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

2: 

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

4: \newcommand{\asca}{{\it ASCA}}

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

6: \newcommand{\ee}[1]{\times10^{#1}}

7: \newcommand{\dem}{\mathcal{D}}

8: \newcommand{\mdot}{\dot{M}}

9: \newcommand{\vinf}{v_\infty}

10: \newcommand{\nH}{n_{\rm H}}

11: \newcommand{\mump}{\mu m_p}

12: \newcommand{\kms}{km\,s$^{-1}$}

13: \newcommand{\ergs}{erg\,s$^{-1}$}

14: %\newcommand{\bfx}{{\bf x}}

15: \newcommand{\nuth}{\nu_{\rm th}}

16: \newcommand{\stdplot}[1]{\includegraphics[angle=-90,width=3.5in]{#1}}

17: \newcommand{\tripspec}[1]{\includegraphics[angle=-90,width=3.0in]{#1}}

18: \newcommand{\nulll}[1]{}

19: \newcommand{\tnm}[1]{\tablenotemark{#1}}

20: \newcommand{\ud}[2]{$^{+#1}_{-#2}$}

21: 

22: \slugcomment{{\Large DRAFT: \today}}

23: 

24: \begin{document}

25: 

26: \title{Constraints on the Velocity and Spatial Distribution of Helium-like

27: Ions in the Wind of SMC X-1 from Observations with XMM-Newton/RGS}

28: 

29: %\title{Observations of the High-Mass X-ray Binary SMC~X-1 with \xmm{}

30: %RGS} 

31: 

32: \author{Patrick S. Wojdowski\altaffilmark{1}}

33: 

34: \affil{Kavli Institute for Astrophysics and Space Research,

35:   Massachusetts Institute of Technology} 

36: 

37: \email{pswoj@space.mit.edu}

38: 

39: \author{Duane A. Liedahl}

40: 

41: \affil{Department of Physics and Advanced Technologies, Lawrence

42: Livermore National Laboratory}

43: 

44: \author{Timothy R. Kallman}

45: 

46: \affil{NASA Goddard Space Flight Center}

47: 

48: \altaffiltext{1}{current address: Aret\'e Associates, P.O. Box 6024, Sherman Oaks, CA 91413} 

49: 

50: \begin{abstract}

51: We present here X-ray spectra of the HMXB SMC~X-1 obtained in an

52: observation with the \xmm{} observatory beginning before eclipse and

53: ending near the end of eclipse.  With the Reflection Grating

54: Spectrometers (RGS) on board \xmm{}, we observe emission lines from

55: hydrogen-like and helium-like ions of nitrogen, oxygen, neon,

56: magnesium, and silicon.  Though the resolution of the RGS is

57: sufficient to resolve the helium-like $n$=2$\to$1 emission into three

58: line components, only one of these components, the intercombination

59: line, is detected in our data.  The lack of flux in the forbidden

60: lines of the helium-like triplets is explained by pumping by

61: ultraviolet photons from the B0 star and, from this, we set an upper

62: limit on the distance of the emitting ions from the star.  The lack of

63: observable flux in the resonance lines of the helium-like triplets

64: indicate a lack of enhancement due to resonance line scattering and,

65: from this, we derive a new observational constraint on the

66: distribution of the wind in SMC~X-1 in velocity and coordinate space.

67: We find that the solid angle subtended by the volume containing the

68: helium-like ions at the neutron star multiplied by the velocity

69: dispersion of the helium-like ions must be less than

70: 4$\pi$\,steradians\,\kms{}.  This constraint will be satisfied if the

71: helium-like ions are located primarily in clumps distributed

72: throughout the wind or in a thin layer along the surface of the B0

73: star.

74: 

75: \end{abstract}

76: 

77: \section{Introduction}

78: 

79: In isolated early-type stars winds are driven as ultraviolet photons

80: from the stellar surface impart their outward momentum to the wind in

81: resonance line transitions.  In a high-mass X-ray binary (HMXB), the

82: wind is ionized by X-rays from the compact object.  If the X-radiation

83: is intense enough, the resulting ions will not have transitions in the

84: ultraviolet and this greatly affects the dynamics of the wind

85: \citep[see, e.g., ][ and references therein]{blo94}.  The

86: SMC~X-1/Sk~160 system, which consists of a 0.71 second X-ray pulsar

87: and a B0I star together in a 3.9-day orbit, is the most X-ray luminous

88: known HMXB and, therefore, presumably, an extreme example of wind

89: disruption by X-ray ionization.

90: 

91: The behavior of the wind of an early-type star under the influence of

92: ionization from an X-ray emitting companion has been the subject of

93: many theoretical studies.  For SMC~X-1, the most relevant and detailed

94: such study is, arguably, the one by \citet{blo95} which included a

95: numerical hydrodynamic simulation.  The features that appeared in that

96: simulation included a regular, UV-driven wind on the X-ray shadowed

97: side of the star, a thermal wind on the X-ray illuminated side,

98: transition regions between the two types of winds and dense,

99: finger-like structures in the equatorial plane.  However, even this

100: simulation includes significant approximations: the gravity of the

101: compact object is not included and X-ray photoionization and its

102: dynamical effects are treated in a very approximate way.  Because of

103: the complexity of these systems and the difficulty of accounting for

104: all of the relevant physics, observations are critical to

105: characterizing the behavior of HMXB winds.

106: 

107: %Formerly 1st part of 3rd paragraph

108: Though previous X-ray observations of SMC~X-1 have had spectral

109: resolving powers ($\equiv\lambda/\Delta\lambda=E/\Delta{}E$) of

110: $\sim$50 or less, X-ray spectroscopy has already revealed much about

111: the wind.  X-ray spectroscopic observations of this resolving power

112: can, through measurements of line fluxes, indicate the quantity and

113: ionization level of X-ray emitting material.  In observations of SMC

114: X-1 with \asca{}, emission lines were not detected and upper limits

115: were set on the quantity of material of intermediate ionization level

116: in the wind of SMC~X-1, excluding the presence of finger-like

117: structures as they appeared in the simulation of \citeauthor{blo95}

118: \citep{woj00}.  Further observations of SMC~X-1 with the Advanced

119: Camera for Imaging Spectroscopy (ACIS) on-board the {\it Chandra X-ray

120: Observatory}, however, have detected emission lines and, therefore,

121: the presence of material of intermediate ionization in the wind of

122: SMC~X-1 \citep{vrt01}.  These results indicate that overdense regions

123: do exist in the wind of SMC~X-1 and that while the simulations of

124: \citeauthor{blo95} may not be accurate in detail, the general form of

125: the structure predicted by \citeauthor{blo95} may, in fact, be

126: present.  To develop further constraints on models of the wind in

127: SMC~X-1, it is desirable to have constraints in addition to the

128: quantity and ionization level of the X-ray emitting plasma.

129:    

130: %Formerly 2nd part of 3rd paragraph

131: High-resolution spectroscopic observations with resolving powers

132: $\gtrsim$200 have the potential to provide useful constraints on the

133: kinematics of the high-ionization wind in HMXBs through direct

134: measurements of Doppler line shifts and broadening.  In addition, the

135: $n=2\to1$ triplets of He-like ions are resolved at high resolution and

136: measurements of the fluxes of the individual lines of these triplets

137: provide a constraint on the structure and kinematics of the X-ray

138: emitting material.  Of the three components of the helium-like

139: triplet, only one --- the resonance line --- has a large oscillator

140: strength and can be enhanced by resonant line scattering.  Because

141: resonant line scattering saturates and because this saturation depends

142: on the distribution of the scatterers in physical and velocity space,

143: constraints on the distribution of the wind in physical and velocity

144: space can be derived from the relative fluxes of these three lines

145: \citep{woj03}.  The flux ratios of the helium-like triplets are also

146: affected by ultraviolet radiation \citep*{blu72} and, therefore, the

147: flux ratios measured from an HMXB constrain the distance from the

148: photosphere of the high-mass star to the emission region.

149: 

150: Because of the location of SMC X-1 outside of the plane of the Galaxy,

151: the column density of interstellar material to it is small and, unlike

152: the HMXBs in the Galaxy, it can be observed in the wavelength range

153: 15--35\,\AA.  The $n=2\to1$ transitions of the hydrogen-like and

154: helium-like ions of oxygen and nitrogen as well as hydrogen-like

155: carbon lie in this range.  Therefore, unlike with the Galactic HMXBs,

156: it is possible to derive observational constraints on the density and

157: velocity distributions of the regions of moderate ionization in

158: SMC~X-1 where these ions exist.  Because of the high spectral

159: resolution (FWHM of $\sim$65\,m\AA, corresponding to resolving powers

160: in the range 230--540) and large effective area (total of

161: $\sim$80\,cm$^2$) of the Reflection Grating Spectrometer on \xmm{} in

162: this band, the RGS on \xmm{} is particularly well-suited to

163: spectroscopic study there.  We present here observations of SMC~X-1

164: with \xmm{} beginning before an eclipse and ending near the end of

165: that eclipse.  We present an analysis in which we focus on the data

166: from the RGS and derive constraints on the space and velocity

167: distribution of material in the wind.  In \S\ref{sec:empir} we

168: describe our observations and an empirical analysis of the line

169: emission in which we measure line fluxes, shifts, and widths.  In

170: \S\ref{sec:phys} we derive constraints on the distribution of the wind

171: in physical and velocity space from the line fluxes we measure.  In

172: \S\ref{sec:discuss}, we discuss the implications of our results for

173: models of HMXB winds.

174: 

175: \section{Observations \& Empirical Analysis}

176: \label{sec:empir}

177: 

178: SMC~X-1 was observed from on 2001-05-31 from approximately 02:14 to

179: 18:50 UT with \xmm.  The actual times at which the first and last

180: observational data were received with the various instruments differ

181: from those times by as much as an hour.  However, data were collected

182: with each instrument 

183: during at least 90\% of the time between the nominal beginning and end

184: of the observations.  In Figure~\ref{fig:m1_lc} we show plots of count

185: rates from the source region and a background region on the MOS1

186: detector.

187: \begin{figure}

188: \includegraphics[angle=-90,width=5in]{f1.eps}

189: \caption{Count rates from the MOS1 detector.  We show the count rate

190:   from a source region in black and from a background in red.  An

191:   increase at late time in both count rates is indicative of an

192:   increase in the non-X-ray background.  Each data point represents

193:   the average count rate over 128\,s.  Time zero on the abscissa is

194:   2001-05-31 02:22:16.9357 UT.}

195: \label{fig:m1_lc}

196: \end{figure}

197: In the beginning, the count rate in the source region decreases

198: through X-ray eclipse ingress.  Toward the end of the observation,

199: the count rates for both the source and the background regions

200: increase while the difference stays approximately constant indicating

201: an increase in the non-X-ray background.  We divided the data into the

202: three time intervals noted on the figure.  On the time scale of the

203: figure, the times which divide the intervals are 10609\,s and

204: 42289\,s.  The intervals, labeled ``ingress'', ``center'', and

205: ``end'' were chosen to divide the eclipse ingress from the time of

206: complete X-ray eclipse and the time of high background from the rest

207: of the observation.  We then extracted spectra from each of the five

208: \xmm{} X-ray instruments for each of the three intervals and from the

209: interval consisting of the center and end intervals which we label

210: ``eclipse''.

211: 

212: We focus here on the spectral data from the RGS.  However, in

213: Figure~\ref{fig:m1_ecl_spec} we plot the spectrum from MOS 1 for the eclipse

214: interval.

215: \begin{figure}

216: \stdplot{f2a.eps}

217: \stdplot{f2b.eps}

218: \caption{MOS 1 Eclipse spectrum of SMC X-1.  Labels indicate the

219: energies of spectral features which may or may not be detected in this

220: spectrum.  The labels denote the energies of lines except where the

221: ion is followed by ``e'' indicating an edge.  The line energies

222: indicated are $n=2\to1$ transitions except for \ion{0}{8} at

223: 0.77\,keV, \ion{Ne}{10} at 1.21\,keV, and \ion{Mg}{12} at 1.74\,keV

224: which are $n=3\to1$ transitions.}

225: \label{fig:m1_ecl_spec}

226: \end{figure}

227: This spectrum is similar to those spectra of SMC~X-1 obtained by

228: \citet{vrt01} with \chandra/ACIS. In Figure~\ref{fig:rgs_ecl} we

229: show our RGS spectrum for the eclipse interval.

230: \begin{figure}

231: %\stdplot{ecl_05_12_rgs_spec.ps}

232: %\stdplot{ecl_12_20_rgs_spec.ps}

233: %\stdplot{ecl_20_28_rgs_spec.ps}

234: %\stdplot{ecl_28_38_rgs_spec.ps}

235: \noindent

236: \includegraphics[angle=-90,width=6.5in]{f3a.eps}

237: \includegraphics[angle=-90,width=6.5in]{f3b.eps}

238: \caption{The RGS eclipse spectrum of SMC~X-1.}

239: \label{fig:rgs_ecl}

240: \end{figure}

241: Line emission from hydrogen-like and helium-like ions of several

242: elements is apparent.  The spectral resolution of the \xmm{} RGS is

243: such that the $n=2\to1$ transitions of helium-like ions are resolved

244: into a triplets consisting of, in order of wavelength, the resonance

245: ($1s2p\,^1\!P_1\to$ground), intercombination

246: ($1s2p\,^3\!P_{2,1}\to$ground), and forbidden

247: ($1s2s\,^3\!S_1\to$ground) lines.  However, in our RGS spectra, only

248: the intercombination lines are apparent.  Before attempting to explain

249: this fact or to use it to derive constraints on the properties of the

250: SMC~X-1 system, we first make a quantitative characterization of the

251: line emission.

252: 

253: In order to measure the fluxes, widths and shifts of these lines, we

254: fit the wavelength channels within 0.5\,\AA{} of the rest

255: wavelengths of each of the $n=2\to1$ emission line complexes.  We fit

256: the spectrum in the vicinity of each line complex using a power law

257: continuum and Gaussian lines.  We express the width (sigma) and

258: centroid of the Gaussians in velocity coordinates and fix the width

259: and shifts of the lines in a single complex to be the same, i.e.

260: \footnote{This awkward model for the continuum flux is due to the fact

261: that the power law spectral model is defined in energy space

262: ($F_E\equiv{}K(E/{\rm keV})^\alpha$).  Our fits to the continuum

263: flux are not a focus of this work nor do we make any use of them

264: here so we proceed with this continuum model.}  :

265: \begin{equation}

266: F_\lambda=K(hc)^{\alpha+1}({\rm keV})^\alpha{}\lambda^{-(\alpha+2)}

267: +\sum_k\frac{1}{\lambda_k(v_r/c)\sqrt{\pi}} I_k 

268: \exp\left(\frac{(\lambda-\lambda_k(1+v_r/c))^2}{2\lambda_k^2(\sigma_{v,r}/c)^2}\right) 

269: \label{eqn:line_model}

270: \end{equation}

271: where $F_\lambda$ is the photon (not energy) flux, the $\lambda_k$s

272: are the rest wavelengths of the lines, and $K$, $\alpha$, $v_r$,

273: $\sigma_{v,r}$, and the $I_k$s are fit parameters.  We assume that

274: lines appear in emission and therefore constrain the fluxes of all

275: lines to be greater than or equal to zero.

276: 

277: %a model consisting of a power-law continuum and four Gaussian

278: %emission features.  We fix the centers of the Gaussians to be the

279: %nominal line wavelengths of the four lines times a factor $1+v_r/c$

280: %where $v_r$, the radial velocity, is a variable parameter and $c$ is

281: %the speed of light.  We fix the $\sigma$ parameters of the four

282: %Gaussians to be a single variable value $\sigma_{v,r}$.

283: 

284: The intercombination line is actually an unresolved doublet.

285: Therefore, we fit the ingress and eclipse spectra in the vicinity of

286: the helium-like triplets using the model described by

287: equation~\ref{eqn:line_model} with four Gaussians: one for each of the

288: resonance and forbidden lines and one for each of the two components of

289: the intercombination line.  However, we fix the ratio of the fluxes of

290: the two components of the intercombination line to be equal to that

291: predicted by \citet{kin03} for recombination using temperatures given

292: in Table~\ref{tab:parconst}.  Furthermore, for reasons we describe

293: later, we formulate our model not in terms of the individual fluxes of

294: the three lines of the triplets but in terms of the total triplet line flux $I$ and

295: the ratios of the line fluxes $G$ and $R$ where

296: \begin{equation}

297: G\equiv\frac{I_{\rm i}+I_{\rm f}}{I_{\rm r}}

298: \end{equation}

299: and

300: \begin{equation}

301: R\equiv\frac{I_{\rm f}}{I_{\rm i}}

302: \end{equation}

303: where the subscripts ``r'', ``i'', and ``f'' indicate, respectively,

304: the resonance, intercombination, and forbidden lines.  In the fits, we

305: restrict the value of $\sigma_{v,r}$ to be less than 1500\,\kms{},

306: $v_r$ to be between $-$1000 and 1000\,\kms{} and we restrict $I$, $G$,

307: $R$, and $K$ to be greater

308: than or equal to zero.  In Figures~\ref{fig:trips_ing} and

309: \ref{fig:trips_ecl},

310: \begin{figure}

311: \tripspec{f4a.eps}

312: \tripspec{f4b.eps}

313: \tripspec{f4c.eps}

314: \tripspec{f4d.eps}

315: \tripspec{f4e.eps}

316: \caption{Helium-like triplets for the ``ingress'' interval.}

317: \label{fig:trips_ing}

318: \end{figure}

319: \begin{figure}

320: \tripspec{f5a.eps}

321: \tripspec{f5b.eps}

322: \tripspec{f5c.eps}

323: \tripspec{f5d.eps}

324: \tripspec{f5e.eps}

325: \caption{Helium-like triplets for the ``eclipse'' interval.}

326: \label{fig:trips_ecl}

327: \end{figure}

328: we show our best fits to these triplets and in

329: Table~\ref{tab:he_lines} we show the best-fit values and confidence

330: intervals for these line fluxes, shifts and widths.  The fact that we

331: detect only the intercombination lines of the helium-like triplets is

332: reflected in the fact that we obtain only lower limits on $G$ and only

333: upper limits on $R$.

334: 

335: \begin{deluxetable}{lccccccc}

336: \tablewidth{0pt}

337: \tablecaption{Measured Helium-like Triplet Parameters} 

338: \tablehead{ & & & & & & \multicolumn{2}{c}{power-law} \\

339: \cline{7-8}

340: Element & $I$\tnm{a} & $G$ & $R$ &

341: $\sigma_{v,r}$(\kms) & $v_{\rm r}$(\kms) & K & $\alpha$ } 

342: \startdata

343: \cutinhead{ingress}

344: N & 1.2\ud{0.7}{0.6} & $>$4 & \nodata & $<$800 & $-$300\ud{400}{300} & 

345: 36\ud{10}{36} & 0\ud{\infty}{2} \\

346: O & 2.6\ud{1.4}{1.5} & $>$3& $<$0.3& 500\ud{600}{500} & $-$500\ud{600}{\infty}

347: & 135\ud{14}{111} & 0\ud{\infty}{3} \\

348: Ne & 1.4\ud{4.2}{1.4} & \nodata & \nodata & \nodata & \nodata & 170$\pm$30 &

349: $-$0.9\ud{\infty}{2.0} \\

350: Mg & 2\ud{6}{2}& \nodata& \nodata& \nodata& \nodata& 205\ud{16}{50} & $-$1.1\ud{1.1}{0.3} \\

351: Si & 22\ud{17}{16} & \nodata & \nodata & \nodata & \nodata & 280\ud{90}{70} & $-1.3\pm0.5$\\

352: \cutinhead{eclipse}

353: N & 0.22$\pm$0.18 & $>$1.6& $<$0.63& $<$500& $-$400\ud{500}{200} &

354: $<$1.3& \nodata \\

355: O & 1.1$\pm$0.4 & 18\ud{\infty}{14} & 0.26\ud{0.34}{0.26} & $<$300 &

356: 90\ud{130}{270} & $<$0.0002 & 21.5\ud{0.6}{13.6} \\

357: Ne & 1.8\ud{0.7}{0.8} & $>$9& 0.2\ud{0.3}{0.2} & $<$600& $-$400\ud{500}{300}

358: & 6.5$\pm$0.7 & -6\ud{5}{\infty} \\

359: Mg & 2.2\ud{0.9}{0.7} & $>$2& $<$0.25& $<$1300& $-$600\ud{600}{\infty}

360: & 3.8\ud{1863.5}{0.6} & 0\ud{\infty}{3} \\

361: Si & 2.4\ud{2.4}{2.0} & \nodata & \nodata & \nodata & \nodata & 2.5\ud{1863.9}{0.8} 

362: & 0\ud{\infty}{9} \\

363: \enddata

364: \label{tab:he_lines}

365: \tablecomments{As a confidence limit on $\sigma_{v,r}$, $v_{\rm r}$, or

366: $\alpha$, $\infty$ indicates that parameter boundaries are within

367: the confidence limits.  The entry ``\nodata'' indicates that the data do

368: not constrain this quantity.}

369: \tablenotetext{a}{Total triplet line flux in units of $10^{-6}\,$photons$\,$cm$^{-2}$s$^{-1}$. Not corrected for interstellar photoelectric absorption.}

370: %\tablenotetext{b}{Number of photons s$^{-1}$cm$^{-2}$keV$^{-1}$ is

371: %$KE^{-\alpha}$ where $E$ is energy.} 

372: %% \tablenotetext{b}{I will probably not include the power-law fit data

373: %% when I submit the paper}

374: \end{deluxetable}

375: 

376: %%%%Fitting H-lines

377: The hydrogen-like $n=2\to1$ transitions consist only of an unresolved

378: doublet and we fit these doublets using two Gaussians with the ratio of

379: the fluxes of those two lines fixed at 2:1, proportional to the

380: statistical weights of the upper levels and 

381: the ratio expected for recombination.

382: Furthermore, we restrict $\sigma_v$ to be less than 2500\,\kms{} and

383: $v_r$ to be between $-$2500 and 2500\,\kms.  In

384: Table~\ref{tab:h_lines}, we show the fit parameters for the

385: hydrogen-like lines.  Again, $I$ indicates the total photon flux for

386: the doublet.

387: 

388: \begin{deluxetable}{lccccccc}

389: \tablewidth{0pt}

390: \tablecaption{Measured Hydrogen-like Line parameters}

391: \tablehead{ & & & & \multicolumn{2}{c}{power-law} \\

392: \cline{5-6}

393:  $Z$ & $I$($10^{-6}$s$^{-1}$cm$^{-1}$) & $\sigma_{v,r}$(\kms) & 

394: $v_{\rm r}$(\kms) & norm & $\alpha$ }

395: \startdata

396: \cutinhead{ingress}

397: 6 & 0.5\ud{0.6}{0.5} & \nodata & \nodata & $<$1900 &

398: \nodata \\ 

399: 7 & 1.0\ud{0.7}{0.8} & $<$1200 & 600\ud{500}{1800} & 174\ud{17}{159} &

400: 0\ud{\infty}{3} \\ 

401: 8 & 4\ud{3}{2} & 900\ud{1100}{900} & $-$2300\ud{900}{\infty} &

402: 16\ud{26}{13}& $-$6\ud{4}{\infty} \\ 

403: 10 & 7$\pm$5 & 2500\ud{\infty}{2000} & \nodata& 171\ud{14}{13}

404: & 0.0\ud{\infty}{1.8} \\

405: 12 & 1.4\ud{9.4}{1.4}& \nodata& \nodata& 170$\pm$30&

406: $-$0.4\ud{\infty}{0.3}  \\

407: 14 & 44\ud{16}{22}& \nodata& 2000\ud{\infty}{4000}&

408: 29000\ud{42000}{17000}& $-$8.6\ud{1.4}{1.2} \\ 

409: \cutinhead{eclipse}

410: 6 & 0.3$\pm$0.3 & 1300\ud{\infty}{900} & $-$1500\ud{1900}{\infty}& $<$8

411: & \nodata \\  

412: 7 & 0.7$\pm$0.2& $<$400& 0$\pm$200& 9$\pm$3& \nodata \\

413: 8 & 1.7$\pm$0.3 & $<$400 & 160\ud{70}{140} & 8.4$\pm$1.9 &

414: 0\ud{\infty}{3} \\ 

415: 10 & 1.0$\pm$0.6& \nodata& 200\ud{900}{1100}& 7.5$\pm$1.4 & 0\ud{\infty}{5}

416: \\

417: 12 & 0.9\ud{1.0}{0.9}& \nodata & \nodata & 17\ud{1892}{4}& \nodata \\ 

418: 14 & $<$2.0 & \nodata & \nodata & 3.3\ud{2632.6}{1.6} & \nodata \\ 

419: \enddata

420: \tablecomments{The entry ``\nodata'' indicates that the data do not

421: constrain this quantity.}

422: \label{tab:h_lines}

423: \end{deluxetable}

424: 

425: 

426: \section{Line Fluxes as Measures of the Distribution of Scatters in

427:   Physical and Velocity Space}

428: \label{sec:phys}

429: 

430: The emission lines from high-mass X-ray binaries result from

431: recombination and resonant line scattering in wind of the high-mass

432: star.  The line emission from these processes depends on the

433: distribution of the wind material in coordinate space and, in the case

434: of resonant line scattering, in velocity space.  Therefore, the line

435: fluxes we measure provide a constraint on these distributions of the

436: wind material.  However, the equations that describe these

437: distributions are complex and rather than attempt to derive

438: constraints directly from the line fluxes, we believe it will be more

439: illuminating to create a simple model described by a few parameters

440: and then derive constraints on these parameters.  In \S\ref{sec:rnr},

441: we derive equations for line luminosities due to recombination and

442: resonant line scattering.  In \S\ref{sec:helike}, we describe a

443: modification to those luminosities for the case of the helium-like

444: triplets due to ultraviolet radiation of the high-mass star.  In

445: \S\ref{sec:models} we describe the simple models and our constraints

446: on their parameters.

447: 

448: \subsection{Recombination and Resonant Line Scattering}

449: \label{sec:rnr}

450: 

451: In a plasma in photoionization equilibrium, photons are absorbed from

452: the continuum in photoionizing transitions and line photons are emitted

453: as electrons and ions recombine.  If the plasma is optically thin in

454: the photoionizing continuum, the recombination line luminosity may be

455: expressed as

456: \begin{equation}

457: L_{{\rm rec}} =  h\nu_0\frac{\pi{}e^2}{m_ec}

458: (4\pi)^{-1}\int\!\!\!\int \eta n_{{\rm Ion}}ds\,d\Omega

459: \int_{\nuth}^\infty\frac{L_\nu}{4\pi{}h\nu}\frac{df}{d\nu}d\nu 

460: \label{eqn:lrec}

461: \end{equation}

462: where $h$ is Planck's constant, $\nu_0$ is the frequency of the line,

463: $e$ and $m_e$ are the electronic charge and mass, $n_{{\rm Ion}}$ is

464: the density of the ion that emits the line, $\eta$ is fraction of

465: recombinations to the ion which result in emission of the line,

466: $L_\nu$ is the specific luminosity of the compact X-ray source,

467: $\nuth$ is the ionization threshold frequency of the emitting ion in

468: its ground state, and $df/d\nu$ is its continuum oscillator strength,

469: and $s$ and $\Omega$ are the distance from and solid angle subtended

470: at the compact X-ray source.

471: 

472: In any plasma, line photons are absorbed and reemitted in resonance

473: line transitions.  The net effect of this is only to redirect line

474: photons.  However, depending the geometry of the plasma and radiation

475: source, resonant line scattering may result in absorption or emission

476: lines being observed.  For the geometry of interest in this work --- an

477: occulted compact radiation source with radiation scattered into the

478: observer's line of sight by plasma --- resonant scattering results in

479: emission lines and the line luminosity due to scattering may be

480: expressed as

481: \begin{equation}

482: L_{{\rm scat}} = L_{\nu_0}(4\pi)^{-1}\int\!\!\!\int

483: (1-e^{-\tau(\nu,{\bf \Omega})})d\nu{}d\Omega 

484: \label{eqn:lscat}

485: \end{equation}

486: where $\tau$ is the line optical depth through the plasma along a line

487: from the compact radiation source.  The line optical depth is given by

488: \begin{equation}

489: \tau=f\frac{\pi{}e^2}{m_ec}\int{}n_{\rm Ion}\phi(\nu)ds

490: \end{equation}

491: where $f$ is the line oscillator strength and the function

492: $\phi$ describes the line profile.  Neglecting natural line

493: broadening and considering only Doppler line broadening, we have

494: \begin{equation}

495: \phi(\nu)\equiv\frac{c}{\nu_0}g(c(\nu/\nu_0-1))

496: \end{equation}

497: where $g$ is the local distribution of the ions' velocity component in

498: the direction along the line of sight from the compact X-ray source

499: and is normalized such that

500: \begin{equation}

501: \int_{-\infty}^{\infty}g(v)dv\equiv1. 

502: \end{equation}

503: However, we assume that ion velocities are very small compared to $c$

504: so that $g(v)$ is non-zero only for $|v|\ll{}c$.  With this definition,

505: \begin{equation}

506: \int_0^{\infty}\phi(\nu)d\nu=1.

507: \end{equation}

508: 

509: As we will ultimately be interested in calculating observed line

510: fluxes the spatial integrals in equations~\ref{eqn:lrec} and

511: \ref{eqn:lscat} should be restricted to the volume that is both

512: illuminated by the compact radiation source and observable (i.e., not

513: occulted by the star).  In principle, the radiation from the compact

514: source that reaches the observable plasma may be depleted at the line

515: frequency by line scattering in the occulted region, reducing the

516: observable line luminosity relative to that described by

517: equation~\ref{eqn:lscat}.  However, as scattering does not destroy

518: photons but merely redirects them, for most plasma geometries, the

519: total number of line photons reaching the observable region will not

520: be significantly affected by scattering in the occulted region.  Only

521: for a specialized plasma geometry: a long, narrow distribution of

522: plasma, oriented along a line of sight from the compact radiation

523: source will equation~\ref{eqn:lscat} be significantly inaccurate.

524: Furthermore, if the part of the wind where the line emission occurs is

525: moving supersonically (as HMXB winds are known to do) the line

526: frequencies in different parts of the wind will be shifted apart so

527: that resonant line scattering in one part of the wind will not be

528: affected by resonant line scattering in another part.

529: 

530: In equation~\ref{eqn:lrec}, we have $\eta$ inside

531: the spatial integral, allowing for the possibility that the

532: recombination efficiencies may depend on the local conditions.  In

533: fact, for helium-like lines, the forbidden line

534: may be ``pumped'' into the intercombination line at high densities or

535: high ultraviolet radiation intensities and we define $\eta$ so that

536: the effects of this pumping are absorbed into it.  Near a hot star such as

537: Sk~160, the B0 companion of SMC~X-1, the ultraviolet radiation

538: intensity is high.  Therefore, the line luminosities due to resonant

539: scattering depend on the distance from the star.  In the next section,

540: we describe this in more detail.

541: 

542: 

543: \subsection{Helium-like Triplets}

544: \label{sec:helike}

545: As can be seen from the equations above, the observed flux of any one

546: line depends in a complicated way on the distribution of the emitting

547: ions in both coordinate and velocity space.  Therefore, the constraint

548: that can be derived from a single line is quite complex.  However,

549: because the resonance line of each helium-like triplet has a

550: significant oscillator strength while the other two lines do not,

551: only the resonance line is affected by resonant scattering.

552: Therefore, only the flux of the resonance line depends on the velocity

553: distribution of the emitting ions and by measuring the flux of all

554: three lines of the triplet, it is possible to derive more specific

555: constraints on the distribution of the emitting ions in space and

556: velocity than can be obtained from the flux of any one line.

557: 

558: %Because of the different characteristics of the lines of the

559: %helium-like triplets, their observed fluxes and relative fluxes

560: %provide useful constraints on the distribution of the emitting

561: %helium-like ions and on the ambient conditions in the region

562: %containing them.

563: 

564: % In order of wavelength, these three lines are the forbidden,

565: % intercombination, and recombination lines.  The intercombination

566: % line is actually a doublet consisting of the transitions from the

567: % upper levels $^3P_{2}$ and $^3P_{1}$.  The resonance and forbidden

568: % lines are produced by transitions from, respectively, the $^1P_1$

569: % and $^3S_1$ states.  Following the notation of \citet{por01} we will

570: % sometimes refer to the levels $^3P_{k}$ as $p_k$, to the $^3S_1$

571: % level as $m$ for ``metastable'', and to the ground level as $g$.

572: 

573: As previously mentioned, the recombination efficiencies of the

574: intercombination and forbidden lines are not constant but depend on

575: the local conditions.  This is due to the fact that the upper level of

576: the forbidden line, the $^3S_1$ level, is metastable and an ion in

577: that state may be excited by absorption of an ultraviolet photon or

578: collision with an electron to any of the $^3P$ levels.  Two of these

579: levels ($^3P_2$ and $^3P_1$) make up the upper levels of

580: the intercombination line.  Therefore, at high densities or high

581: ultraviolet radiation intensities, the effective recombination

582: efficiency of the intercombination line is increased and the effective

583: recombination efficiency of the forbidden line is decreased.  This

584: effect has been used by several authors to estimate the distance of

585: emitting ions from isolated hot stars.  However, several of the

586: calculations have neglected the fact that once an ion is excited from the

587: $^3S_1$ state to a $^3P$ state, it may decay back to the $^3S_1$ state by

588: emission of an ultraviolet photon rather than decay to ground by

589: emission of an intercombination line photon.  We write the

590: recombination line efficiencies including this effect as 

591: \begin{eqnarray}

592: \eta_{\rm i}&=&\eta_{\rm i}^\prime+\eta_{\rm f}^\prime(1-b_{\rm f}) \\

593: \eta_{\rm f}&=&\eta_{\rm f}^\prime{}b_{\rm f}

594: \end{eqnarray}

595: where $\eta_{\rm i}^\prime$ and $\eta_{\rm f}^\prime$ are the

596: recombination efficiencies without $^3S_1\to{}^3P$ excitation and

597: $b_{\rm f}$ is the branching ratio for radiative decay of the $^3S_1$

598: state to ground.  In order to derive an expression for this branching

599: ratio, we solve the rate equations of \citet[][ equations 18 and

600: 19]{mew78} and neglect inner-shell ionization, all collisions, and

601: induced radiative decay.  To write our expression, we use the notation

602: of \citet{por01} and denote the $^3S_1$ level by $m$ (for

603: ``metastable''), the ground ($1s^2\,^1S_0$) by $g$ and denote the

604: level $^3P_k$ by $p_k$ where $k$ may be 0, 1, or 2 and get

605: \begin{equation}

606: b_{\rm f}\equiv\frac{A_{mg}}{A_{mg}+T}

607: \end{equation}

608: where 

609: \begin{equation}

610: T\equiv\sum_{k=0}^2w_{mp_k}\frac{A_{p_kg}}{A_{p_kg}+A_{p_km}}

611: \end{equation}

612: where $w_{mp_k}$ is the excitation rate from the $^3S_1$ to

613: $^3P_k$\footnote{Our expression for the forbidden line efficiency is

614: consistent with expressions for forbidden line intensities given by

615: \citet{mew78} and by \citet{por01}.  However, our expression for the

616: intercombination line is not consistent with those works.  As they are

617: written, it is difficult to compare our line efficiencies with those

618: line intensities.  Therefore, we note that our solution of the rate

619: equations implies

620: $I_i=\sum_{k=1}^2C_{gp_k}(1-BR_{p_km}BR_{mg})+C_{gm}(1-BR_{mg})$ for

621: equation 5 of \citet{por01}.  We are encouraged in using our

622: expression for the intercombination line because, with our expression,

623: the sum of the intercombination and forbidden lines, as expected, does

624: not depend on the magnitude of the $^3S_1\to^3P$ excitation.  In any

625: case, the final results for both of those works are based not on those

626: expressions but on solutions of the full rate matrix and would not be

627: affected by any error in the expressions.}.

628: 

629: For photoexcitation, the $w$ values are given by

630: \begin{equation}

631: w_{mp_k}=\frac{4\pi{}^2e^2}{m_ec}f_{mp_k}\frac{J_{\nu_{mp_k}}}{h\nu_{mp_k}}

632: \end{equation}

633: where $J_{\nu_{mp_k}}$ is the mean intensity of radiation (i.e., the

634: intensity averaged over solid angle) at the frequency $\nu_{mp_k}$ and

635: $\nu_{mp_k}$ and $f_{mp_k}$ are, respectively, the frequency and

636: oscillator strength of the transition $m\to{}p_k$.  The frequencies

637: $\nu_{mp_k}$ for elements from carbon to iron range from the near to

638: the extreme ultraviolet.  Because the quantity $\eta_{\rm i}+\eta_{\rm

639:   f}$ and the value of $\eta_{\rm r}$ do not depend on the ultraviolet

640: pumping, neither does the value of the $G$ ratio but the value of the

641: $R$ ratio does.

642: 

643: \subsection{Simple Model}

644: \label{sec:models}

645: From the observed fluxes of the three lines of a given helium-like

646: triplet, it is possible to derive constraints on the distribution of

647: the helium-like ions in coordinate and velocity space.  However,

648: because of the complexity of the equations describing the line

649: luminosities, the constraints that can be derived are also complex.

650: Therefore, we compute line fluxes for two simple parameterized models

651: for the distribution of material.  Then we derive constraints on those

652: parameters from the observed line fluxes.  We do not actually expect

653: the distribution of material in the wind of SMC~X-1 to be described by

654: these simple models in detail.  However, from these model parameter

655: constraints it is possible to make approximate inferences about the

656: distribution of the wind of SMC~X-1 in coordinate and velocity space.

657: 

658: In our models, each ion exists exclusively within a solid angle

659: $\Omega$ subtended at the neutron star, and is distributed such that

660: the column density along lines of sight from the neutron star has the

661: single value $N$ within the solid angle $\Omega$.  The distribution of

662: ion velocities along lines of sight from the neutron star in our model

663: is uniform over some range with width $\Delta{}v$ (which need not be

664: centered on zero).  Also, to simplify the calculation of ultraviolet

665: pumping, we take the entire plasma to be at a single distance $r$ from

666: the surface of the companion star which we take to be a sphere

667: emitting as a blackbody with a temperature of $T_\star=$30,000\,K, the

668: approximate effective temperature for a star of type B0 such as

669: Sk~160.

670: 

671: For this model, the UV radiation density is

672: \begin{equation}

673: J_{\nu}=W(r)B_\nu(T_\star)

674: \end{equation}

675: where $B_\nu$ is the Planck function,

676: \begin{equation}

677: W(r)\equiv\frac{1}{2}

678: \left[1-\left(1-\left(\frac{r_\star}{r}\right)^2\right)^{1/2}\right],

679: \end{equation}

680: and $r_\star$ is the stellar radius.  The luminosity of a line due to

681: recombination is then

682: \begin{equation}

683: L_{{\rm rec}}=h\nu_0\frac{\pi{}e^2}{mc}

684: (4\pi)^{-1}\eta\Omega{}N 

685: \int_{\nuth}^\infty\frac{L_\nu}{4\pi{}h\nu}\frac{df}{d\nu}d\nu 

686: \end{equation}

687: and the line luminosity due to scattering is

688: \begin{equation}

689: L_{{\rm scat}}=L_{\nu_0}(4\pi)^{-1}\frac{\nu_0}{c}\Omega\Delta{}v

690: (1-e^{-f\frac{\pi{}e^2}{m\nu_0}\frac{N}{\Delta{}v}}).

691: \label{eqn:lscat_model}

692: \end{equation}

693: or, equivalently,

694: \begin{equation}

695: L_{{\rm scat}}=L_{\nu_0}(4\pi)^{-1}\frac{\nu_0}{c}\Omega{}N

696: \frac{1-e^{-f\frac{\pi{}e^2}{m\nu_0}\frac{N}{\Delta{}v}}}{N/\Delta{}v}.

697: \label{eqn:lscat_model2}

698: \end{equation}

699: 

700: It can be seen from the above expressions that it is not necessary to

701: specify all of the model parameters in order to determine the line

702: luminosities.  It is sufficient to specify only the parameter

703: combinations $\Omega{}N$ and $N/\Delta{}v$ rather than the values of

704: the three parameters contained in those two combinations.  If we replace the

705: three parameters with the two parameter combinations then the fluxes

706: of the lines depend linearly on $\Omega{}N$, the sum of the

707: luminosities of the intercombination and forbidden lines depends only

708: on $\Omega{}N$, the $G$ ratio depends only on $N/\Delta{}v$, and the

709: $R$ ratio depends only on $r$.

710: 

711: In principle, it should be possible to derive constraints on the model

712: parameters from the constraints on the line fluxes in

713: Table~\ref{tab:he_lines}.  However, for the line emission mechanisms

714: we consider here, the $G$ and $R$ ratios can only take on values

715: within a finite range (e.g., see \citealt{woj03} Figure~8 for the

716: range of the $G$ ratio for \ion{Si}{13}).  In at least one case, our

717: best fit value of $G$ falls outside of this range.  Therefore, in

718: order to derive meaningful constraints on the physical model

719: parameters, we redo the fits with the $G$ and $R$ values constrained

720: to be within the range allowed by scattering and recombination as

721: described above.  In all cases, we are able to obtain good fits with

722: these constraints imposed.

723: 

724: In order to relate our model parameters to line fluxes and derive

725: constraints on our model parameters, we use the following data.  We

726: get recombination efficiencies (without $^3S_1\to^3P$ pumping) by dividing

727: effective line recombination rates from \citet{kin03} by total

728: recombination rates from \citet{ver96c}.  We have written equations

729: for line luminosities in terms of the specific luminosity of the X-ray

730: source.  Of course, we measure fluxes and any luminosities we derive

731: are subject to uncertainty in the distance to the object.  However,

732: if we write equations~\ref{eqn:lrec} and \ref{eqn:lscat} in terms of

733: fluxes, factors of distance cancel.  Therefore, in our calculations we

734: use fluxes and our results do not depend on the distance to SMC~X-1.

735: For the specific luminosity of the SMC~X-1 neutron star $L_\nu$ we use

736: \begin{equation}

737: {\cal F}=K\left(\frac{\epsilon}{1\,{\rm keV}}\right)^{-\alpha}

738: \end{equation}

739: where ${\cal F}$ is the specific photon flux per unit photon energy

740: interval which is related to $L_\nu$ by

741: $L_\nu=(4\pi{}d^2)^{-1}h^2\nu{\cal F}$ and $\epsilon$ is photon energy with

742: $K=0.0267$\,photons\,cm$^{-2}$\,keV$^{-1}$ and $\alpha=1$ from a fit

743: to the uneclipsed {\it ASCA} spectrum of SMC~X-1 \citep{woj00}.  The

744: value of 1 we use for $\alpha$ is an approximation to the best-fit

745: value of 0.94$\pm$0.02.  

746: %Multiplying ${\cal F}$ by $h^2\nu$ gives the specific energy flux per

747: %frequency.  

748: For the continuum oscillator

749: strength of the helium-like ions above the ionization threshold we use

750: \begin{equation}

751: \frac{df}{d\nu}=2\nuth^2\nu^{-3}

752: \end{equation}

753: \citep[c.f.,][]{woj03}.  For all elements, we take the oscillator

754: strength of the resonance line to be 0.7, which is accurate to 5\%.

755: For line frequencies (wavelengths) and spontaneous decay rates we use

756: values from the Astrophysical Plasma Emission Database

757: \citep[APED,][]{smi01}.  We take the oscillator strengths $f_{mp_0}$ to

758: be zero and derive the oscillator strengths $f_{mp_1}$ and $f_{mp_2}$

759: from the spontaneous deexcitation rates of APED.  We give the results

760: of our model parameter constraints in Table~\ref{tab:parconst}.

761: 

762: \begin{deluxetable}{lccccccccccc}

763: \tablewidth{0pt}

764: \tablecaption{Simple Model Parameters Derived from Helium-like Line Emission}

765: %\rotate

766: \tabletypesize{\scriptsize}

767: \tablehead{element & $T_{\rm rec}$(eV) & $\Delta\tau$ &

768: $r/r_\star$\tnm{a} &

769: $N/\Delta{}v$ & $\Omega(4\pi)^{-1}N$ & $\Omega(4\pi)^{-1}\Delta{}v$ &

770: $T_{\rm rec}$ \\

771:  & & &  &

772: ($10^{15}$\,cm$^{-2}$(\kms)$^{-1}$) &

773: ($10^{15}$\,cm$^{-2}$) & (\kms) & reference}

774: \startdata

775: N  & 3.0 & $>$4.2  & $<$57         & $>1.0$  &

776: 0.22$\pm$0.18    & $<$0.39 & 1 \\

777: O  & 4.0 & $>$20   & 18\ud{11}{17} & $>6.4$  & 1.3$\pm$0.4      & $<$0.27 & 1 \\

778: Ne & 12  & $>$22   & $<$6.6        & $>11$   &

779: 2.2\ud{0.9}{1.0} & $<$0.27 & 2 \\ 

780: Mg & 21  & $>$4.2  & $<$1.4        & $>3.2$  &

781: 3.4\ud{1.4}{1.1} & $<$1.5  & 2 \\

782: Si & 75  & \nodata    & \nodata    & 5.1\ud{5.1}{4.2} & 

783: \nodata{} & \nodata & 2 \\

784: \enddata

785: \tablenotetext{a}{$r$ is the distance from center of the star.}

786: \label{tab:parconst}

787: \tablerefs{(1)~\citealt{kin03}; (2)~\citealt{sak99}}

788: \end{deluxetable}

789: 

790: \section{Summary/Discussion}

791: \label{sec:discuss}

792: 

793: We have observed SMC~X-1 in eclipse with \xmm{} and, with the RGS,

794: resolved the helium-like triplets of nitrogen, oxygen, neon, and

795: magnesium.  To our knowledge, this is the first time helium-like

796: triplets from SMC~X-1 have been resolved and the first time the

797: helium-like triplets of nitrogen, oxygen, or neon have been resolved

798: for any HMXB.  In all cases, only one component of the triplet, the

799: intercombination line, is observably present.  The lack of observable

800: forbidden line fluxes in the triplet is easily explained by

801: photoexcitation pumping of the forbidden line into the

802: intercombination line by the ultraviolet radiation of the B0 star

803: Sk~160 and allows upper limits to be set on distance of the emitting

804: helium-like ions from Sk~160.  The absence of observable fluxes in the

805: resonance lines is consistent with what we expect from recombination

806: in the photoionized wind.  However, the fact that the resonance lines

807: are not enhanced by resonant scattering implies a lower limit on the

808: optical depth of the wind in the resonance line and, therefore,

809: constraints on the structure and kinematics of the wind are also

810: implied.

811: 

812: We have modeled the helium-like line emission as recombination and

813: scattering emission from a region that is partly obscured by the

814: companion star and is described by a single solid angle ($\Omega$) and

815: column density ($N$) and the velocities along lines of sight from

816: the neutron star have a boxcar distribution with width $\Delta{}v$ for

817: each of the ions.  We have inferred

818: infer the quantity $\Omega\Delta{}N$ from the sum of the flux of the

819: intercombination and forbidden lines which are not affected by

820: resonant line scattering.  We infer the quantity

821: $N/\Delta{}v$ from flux of the resonance line relative to

822: the sum of the other two (the inverse of the $G$ ratio).  From the

823: constraints on these two quantities, we also obtain a constraint on

824: the quantity $\Omega\Delta{}v$.  Because we do not detect the

825: resonance line, we obtain only lower limits on $N/\Delta{}v$

826: and only upper limits on $\Omega{}\Delta{}v$.  

827: 

828: As previously mentioned, in our analysis we have assumed that photons

829: escape isotropically along lines from the location of their production

830: or first scattering.  While we do expect recombination emission to be

831: isotropic, resonant line scattering is not isotropic.  The angular

832: distribution for resonant line scattering is a linear combination of

833: an isotropic distribution and an dipole distribution \citep[see,

834: e.g.,][Ch.\ 19]{cha60}.  Furthermore, for large optical depths that

835: are comparable to or greater than unity such as we infer, photons may

836: undergo several scatterings before escaping.  In this case, photons

837: will escape preferentially along directions where the optical depth is

838: least.  In general, the calculation of photon escape direction is also

839: complicated by the fact that photons may travel large distances across

840: the plasma before escaping.  While it is difficult to include this

841: possibility in analytic calculations, it is straightforward using

842: Monte Carlo techniques.  However, if the region that emits the

843: helium-like line emission has supersonic velocity differentials, as we

844: know that the bulk of the wind does, the transfer of resonant line

845: photons is essentially local (the Sobolev approximation,

846: \citealt{cas70}) and can be approached analytically.  For a

847: supersonically expanding wind, the optical depth in a direction given

848: by the coordinate $x$ is inversely proportional to $dv_x/dx$.  If the

849: flow moves along lines from a single point and $y$ is the distance

850: from that point, then these derivatives, in directions parallel and

851: perpendicular to the flow, respectively, are $dv/dy$ and $v/y$.  While

852: these quantities will not generally be equal, they will generally be

853: of the same order over the bulk of the wind and, therefore, if we

854: assume complete redistribution in frequency and direction for

855: individual scatterings, then the radiation will escape isotropically

856: \citep[see][]{cas70}.  In fact, individual scattering events do not

857: redistribute frequency and direction completely and so even with

858: $v/s=dv/ds$, radiation does not escape isotropically.  However the

859: difference from the case of complete redistribution is no greater than

860: 10\%--20\% (\citealt{car72}, see also \citealt{mih80}).  In light of

861: these facts, we proceed to explore the implications of our results.

862: However, it must be kept in mind that the approximations we have made

863: for the radiative line transfer may quantitatively affect our

864: conclusions.

865: 

866: The implication of our results on fundamental wind parameters, such as

867: the mass-loss rate and the terminal wind velocity, is complex owing to

868: factors such as the complex dependence of the ion fractions on density

869: in photoionization balance.  Therefore, interpreting these results in

870: terms of fundamental wind parameters is difficult to do without

871: computing line fluxes for complete wind models which is beyond the

872: scope of this work.  However, the upper limits on

873: $(\Omega/4\pi)\Delta{}v$ that we obtain, approximately 1\,\kms, are

874: quite small compared to 1000\,\kms{}, the order of the terminal wind

875: velocities of massive stars and the wind velocity in a simulation of

876: the wind of SMC~X-1 by \citet{blo95}.  Therefore, the part of the wind

877: that emits the helium-like lines cannot be homogeneously distributed

878: around the entire wind.  Instead, we believe that our results indicate

879: either that the volume (or volumes) containing the helium-like ions of

880: the various elements subtend a small solid angle at the neutron star

881: or that the helium-like ions exist primarily in a region where the

882: wind is at a small fraction of its terminal velocity or, perhaps, some

883: combination of both.

884: 

885: One possible explanation of our results is that the helium-like ions

886: exist primarily in dense clumps throughout the wind.  Dense clumps in the

887: wind of an HMXB have previously been invoked in the case of Vela~X-1

888: by \citet{sak99} in order to explain the strong fluorescence lines

889: observed from that system.  Another possible explanation is that

890: helium-like ions exist mainly near the surface of the star where the

891: density is higher, the ionization less, and the velocity lower than in

892: the outer parts of the wind \citep[see, e.g.,][]{lie01}.  Furthermore,

893: the part of a volume along the surface of the companion star with a

894: thickness significantly less than one stellar radius that is visible

895: during eclipse would subtend a solid angle significantly less than

896: $4\pi$ from the neutron star.  In Figure~\ref{fig:clump_face} we

897: illustrate both of these possibilities.

898: \begin{figure}

899: \begin{center}

900: \includegraphics[angle=-90,width=2.4in]{f6a.eps}\hspace{0.5in}

901: \includegraphics[angle=-90,bb=-83 -27 295 263,width=2.4in]{f6b.eps}

902: \end{center}

903: \caption{In this figure we illustrate to possible distributions of the

904: helium-like ions consistent with the helium-like line fluxes we

905: measure.  In the first panel we illustrate the possibility that the

906: helium-like ions are in clumps distributed throughout the wind.  In

907: the second panel we illustrate the possibility that the helium-like

908: ions are near the surface of the companion star.}

909: \label{fig:clump_face}

910: \end{figure}

911: Definitive tests of these hypotheses would require calculations of the

912: line emission from detailed wind models and that is beyond the scope

913: of this work.  

914: However, we expect that our results, including the

915: remarkably small values of $\Omega\Delta{}v$ are consistent with

916: current expectations about the nature of winds in HMXBs.

917: 

918: \bibliographystyle{apj} 

919: \bibliography{hmxb}

920: 

921: \end{document}

922: