1: \documentclass[12pt,preprint]{aastex}

2: 

3: %\documentclass{aastex}

4: %\usepackage{natbib,psfig}

5: %\usepackage{emulateapj5,apjfonts,natbib,psfig}

6: %\usepackage{graphicx}

7: %

8: %

9: %\documentclass[apj]{emulateapj}

10: %\usepackage{mathptmx}

11: %

12: 

13: \newcommand{\asca}{{\small \it ASCA}}

14: \newcommand{\ginga}{{\small \it Ginga}}

15: \newcommand{\rosat}{{\small \it ROSAT}}

16: \newcommand{\hri}{{\small HRI}}

17: \newcommand{\sax}{{\small \it BeppoSA$\!$X}}

18: \newcommand{\cie}{{\small CIE}}

19: \newcommand{\pie}{{\small PIE}}

20: \newcommand{\dem}{{\small DEM}}

21: \newcommand{\ccd}{{\small CCD}}

22: \newcommand{\agn}{{\small AGN}}

23: \newcommand{\ngc}{{\small NGC~3783}}

24: \newcommand{\iras}{{\small IRAS~13349+2438}}

25: \newcommand{\uta}{{\small UTA}}

26: \newcommand{\hullac}{{\small HULLAC}}

27: \newcommand{\xstar}{{\small XSTAR}}

28: \newcommand{\fwhm}{{\small FWHM}}

29: \newcommand{\mekal}{{\small MEKAL}}

30: \newcommand{\pexrav}{{\small PEXRAV}}

31: \newcommand{\xspec}{{\small XSPEC}}

32: \newcommand{\rrc}{{\small RRC}}

33: \newcommand{\clr}{{\small CLR}}

34: \newcommand{\xmm}{{\it XMM-Newton}}

35: \newcommand{\rgs}{{\small RGS}}

36: \newcommand{\chandra}{{\it Chandra}}

37: \newcommand{\letg}{{\small LETG}}

38: \newcommand{\hetgs}{{\small HETGS}}

39: \newcommand{\heg}{{\small HEG}}

40: \newcommand{\meg}{{\small MEG}}

41: \newcommand{\aciss}{{\small ACIS-S}}

42: \newcommand{\ciao}{{\small CIAO}}

43: \newcommand{\cxc}{{\small CXC}}

44: \newcommand{\oiii}{O~[{\small III}]}

45: \newcommand{\uv}{{\small UV}}

46: \newcommand{\x}{X-ray}

47: \newcommand{\xs}{X-rays}

48: \newcommand{\cmsq}{cm$^{-2}$}

49: \newcommand{\kms}{km~s$^{-1}$}

50: \newcommand{\etc}{$\eta$~Car}

51: \newcommand{\msol}{$M_{\odot}$}

52: \newcommand{\msolyr}{$M_{\odot}$~yr$^{-1}$}

53: \newcommand{\etal}{et~al.}

54: 

55: \begin{document}

56: 

57: \title{A hot transient outflow from $\eta$ Carinae}

58: \author{Ehud Behar\altaffilmark{1}, Raanan Nordon\altaffilmark{1},

59: Eyal Ben-Bassat\altaffilmark{1}, and Noam Soker\altaffilmark{1}}

60: 

61: \altaffiltext{1}{Department of Physics,

62:                  Technion, Haifa 32000, Israel.

63:                  behar@physics.technion.ac.il (EB).}

64: %\received{} \revised{7/1/04} \accepted{}

65: 

66: \shorttitle{Outflow from $\eta$ Carinae} \shortauthors{Behar et

67: al.}

68: 

69: \begin{abstract}

70: $\eta$~Carinae (\etc) is a stellar binary system with a period of

71: 5.54 years.  It harbors one of the brightest and most massive

72: stars in our galaxy.  This paper presents spectroscopic evidence

73: for a fast (up to 2,000 \kms) X-ray outflow of ionized gas

74: launched from \etc\ just before what is believed to be the binary

75: periastron (point of smallest binary separation). The appearance

76: of this high-velocity component, just as the irregular flares in

77: the \x\ light curve, can not be explained by the simple continuous

78: binary wind interaction, adding to the intrigue of the \etc\

79: system.

80: %By analogy with

81: %jets in other astrophysical systems, it is possible to attribute

82: %the jet to the onset of a phase of accretion.

83: \end{abstract}

84: 

85: \keywords{techniques: spectroscopic --- X-rays: stars --- stars:

86: mass loss --- stars: individual: \etc inae}

87: 

88: \section{INTRODUCTION}

89: \label{sec:intro}

90: 

91: Despite being studied for many years and in all wavebands, the

92: $\eta$ Carinae system (\etc), best known for its spectacular

93: nebula \citep{morse98} that can be tracked back to the Great

94: Eruption of 1840 \citep{davidson97} is still much of a mystery. It

95: is now believed to be a stellar binary system with a period of

96: 5.54 years \citep{damineli96}. The primary star is one of the

97: brightest and most massive ($\sim$ 120~\msol) stars in our galaxy.

98: The system is currently in a phase of immense mass ejection. The

99: \x s detected from \etc\ can be explained by the collision of the

100: two stellar winds \citep{corcoran01, pittard02}, not so the 70 day

101: X-ray minimum observed periodically around periastron

102: \citep{corcoran05}.

103: 

104: Currently, the mass loss rate of the primary exceeds

105: $10^{-4}$\msolyr\ \citep{smith03} and that of the secondary is an

106: order of magnitude lower \citep{pittard02}, but still unusually

107: high. The wind of the secondary is much faster \citep[few 1,000

108: \kms,][]{pittard02} than that of the primary \citep[$\sim$~600

109: \kms\ and depending on latitude,][]{smith03}. The \x\ flux between

110: 2 and 10~keV during the past 12 years is normally at a level of

111: 5$\times 10^{-11}$ erg~s$^{-1}$~cm$^{-2}$, as expected from the

112: collision of the two stellar winds \citep{usov92, corcoran01,

113: pittard02}. The \x\ light curve is roughly constant throughout

114: most of the 5.54 year orbital period, but rises gradually by a

115: factor of 3--4 upon approach to periastron at which time short

116: flares appear \citep{corcoran97}, before it drops sharply and

117: stays in an \x\ low state ($<$~20\% of normal brightness) for

118: approximately 70~days. The full X-ray light curve can be found in

119: \citet{corcoran05}. Although absorption by the dense primary wind

120: may play a certain role in attenuating the X-ray emission from

121: \etc, both spectral and temporal arguments have been put forward

122: to reject absorption or an eclipse as the prime reason for this

123: persisting low state \citep{hamaguchi04, soker05, akashi06}, which

124: is unique to \etc. Recently, it has been proposed that if the

125: binary separation during periastron is small enough, the massive

126: primary wind could smother the secondary wind and thus shut down

127: the X-ray emission \citep{soker05, akashi06}. However, so far

128: there has been no direct evidence for accretion.

129: 

130: In this work, we wish to add to the temporal aspects of the \etc\

131: tale a high-resolution \x\ spectroscopic dimension. This is

132: achieved by using five deep exposures of \etc\ with the High

133: Energy Grating Spectrometer (\hetgs) on board the \chandra\ \x\

134: Observatory (CXO), only the first of which has been published

135: \citep{corcoran01,corcoran01b}. Theoretical variations of spectral

136: line profiles in binary colliding wind systems have been modeled

137: by \citet{henley03}. The varying line profiles presented here for

138: \etc, however, do not follow these models, at least not in a

139: straightforward way, as we explain in \S 3.

140: 

141: \section{Observations \& Results}

142: 

143: The log of the \chandra\ observations is given in Table~1.   Two

144: observations (3745 and 3748) coincide with the short intense

145: flares just before periastron.  The exact time of periastron is

146: not well determined and depends on the uncertain orbital

147: parameters of the system. Nonetheless, for the sake of

148: presentation, we assume throughout this paper a period of

149: 5.54~years (2024~days) and set phase zero to be at 2003-06-29. The

150: X-ray low state, thus, occurs during phase 0 -- 0.035. The data

151: were retrieved from the CXO archive and processed using standard

152: software (CIAO version 3.2.2.). \chandra 's HEG and MEG (high and

153: medium energy gratings) spectrometers operate simultaneously in

154: the 3.5 -- 7.5 \AA\ band of interest here and their spectra are

155: consistent with each other. Aiming at the highest possible

156: kinematic resolution, however, the higher spectral resolving power

157: of the HEG in this band (factor of $\sim$ 2) is favored over the

158: slightly higher effective area of the MEG (factor of $\sim$ 2),

159: especially since the co-added plus and minus 1st order HEG data

160: are already of sufficiently high signal-to-noise ratio (S/N).

161: 

162: To reveal the dynamics of the \x\ gas, we transferred each

163: spectral line profile to line-of-sight velocity space ($v_\|$ in

164: \kms) using the simple non-relativistic Doppler shift $v_\| =

165: c(\lambda - \lambda _0)/ \lambda _0$, where $\lambda$ and $\lambda

166: _0$ represent the observed and rest-frame wavelengths of the line,

167: respectively, and $c$ is the speed of light. Negative and positive

168: velocities represent approaching and receding gas, respectively.

169: Fig.~1 shows the broad and variable \x\ line profiles observed in

170: \etc\ as demonstrated on the Si$^{+13}$ Ly$\alpha$ unresolved

171: doublet at 6.18~\AA. As expected, the MEG profile represents a

172: smoothed version of the higher-resolution HEG profile. The

173: observed line is much broader than the instrumental line spread

174: function (also shown). Fig.~1 also demonstrates how the line

175: profile varies dramatically from assumed phase $\phi$=~--0.470

176: (Obs. 632) to $\phi$=~--0.028 (Obs. 3745). The early-phase

177: ($\phi$=~--0.470) line is centered around zero velocity with no

178: significant emission beyond 700~\kms\ (\fwhm\ = 1000

179: $\pm$~100~\kms). The line closer to periastron ($\phi$=~--0.028)

180: features high velocity gas up to $\sim$2000~\kms. A gaussian fit

181: yields a centroid shift of --550 $\pm$~90~\kms, although the

182: profile is clearly asymmetric.

183: 

184: %we show a sample spectrum (Obs. 3748 at assumed orbital phase

185: %--0.006) with the main lines identified to exemplify the

186: %high-quality line-resolved spectra. The steep decline of the

187: %continuum towards long wavelengths manifests the photoelectric

188: %absorption along the line of sight, most likely by the primary

189: %wind.

190: %The inset in Fig.~1 highlighting the

191: %Si lines demonstrates the abnormal line profiles and

192: %approaching-velocity Doppler shifts observed when the system is

193: %presumably close to periastron.

194: 

195: In each observation, the kinematic profiles of all lines are

196: fairly similar. This is demonstrated in Fig.~2, where we present

197: four bright spectral lines from Obs.~3745 ($\phi$= --0.028). The

198: profiles are unambiguously broad and asymmetric with approaching

199: velocities reaching 2,000~\kms, but with no significant receding

200: emission beyond $\sim$~700~\kms. In order to understand how the

201: line profiles vary between observations and as a function of phase

202: we co-added the profiles of nine bright lines in each observation,

203: namely the Ly$\alpha$ (1s-2p), He$\alpha$ resonance (1s$^2$-1s2p)

204: and He$\alpha$ forbidden (1s$^2$-1s2s) lines of the Si, S, and Ar

205: ions. Since the profiles of all these lines in a given observation

206: are roughly similar (e.g., Fig.~2), co-adding gives an average

207: profile with improved S/N. Individual Fe lines below 2~\AA\ could

208: not be included in this analysis, despite their high intensity,

209: since the \hetgs\ resolving power decreases strongly at these low

210: wavelengths to the extent that the He-like Fe line complex is

211: unresolved.  The blue wing of the He$\alpha$ Fe resonance line is

212: consistent with, though not resolved as well as, the other line

213: profiles presented here. The resulting mean profiles have been

214: normalized (except for Obs. 3747) to facilitate the comparison and

215: are presented in Fig.~3. The observed, mean, unnormalized peak

216: intensities for the assumed phases: --0.470, --0.130, --0.028,

217: --0.006, and +0.044 are, respectively, 2.0, 3.9, 3.9, 1.9, and

218: 0.3, $\times$~10$^{-3}$~ph~s$^{-1}$~cm$^{-2}$~\AA$^{-1}$. However,

219: any interpretation of the {\it absolute} line fluxes in terms of

220: the orbital parameters would be risky, as the observed variability

221: is a result of both abrupt flaring and varying absorption.

222: 

223: A clear trend can be seen in Fig.~3. Far from periastron, the line

224: profile is relatively symmetrical and narrow. Closer to

225: periastron, bright components of gas moving towards us at

226: velocities as high as --2,000~\kms\ start to develop. As the

227: system further approaches periastron, the outflow dominates the

228: line profile to the point where the bulk of the emission is

229: clearly blue-shifted. After periastron absorption considerably

230: attenuates the emission lines to the low level seen in Fig.~3.

231: Naive gaussian fits to the first two profiles yield centroids at

232: --25 $\pm$~25~\kms\ (for $\phi$=~--0.470) and at --50

233: $\pm$~20~\kms\ ($\phi$=~--0.130) both with \fwhm\ = 700

234: $\pm$~50~\kms. The late-phase asymmetric profiles are

235: non-gaussian. For the mere purpose of quantifying the centroid

236: shift, we performed two-gaussian fits, which yield peaks at --120

237: $\pm$~50~\kms\ and --800 $\pm$~100~\kms\ for $\phi$=~--0.028 and

238: at --170 $\pm$~50~\kms\ and --1000 $\pm$~100~\kms\ for

239: $\phi$=~--0.006. It is hard to rule out high-velocity {\it

240: receding} gas before the \x\ minimum, since in agreement with

241: \citet{hamaguchi04}, we find that the column density toward the

242: \x\ source rises from 5$\times$10$^{22}$~\cmsq\ long before

243: periastron to 3$\times$10$^{23}$~\cmsq\ just after it. Emission

244: from the far side of the system, thus, could be strongly absorbed.

245: 

246: %as the \x\ absorption is observed to increase

247: %significantly before the minimum.

248: %Measuring the absorbing column density, we find that it rises from

249: %5$\times$10$^{22}$~\cmsq\ before periastron to

250: %3$\times$10$^{23}$~\cmsq\ right after it, where the emission lines

251: %are almost completely quenched, and in agreement with previous

252: %reports \citep{hamaguchi04}.

253: 

254: \section{Interpretation \& Discussion}

255: 

256: The colliding wind binary model, which is generally accepted for

257: explaining the \x\ emission of \etc, has several clear predictions

258: regarding the spectral line profiles expected at different orbital

259: phases.  The Doppler shifts are the result of shocked gas flowing

260: along the contact discontinuity (CD) surface and away from the

261: stagnation point (SP). If the shock opening angle is between

262: $45^\circ\ - 90^\circ$ \citep{pittard02}, the broadest line

263: profiles are expected when the system is observed perpendicular to

264: the line connecting the two stars, as the hottest gas formed

265: around the SP flows directly towards and away from the observer.

266: In this case, the lines are centered at zero velocity and are

267: intrinsically symmetrical, but could be skewed by absorption of

268: the far (red) emission.  On the other hand, when the system is

269: observed from behind one of the stars, shifted centroids are

270: expected. If the system is observed from behind the weaker

271: (stronger) wind, downstream shocked gas is flowing along the CD

272: surface in the general direction of (opposite) the observer,

273: producing blue- (red-) shifted centroids. The maximum velocities

274: in this case would be less than those in the first (orthogonal)

275: orientation. For a complete and detailed calculation of line

276: profiles from colliding winds see Henley et al. (2003). In the

277: following, however, we argue that the current line profiles

278: observed at the different phases of the \etc\ orbit are in

279: contradiction, even qualitatively, with the simple colliding wind

280: scenario.

281: 

282: If the major axis of the \etc\ binary is perpendicular to our line

283: of sight \citep{smith04}, the broadest profiles are expected at

284: apastron (see above), but that is when the observed profiles are

285: narrowest ($\phi =-0.470$).  Also, with this geometry, the

286: secondary would have to pass in front of the primary before

287: periastron (in contrary to the picture of \citet{smith04}) to

288: explain the blue-shifted centroids at $\phi = -0.028, -0.006$, but

289: that would produce symmetrical (blue-shifted) profiles as

290: absorption through the secondary wind is weak, while strongly

291: asymmetrical profiles are observed.  In any case, the lower

292: velocities observed near apastron appear to rule out this

293: orientation if the colliding wind geometry is to produce the

294: observed \x\ line profiles.

295: 

296: If alternatively, the projection of the line of sight on the

297: orbital plane is more or less along the major axis and the viewing

298: angle at apastron is from behind the secondary \citep{corcoran01},

299: the observed blue-shifts at $\phi = -0.028, -0.006$ could be

300: ascribed to gas flowing from the SP towards the observer and the

301: unobserved red wing of the line to absorption by the primary wind.

302: However, according to this scenario, at apastron one would expect

303: blue-shifted centroids of at least a few 100~\kms\ due to

304: downstream gas (see above), which is not observed ($\phi =

305: -0.470$). For an observer above the binary plane ($i < 90^\circ$),

306: as seems to be the case in \etc\ \citep{corcoran01}, the problem

307: becomes more severe (higher blue-shift is expected, but not

308: observed).

309: 

310: In short, the observed behavior of the \x\ line profiles of \etc\

311: appear qualitatively inconsistent with a naive wind-collision

312: scenario, since the interpretation of the high blue-shifts from a

313: post-shock stream can not be made consistent with all phases.

314: \citet{henleyphd}, comparing with line profiles calculated from

315: hydrodynamical simulations, also reached the conclusion that the

316: observed profiles are not as expected from simple models of the

317: geometry of the wind-wind collision. He suggested that including

318: the Coriolis effect may bring the models to better agreement with

319: observations. \citet{kashi07} suggest the effect of orbital

320: motion, which can account for small centroid shifts, but can not

321: explain the currently observed \x\ profiles. Here, we wish to

322: propose an alternative explanation. In particular, we note that

323: the asymmetric line profiles are observed during strong peaks in

324: the X-ray light curve. These peaks too can not be accounted for by

325: the simple colliding wind model \citep{pittard02, henley03}, or by

326: the Coriolis force.

327: 

328: We therefore prefer to interpret the observed high-velocity gas up

329: to --2000~\kms\ in terms of a transient, collimated fast wind, or

330: a jet, ejected from the immediate vicinity of the binary system

331: through the interaction of the two stars. The peaks in the light

332: curve hint that the outflow may be in the form of blobs. The

333: unresolved \chandra\ images and the rapid variations both indicate

334: that the outflow is restricted to within approximately

335: 2$\times$10$^{16}$~cm from the center (assuming 0.5\arcsec\ at

336: \etc 's distance of 2.3~kpc). The appearance of the same charge

337: states in the spectra attained during all phases implies that the

338: temperatures of the X-ray gas and line-emitting outflow remain in

339: the range of $kT$~= 2~-- 5~keV. Consequently, we speculate that

340: the fast component of the outflow consists of gas shocked by the

341: collision of the winds (as observed throughout the orbit) that is

342: launched and collimated near periastron. The widths of the major

343: peaks seen in Figs.~2 and 3 suggest that the outflow is only

344: moderately collimated to within $\sim 30^\circ$. The condensation

345: of shocked gas provides a natural explanation for the rise in

346: X-ray intensity during the short intermittent flares observed in

347: the X-ray light curve. A collimated outflow from the secondary has

348: also been suggested to explain the enhanced He~II $\lambda 4686$

349: emission before periastron \citep{soker06}. Interestingly, both

350: proper motion and Doppler shifts have been measured for

351: high-velocity visible-light knots much further out from the center

352: \citep{walborn78, meaburn93}. The optical knots were ejected more

353: than a hundred years ago along the minor axis of the nebula

354: \citep{meaburn93}. We would expect the present outflow to be

355: launched along the major axis of the nebula although we have no

356: pertinent information on its present direction with respect to the

357: system's geometry. If indeed that is the case and if the

358: inclination angle of the binary plane to our line of sight is $i=

359: 42^\circ$ \citep{smith02}, the actual outflow velocity would be as

360: high as $\sim$2,700~\kms.

361: 

362: %X-ray jets have been observed in accreting stellar systems

363: %\citep{kastner04}, and also in X-ray binaries with a neutron star

364: %or black-hole companion (microquasars), but not from evolved hot

365: %stars. Although not well understood, it is known

366: %phenomenologically that jets are typically found in accreting

367: %systems.  The current discovery, therefore, makes for

368: %circumstantial evidence of accretion in the \etc\ system, and as

369: %such could validate the scenario proposed in \citet{soker05} and

370: %in \citet{akashi06}. Interestingly, the wind from the progenitor

371: %of the well studied supernova (SN) 1987A was strongly asymmetric

372: %\citep[e.g.,][]{allen89}. Furthermore, the galactic SN remnant

373: %Cassiopeia A has alpha-element-rich jets as well \citep{hwang04},

374: %which speculatively, could be the scaled up version of those we

375: %are seeing in \etc.

376: 

377: As the massive \etc\ primary is a short-lived star shedding

378: considerable mass and as it is now understood that core collapse

379: supernovae can not expel more than a few solar masses, implying

380: that higher mass stars must shed most of their mass prior to the

381: explosion, the winds and fast outflow of \etc\ may be a supernova

382: in the making. At the very least, it highlights the next

383: periastron passage on 2009, January 12 as a faithfully scheduled

384: experiment for the astrophysically common, but poorly understood

385: phenomenon of jet launching.

386: 

387: \acknowledgments This research was supported by grant \#28/03 from

388: the Israel Science Foundation, and by a grant from the Asher Space

389: Research Institute at the Technion.

390: %We thank the Hezbollah for

391: %leaving nothing to do in Haifa but science during the last month

392: %of this project.

393: 

394: \clearpage

395: 

396: \begin{thebibliography}{}

397: 

398: \bibitem[Akashi \etal(2006)]{akashi06} Aksahi, M., Soker, N. \&

399: Behar, E. 2006, \apj, 644, 451.

400: 

401: %\bibitem[Allen \etal(1989)]{allen89} Allen, D.~A., Meikle, W.~P.~S., \& Spyromilio, J.

402: %1989, Nature 342, 403.

403: %

404: \bibitem[Corcoran \etal(1997)]{corcoran97} Corcoran, M. F., \etal\ 1997, Nature 390, 587.

405: 

406: \bibitem[Corcoran \etal(2001a)]{corcoran01} Corcoran, M.~F.,

407: Ishibashi, K., Swank, J.~H. \& Petre, R. 2001a \apj, 547, 1034.

408: 

409: \bibitem[Corcoran \etal(2001b)]{corcoran01b} Corcoran, M. F., \etal\ 2001b, \apj, 562, 1031.

410: 

411: \bibitem[Corcoran(2005)]{corcoran05} Corcoran, M.~F. 2005, \aj, 129, 2018.

412: 

413: \bibitem[Davidson \& Humphreys(1997)]{davidson97} Davidson, K. \&

414: Humphreys, R.~M. 1997, Annu. Rev. Astron. Astrophys. 35, 1.

415: 

416: \bibitem[Damineli(1996)]{damineli96} Damineli, A. 1996, \apjl, 460,

417: L49.

418: 

419: \bibitem[Hamaguchi \etal(2004)]{hamaguchi04} Hamaguchi, K. \etal\ 2004 (astro-ph/0411271).

420: 

421: \bibitem[Henley \etal(2003)]{henley03} Henley, D.~B., Stevens,

422: I.~R., \& Pittard, J.~M. 2003, \mnras, 346, 773.

423: 

424: \bibitem[Henley(2005)]{henleyphd} Henley, D.~B., Ph.D. thesis,

425: http://www.physast.uga.edu/$\sim$dbh/thesis/

426: 

427: %\bibitem[Hwang \etal(2004)]{hwang04} Hwang, U. \etal\ 2004, \apj, 615, L117.

428: %

429: %\bibitem[Iping \etal(2005)]{iping05} Iping, R.,~C., Sonneborn, G.,

430: %Gull, T.~R., Massa, D.~L. \& Hillier, D.~J. 2005, \apj, 633, L37.

431: 

432: \bibitem[Kashi \& Soker(2007)]{kashi07} Kashi, A. \& Soker, N. 2007,

433: astro-ph/0702661

434: 

435: %\bibitem[Kastner \etal(2004)]{kastner04} Kastner, J.~H. \etal\ 2004, Nature 430,

436: %429.

437: %

438: \bibitem[Meaburn \etal(1993)]{meaburn93} Meaburn, J. \etal\ 1993, \aap, 276, L21.

439: 

440: \bibitem[Morse \etal(1998)]{morse98} Morse, J. A., \etal, 1998, \aj, 116,

441: 2443.

442: 

443: \bibitem[Pittard \& Corcoran(2002)]{pittard02} Pittard J.~M. \&

444: Corcoran, M.~F. 2002, \aap, 383, 636.

445: 

446: \bibitem[Smith(2002)]{smith02} Smith, N. 2002, MNRAS 337, 1252.

447: 

448: \bibitem[Smith \etal(2003)]{smith03} Smith, N., Davidson, K., Gull,

449: T.~R., Ishibashi, K. \& Hillier, D.~J. 2003, \apj, 586, 432.

450: 

451: \bibitem[Smith \etal(2004)]{smith04} Smith,~N., Morse,~J.~A., Collins,~N.R., \&

452: Gull,~T.~R. 2004, \apj, 610, L105.

453: 

454: \bibitem[Soker(2005)]{soker05} Soker, N. 2005, \apj, 635, 540.

455: 

456: \bibitem[Soker \& Behar(2006)]{soker06} Soker, N. \& Behar, E. 2006, \apj, 652, 1563.

457: 

458: \bibitem[Usov(1992)]{usov92} Usov, V.~V. 1992, \apj, 389, 635.

459: 

460: \bibitem[Walborn \etal(1978)]{walborn78} Walborn, N.~R., Blanco, B.~M. \& Thackeray, A.~D.

461: 1978, \apj, 219, 498.

462: 

463: \end{thebibliography}

464: 

465: \clearpage

466: 

467: \begin{deluxetable}{lcccc}

468: \tablecolumns{6} \tablewidth{0pt} \tablecaption{CXO/HETG

469: Observation Log \label{tab1}} \tablehead{

470:    \colhead{CXO archive} &

471:    \colhead{Start Time} &

472:    \colhead{Exposure} &

473:    \colhead{Assumed Orbital} &

474:    \colhead{2--10~keV Flux} \\

475:    \colhead{ID \tablenotemark{a}} &

476:    \colhead{} &

477:    \colhead{(ks)} &

478:    \colhead{Phase ($\phi$)\tablenotemark{b}} &

479:    \colhead{(10$^{-10}$ erg~s$^{-1}$~cm$^{-2}$)}}

480: \startdata

481: 632 & 2000-11-19 02:46:40 & 90.69 &  --0.470 & 0.50 \\

482: 3749 &  2002-10-16 08:08:49 & 93.96 &  --0.130 & 0.98 \\

483: 3745 &  2003-05-02 11:56:16 & 97.29 &  --0.028 & 2.2 \\

484: 3748 &  2003-06-16 05:35:28 & 100.1 &  --0.006 & 0.97 \\

485: 3747 &  2003-09-26 22:45:53 & 72.16 &  +0.044 & 0.48 \\

486: \enddata

487: \tablenotetext{a}{ One additional observation was carried out

488: during the X-ray minimum (ID 3746 at 2003-07-20 01:46:22) for

489: which no useful spectrum could be obtained.} \tablenotetext{b}{

490: Phase is approximated based on the assumption of a 5.54~year

491: (2024~day) orbit, where phase zero is set at 2003-06-29 and the

492: X-ray low state occurs during phase 0 -- 0.035.}

493: \end{deluxetable}

494: 

495: \clearpage

496: 

497: \begin{figure}

498:   % Requires \usepackage{graphicx}

499:   \centering

500:   \includegraphics[width=10.0cm, angle=0]{f1.ps}

501:   \bigskip

502:   \caption{Velocity profiles of the Si$^{+13}$ Ly$\alpha$ unresolved doublet

503: (6.180 and 6.186 \AA). {\it Upper panel} shows consistent HEG

504: (higher spectral resolution) and MEG (smoother) data from Obs.

505: 3745 ($\phi$=~--0.028). The profile is clearly broadened up to

506: 2000~\kms, much beyond the instrument line spread functions

507: (dashed lines). {\it Bottom panel} shows the HEG profile of Obs.

508: 632 ($\phi$=~--0.470, dashed line) and Obs. 3745

509: ($\phi$=~--0.028). The scaled up line from the early phase is

510: rather symmetrical and much narrower, showing no line emission

511: beyond 700~\kms.}

512:   \label{f1}

513: \end{figure}

514: 

515: %\begin{figure}

516: %\centerline{\includegraphics[width=8.5cm,angle=-90]{spectrum.ps}}

517: % \caption{\x\ spectrum of \etc\ obtained during Obs.~3748 (assumed phase --0.006)

518: %  by co-adding plus and minus 1st orders of the HEG spectrometer.

519: %  The brightest H-like (Ly$\alpha$) and He-like (He$\alpha$, resonance and forbidden) observed lines are indicated at their rest frame.

520: %  The strong Fe line complex emerges below 2~\AA.  In the inset, a blow-up of the Si lines is given, where the broadening

521: %  of the lines towards shorter wavelengths can be seen.

522: %  }

523: %\label{f1}

524: %\end{figure}

525: %

526: 

527: \clearpage

528: 

529: \begin{figure}

530: \centerline{\includegraphics[width=10cm,angle=0]{f2.ps}}

531: 

532:   \caption{Profiles of four bright spectral lines from Obs.~3745 (assumed phase --0.028).

533:   The consistent asymmetric profiles with blueshifts up to $\sim\ -2,000$~\kms\

534:   are clearly seen.}

535:    \label{f2}

536: \end{figure}

537: 

538: \clearpage

539: 

540: %\begin{figure}

541: %  % Requires \usepackage{graphicx}

542: %  \centering

543: %  \includegraphics[width=13.0cm, angle=-90]{fig3.ps}

544: %  \bigskip

545: %  \caption{Mean spectral line profiles in velocity space for the five CXO/HETG observations of \etc.

546: %  See Table~1 for details. Errors are the mean 1$\sigma$.  The profiles demonstrate a slight shift of --250~\kms\ and the

547: %  conspicuous rise of emission from hot, fast outflowing gas at velocities as high as --2,000~\kms\

548: %  in the two pre-minimum observations.  In the post-minimum spectrum, the emission lines are strongly absorbed.}

549: %  \label{f3}

550: %\end{figure}

551: 

552: \begin{figure}

553:   % Requires \usepackage{graphicx}

554:   \centering

555:   \includegraphics[width=10.0cm, angle=-90]{f3.ps}

556:   \bigskip

557:   \caption{Mean, normalized, velocity profiles constructed from nine different

558:   spectral lines for the five observations of \etc.

559:   Only the profile of Obs. 3747 (assumed phase +0.044) during which \etc\ was strongly

560:   absorbed is not normalized and plotted at its correct scale with respect to Obs. 632

561:  (assumed phase --0.470). Typical 1$\sigma$ errors on each data point are 10 -- 15~\%

562:  (top right hand side).

563: %  The profiles demonstrate a slight centroid shift of --250~\kms\ and the

564: %  conspicuous rise of a fast outflowing emission component at velocities as high as

565: %  --2,000~\kms\ in the two pre-minimum observations.

566: }

567:   \label{f3}

568: \end{figure}

569: 

570: \end{document}

571: