hotter than the stellar photospheres, so there can
307: be some contribution from the compact object.
308: Considering that \niii\ and \feii\ were detected only in \igrj\ while
309: \feiii\ was only in \gx, the radiation field of \igrj\ may be
310: softer than \gx, as demonstrated by the small temperature difference between
311: the two sources (see \S~\ref{sec_highres}).
312: 
313: Perhaps the most natural explanation for the origin of the hot dust
314: component relies on dust formation in the dense outflows from the
315: early-type companions of \igrj\ and \gx, as is the case in most sgB[e]
316: stars (i.e., B-type supergiants with forbidden emission lines).
317: However, the origin of the warm circumstellar dust component is very uncertain, 
318: and B[e] stars seldom show evidence for this type of warm dust.  
319: If both the hot and warm dust components have spherical shell geometres around the central star,
320: then the associated optical extinctions are:
321: \av\ $\simeq$ 4 $\times$ 10$^{-4}$ $Q_{\rm abs}$ ($M_{\rm d}$/10$^{-6}$ \msun)
322: ($T_{\rm d}$/100 K)$^4$ ($L_{\rm UV}$/10$^{39}$ ergs s$^{-1}$),
323: where $M_{\rm d}$ and $T_{\rm d}$ are the mass and temperature of the dust components,
324: $Q_{\rm abs}$ $<$ 1 is the dust absorption coefficient, 
325: and $L_{\rm UV}$ is the ultra-violet luminosity of the central star.
326: This gives the optical extinctions \av\ $\ll$ 1 for both the hot and warm dust components
327: of \igrj\ and \gx. (Here we use 1 $\times$ 10$^{39}$ ergs s$^{-1}$ for $L_{\rm UV}$ for both sources.)
328: Therefore, under the assumption of spherical shell geometry, 
329: both the hot and warm dust components of \igrj\ and \gx\ contribute
330: very little to the total optical extinction, suggesting that
331: the hot and warm dust components are {\em NOT} strongly associated with 
332: the silicate absorption features.
333: This also applies to the dust associated with the warm extended \hh\ gas 
334: since the optical extinction from this component is tiny 
335: (i.e., Av $\ll$ 1; see \S~\ref{sec_highres}). 
336: What's then the origin of the silicate absorption features?
337: If it is due to the foreground ISM, we would expect to see the 
338: $\rm CO_2$ ice feature around 15 $\mu$m, especially for \igrj, based on the intensity ratio
339: between the silicate absorption feature and ice feature found in the ISM \citep{ket05}. 
340: The absence of the ice feature in our spectra
341: supports the interpretation that the silicate absorption features are probably
342: not associated with the foreground ISM.
343: One possibility may be the existence of an undisclosed colder (e.g., $\ll$ 100~K) 
344: circumstellar dust component which is responsible for the silicate absorption features
345: and most of the optical extinction.
346: We need further longer wavelength observatons to confirm this possibility.
347: Considering that the 9.7~$\mu$m
348: silicate absorption feature represents the oxygen-rich material, the
349: existence of the 9.7 micron silicate absorption feature 
350: may indicate that the origin of the potential colder dust component 
351: is related to the nucleosynthesis of the progenitors of \igrj\ and \gx\
352: (or their companions).
353: 
354: The optical/near-IR companion of \igrj\ is a sgB[e] star.  Such stars
355: are known to have hot ($T$ $\sim$ 1000~K) circumstellar dust, and
356: probably are evolving into Luminous Blue Variables (LBVs) or
357: Wolf-Rayet stars.  Our mid-IR spectra of \igrj\ and \gx\ are very
358: similar to that of the LBV P~Cygni which shows many \ion{H}{1} lines
359: and forbidden ionic lines \citep{let96b}.  The difference is that
360: while the mid-IR continuum of P~Cygni is due to the free-free emission
361: in stellar winds, the main mid-IR emission of \igrj\ and \gx\ is from
362: multiple dust continua.  (The SEDs of both sources observed here are
363: inconsistent with the free-free emission, as mentioned in Paper~I.)
364: However, we note that some LBV stars have also been observed to have
365: thermal mid-IR dust emission \citep{let96a}. Based on the fact that B[e]
366: stars seldom show mid-IR forbidden line emission, the B-type
367: supergiant (or hypergiant) companions of \igrj\ and \gx\ may be in the
368: evolutionary track to LBVs, implying that the extremely high
369: obscuration seen in some massive X-ray binaries may be a phenomenon
370: associated with the evolutionary phase.  This scenario is also consistent
371: with the fact that the highly-obscured massive X-ray binary Cygnus~X-3
372: has a Wolf-Rayet star companion, together with the circumstellar dust
373: emission of $T$ $\sim$ 250~K \citep{ket02}.  
374: 
375: \acknowledgments
376: This work is based on observations made with the Spitzer Space Telescope, 
377: which is operated by the Jet Propulsion Laboratory, 
378: California Institute of Technology, under a contract with NASA.
379: D.-S.M thanks Elise Furlan for her help in IRS data analysis 
380: and Marten van Kerkwijk for comments.
381: This research was partly supported by the Discovery
382: Grant (327277) of Natural Science and Engineering Research Council of Canada to D.-S.M.
383: 
384: \begin{thebibliography}{}
385: \bibitem[Alexander et al.(1999)]{aet99}
386:     Alexander, T., Sturm, E., Lutz, D., Sternberg, A., Netzer, H., \&  Genzel, R. 1999, \apj, 512, 204
387: \bibitem[Bodaghee et al.(2007)]{bet07}
388:     Bodaghee, A., et al. 2007, A\&A, accepted (astro-ph/0703043)
389: \bibitem[Chiar et al.(2007)]{cet07}
390:     Chiar, J. E., et al. 2007, \apj, 666, L73
391: \bibitem[Courvoisier et al.(2003)]{coet03}
392:     Courvoisier, T. J.-L., et al. 2003, IAU Circ., 8063, 3 
393: \bibitem[Filliatre \& Chaty(2004)]{fc04}
394:     Filliatre, P., \& Chaty, S. \apj, 616, 469
395: \bibitem[Draine(2003)]{draine03}
396:     Draine, B. T. 2003, \araa, 41, 241
397: %\bibitem[Fazio et al.(2004)]{fet04}
398: %    Fazio, G. G., et al. 2004, \apjs, 154, 10
399: \bibitem[Hildebrand et al.(1977)]{het77}
400:     Hildebrand, R. H., et al. 2004, \apjs, 154, 18
401: \bibitem[Houck et al.(1984)]{het84}
402:     Houck, J. R., Shure, M. A., Gull, G. E., \& Herter, T. 1984, \apj, 287, L11
403: \bibitem[Houck et al.(2004)]{jhet04}
404:     Houck, J. R., et al. 2004, \apjs, 154, 18
405: \bibitem[Hummer \& Storey(1987)]{hs87}
406:     Hummer, D. G., \& Storey, P. J. 1987, \mnras, 224, 801
407: \bibitem[Kaplan, Moon, \& Reach(2006)]{kmr06}
408:     Kaplan, D. L., Moon, D.-S., \& Reach, W. T. 2006, \apj, 649, L107 (Paper~I)
409: \bibitem[Kaper et al.(1995)]{ket95}
410:     Kaper, L., et al. 1995, A\&A, 300, 446
411: \bibitem[Koch-Miramond et al.(2002)]{ket02}
412:     Koch-Miramond, L., Abraham, P., Fuchs, Y., Bonnet-Bidaud, J.-M., \& Claret, A. 2002, A\&A, 877, 396
413: \bibitem[Knez et al.(2005)]{ket05}
414:     Knez, C., et al. 2005, \apj, 635, L145
415: \bibitem[Kunze et al.(1996)]{ket96}
416:     Kunze, D., et al. 1996, A\&A, 315, L101
417: \bibitem[Lamers et al.(1996a)]{let96a}
418:     Lamers, H. J. G. L. M., et al. 1996a, A\&A, 315, L225
419: \bibitem[Lamers et al.(1996b)]{let96b}
420:     Lamers, H. J. G. L. M., et al. 1996b, A\&A, 315, L229
421: \bibitem[Mathis (2000)]{m2000}
422:     Mathis, J. S. 2000, in Allen's Astrophysical Quantities, ed. A. N. Cox (London: Athlone Press, Ltd.), 527
423: %\bibitem[Predehl \&\ Schmitt(1995)]{ps95}
424: %    Predehl, P., \& Schmitt, J.~H.~M.~M. 1995, A\&A, 293, 889
425: \bibitem[Revnivtsev et al.(2003)]{ret03}
426:     Revnivtsev, M. G., Sazonov, S. Yu., Gilfanov, M. R., \& Sunyaev, R. A. 2003, Astron. Lett., 29, 587
427: \bibitem[Walter et al.(2003)]{walteret03}
428:     Walter, R., et al. 2003, A\&A, 411, L427
429: \bibitem[Werner et al.(2004)]{wet04}
430:     Werner, M. W., et al. 2004, \apjs, 154, 1
431: \bibitem[Winkler et al.(2003)]{wet03}
432:     Winkler, C., et al. 2003, A\&A, 411, L349
433: \end{thebibliography}
434: 
435: \clearpage
436: \begin{deluxetable}{crr||crr}
437: \tablecolumns{6}
438: \tablewidth{0pt}
439: \tablecaption{Observed Intensities  of \ion{H}{1} Transitions\label{tbl_HI}}
440: \tablehead{
441: \colhead{Transition($\mu$m)} & \colhead{IGR}    & \colhead{GX}     & \colhead{Transition($\mu$m)} & \colhead{IGR}    & \colhead{GX}  \\
442: \colhead{}     & \colhead{J16318} & \colhead{301--2} & \colhead{}     & \colhead{J16318} & \colhead{301--2}}
443: \startdata
444: 17--9(10.26)   &   4.39           &  2.44            &  12--8(10.50)   & 11.57            &  11.40 \\ 
445: 16--9(10.80)   &   5.60           &  2.20            &   9--7(11.31)   & 25.43            &  13.50 \\ 
446: 22--10(11.49)  &   1.74           &  0.78            &  15--9(11.54)   &  6.48            &   1.87 \\ 
447: 20--10(12.16)  &   2.84           &  1.35            &   7--6(12.37)\tablenotemark{a}   & 72.02            &  42.12 \\ 
448: 18--10(13.19)  &   5.15           &  1.70            &  13--9(14.18)   & 10.61            &   3.37 \\ 
449: 23--11(14.30)  &   2.00           &  0.93            & 22--11(14.71)   &  5.02            &   0.87 \\ 
450: 16--10(14.96)  &   5.41           &  1.75            & 20--11(15.82)   &  2.79            &   1.33 \\ 
451: 10--8(16.21)   &  14.66           &  7.32            & 15--10(16.41)   &  3.83            &   1.83 \\ 
452: 19--11(16.59)  &   2.15           &  0.40            &  12--9(16.88)   &  7.35            &   3.17 \\ 
453: 14--10(18.62)  &   4.04           &  1.38            & 16--11(20.92)   &  2.69            &   1.14 \\ 
454: 9--8(27.80)    &  18.71           &  8.86            &  \nodata        &  \nodata         &  \nodata   \\ 
455: \enddata
456: \tablecomments{The values are in 10$^{-14}$ ergs s$^{-1}$ cm$^{-2}$ and are not corrected for extinction.}
457: \tablenotetext{a}{The \ion{H}{1} 11--8 transitions may be blended.}
458: \end{deluxetable}
459: 
460: \clearpage
461: \begin{deluxetable}{crr||crr}
462: \tablecolumns{6}
463: \tablewidth{0pt}
464: \tablecaption{Observed Intensities of Other Lines \label{tbl_other}}
465: \tablehead{
466: \colhead{Line($\mu$m)} & \colhead{IGR}    & \colhead{GX}     & \colhead{Line($\mu$m)} & \colhead{IGR}    & \colhead{GX}  \\
467: \colhead{}     & \colhead{J16318} & \colhead{301--2} & \colhead{}     & \colhead{J16318} & \colhead{301--2}}
468: \startdata
469: \niii(10.68)   &  10.25        &  \nodata  & \niii(12.73)   &  13.33        &  \nodata      \\ 
470: \niii(18.24)   &   1.69        &  \nodata  & \nodata        &  \nodata      &  \nodata      \\ 
471: \neii(12.81)   &  21.44        &  23.41    & \neiii(15.56)  &  2.60         &     15.67     \\ 
472: \feii(17.94)   &  22.54        & \nodata   & \feii(24.52)   &  7.33         &  \nodata      \\
473: \feii(25.99)   &  70.93        & \nodata   & \feii(35.35)   &  15.46        &  \nodata      \\
474: \feiii(22.93)  &  \nodata      &   1.99    & \nodata        & \nodata       &  \nodata      \\
475: \siii(18.71)   &   7.73        &   8.02    & \siii(33.48)   &  40.14        &     25.01     \\
476: \silii(34.82)  &  177.20       &  33.33    & \nodata        & \nodata       &  \nodata      \\
477: \hh(0,0)S(2)(12.28) &  2.88    &  1.44     & \hh(0,0)S(1)(17.03) &  4.66         &    2.05       \\
478: \hh(0,0)S(0)(28.22) & 10.08    &  4.24     & \nodata             &  \nodata      &  \nodata      \\
479: \enddata
480: \tablecomments{The values are in 10$^{-14}$ ergs s$^{-1}$ cm$^{-2}$ and are not corrected for extinction.}
481: \end{deluxetable}
482: 
483: \clearpage
484: \begin{figure}[htf]
485: \plotone{f1.eps}
486: \caption{{\em Main Panel:} The background-subtracted IRS low-resolution spectrum of \igrj.  
487: The solid line represents the observed spectrum, while the dotted line shows the
488: best-fit spectrum combining the hot dust (dashed line), the stellar
489: (dotted-dashed line), and the warm dust (dot-dashed line) components.
490: The SED fit was conducted excluding the wavelengths associated with the silicate absorption
491: features indicated by the arrows around 9.7 and 18 $\mu$m. 
492: {\em Bottom Panel:} The residuals of the SED fits. 
493: Note that for the residual the fluxes of the wavelengths associated with 
494: the silicate absorption features were calculated with the best-fit parameters
495: (see text). The unit of the residual is Jy.}
496: \label{fig:igrj}
497: \end{figure}
498: 
499: \clearpage
500: \begin{figure}[htf]
501: \plotone{f2.eps}
502: \caption{Same as Figure~\ref{fig:igrj}, but for \gx.
503: The silicate absorption features are not as prominent as in \igrj,
504: but are still visible, especially around $\sim$9.7 $\mu$m.} 
505: \label{fig:gx}
506: \end{figure}
507: 
508: \clearpage
509: \begin{figure}[htf]
510: \plotone{f3.eps}
511: \caption{The IRS high-resolution spectrum of \igrj.
512: Note that the background is {\em NOT} subtracted.
513: Some strong lines are marked.}
514: \label{fig:igrjhigh}
515: \end{figure}
516: 
517: \clearpage
518: \begin{figure}[htf]
519: \plotone{f4.eps}
520: \caption{Same as Figure~\ref{fig:igrjhigh}, but for \gx.}
521: \label{fig:gxhigh}
522: \end{figure}
523: 
524: \end{document}
525: