1: \documentclass{emulateapj}

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

3: %\usepackage{epsfig}

4: %\usepackage{lscape}

5: %\usepackage{gensymb}

6: 

7: \usepackage{apjfonts}

8: 

9: %%\setlength{\textwidth}{8.in}

10: 

11: 

12: 

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

14: \def\gsim{\lower 2pt \hbox{$\, \buildrel {\scriptstyle >}\over

15: {\scriptstyle \sim}\,$}}

16: \def\lsim{\lower 2pt \hbox{$\, \buildrel {\scriptstyle <}\over

17: {\scriptstyle \sim}\,$}}

18: \def\rosat{{\sl ROSAT}}

19: \def\ciao{{\sl CIAO~}}

20: \def\ROSAT{{\sl ROSAT}}

21: \def\fuse{{\sl FUSE}}

22: \def\asca{{\sl ASCA}}

23: \def\ASCA{{\sl ASCA}}

24: \def\chandra{{\sl Chandra~}}

25: \def\oix{O~{\scriptsize IX}}

26: \def\oviii{O~{\scriptsize VIII}}

27: \def\ovii{O~{\scriptsize VII}}

28: \def\ovi{O~{\scriptsize VI}}

29: \def\oi{O~{\scriptsize I}}

30: \def\oii{O~{\scriptsize II}}

31: \def\oiv{O~{\scriptsize IV}}

32: \def\oiii{O~{\scriptsize III}}

33: \def\neviii{Ne~{\scriptsize VIII}}

34: \def\neix{Ne~{\scriptsize IX}}

35: \def\nex{Ne~{\scriptsize X}}

36: \def\HI{H~{\scriptsize I}\ }

37: \def\hii{H~{\scriptsize II}\ }

38: \def\aliii{Al~{\scriptsize III} }

39: \def\alii{Al~{\scriptsize II} }

40: \def\feii{Fe~{\scriptsize II}}

41: \def\fexvi{Fe~{\scriptsize XVI}}

42: \def\fexvii{Fe~{\scriptsize XVII}}

43: \def\fexviii{Fe~{\scriptsize XVIII}}

44: \def\source{4U~1820--303}

45: \newcommand{\as}{$^{\prime\prime}~$}

46: \newcommand{\am}{$^{\prime}~$}

47: 

48: \shortauthors{Yao \etal}

49: \shorttitle{\uppercase{Iron abundance in the hot ISM}}

50: 

51: \begin{document}

52: \slugcomment{The Astrophysical Journal, 653, L000-L000, 2006 December 20}

53: 

54: \title{{\sl Chandra} Detection of \fexvii\ in Absorption: Iron Abundance in the Hot Gaseous Interstellar Medium}

55: 

56: \author{Yangsen Yao\altaffilmark{1}, 

57:         Norbert Schulz\altaffilmark{1},

58:         Q. Daniel Wang\altaffilmark{2},

59:         and Michael Nowak\altaffilmark{1}}

60: \altaffiltext{1}{Massachusetts Institute of Technology (MIT) Kavli Institute for Astrophysics and Space Research, 70 Vassar Street, Cambridge, MA 02139; yaoys@space.mit.edu}

61: \altaffiltext{2}{Department of Astronomy, University of Massachusetts, 

62:   Amherst, MA 01003}

63: 

64: \begin{abstract}

65: 

66: The iron depletion level and the gas-phase iron abundance in the hot 

67: ($\sim10^6$ K) interstellar medium (ISM) are critical to our understanding 

68: of its energy balance as well as the thermal sputtering, 

69: cooling, and heating processes of dust grains. 

70: Here we report on the first detection 

71: of the \fexvii\ absorption line at 

72: $15.02$ \AA\ from the hot ISM in the spectrum of the low mass X-ray binary 

73: 4U~1820--303 observed with the {\sl Chandra X-Ray Observatory}. 

74: By jointly analyzing 

75: this absorption line with those from \ovii, \oviii, 

76: and \neix\ ions in the same spectrum, we obtain an

77: abundance ratio as Fe/Ne=0.8(0.4, 2.1) in units of the Anders \& Grevesse

78: solar value (90\% confidence intervals).

79: We find that the result is robust with respect to 

80: different assumed gas temperature distributions. The obtained Fe/Ne abundance

81: ratio, albeit with large uncertainties, 

82: is consistent with the solar value, indicating that there is very little or no

83: iron depleted into dust grains, i.e., most of or all of the dust grains

84: have been destroyed in the hot ISM.

85: 

86: \end{abstract}

87: \keywords{ISM: abundances --- ISM: dust --- 

88: X-rays: ISM --- X-rays: individual (\source)}

89: 

90: \section{Introduction \label{sec:intro}}

91: 

92: As one of the most abundant refractory metals, iron is an important  

93: constituent of the interstellar dust grains and is believed to have the 

94: greatest fraction of its atoms depleted into dust grains \citep{sof94}. 

95: Surveys in ultraviolet (UV) 

96: wavelength band indicate that $\gsim70\%$ of the iron in the cool and warm

97: medium of the Galactic disk/halo could be locked up in solid grains, and 

98: only the remains can be probed through absorption lines of gaseous 

99: \feii\ (see Savage \& Sembach 1996 and references

100: therein). Recent studies on X-ray  

101: absorption edges also provide evidence of iron depletion into dust grains

102: in the interstellar medium (ISM; Juett \etal 2006). 

103: 

104: Iron can be liberated from the dust to the gas phase through thermal 

105: sputtering. Theoretical calculations show that the sputtering caused by high 

106: velocity shocks ($v_s\simeq$ 50-200 ${\rm km\ s^{-1}}$) could destruct several 

107: $\times10\%$ of grains in the ISM, and that with the presence of a magnetic 

108: field, grain-grain collisions can also be a very efficient process for dust 

109: grain destruction, specifically in the case of low-velocity shocks 

110: ($v_s\lsim100\ {\rm km\ s^{-1}}$; 

111: Draine \& Salpeter 1979; Jones \etal 1994). %\citep{dra79, jon94}). 

112: The dust grains can also be fragmented/destroyed through thermal evaporating

113: in the hot ISM.

114: The recycling of atoms back to the gas phase has been evidenced by

115: the difference of the elemental abundance ratios in the Galactic halo from

116: those in the Galactic disk, and by the ionization disparity of the shocked

117: material at various post-shock distance \citep{jen95, sav96}. 

118: A comparison between the observed

119: abundance ratio of (Mg+Fe)/Si in the dust of the halo clouds and that of the 

120: theoretical expectation, together with the observed correlation between the 

121: gas-to-dust ratio and the dust mass carried by Fe, indicates that the dust 

122: grain cores likely contain iron oxides and/or metallic iron. Some 

123: of these resilient cores can survive from the

124: shock destruction during the dust processing in the ISM \citep{sem96, fri03}.

125: 

126: In the hot ISM, the knowledge of the exact amount of iron contained in dust

127: grains, which is reflected in the gas-phase abundance of 

128: iron (e.g., Savage \& Sembach 1996), is of great importance for 

129: understanding many astrophysical processes. 

130: For an emission spectrum of a solar abundance plasma at 

131: temperature $\sim10^6$ K, the contributions of iron ions

132: encompass about 50\% of the total emitting energy and photons

133: in the energy range from 10 eV to 2 keV. %without iron: 0.759/1.733e-9

134: %and produce about 50\% of the total emitting X-ray photons. % 145.9/233.9

135: On the other hand, dust grains are an effective coolant of the hot gas.

136: The thermal energy of the hot gas can be transferred to dust grains via the

137: collisions between electrons and dust grains, resulting in bulk heating 

138: of the dust and infrared (IR) emission. For a solar abundance of the dust 

139: grains, the grain IR radiation is at least comparable to, and could be more 

140: than 10 times more efficient than, the gas X-ray emission in cooling the 

141: hot gas at 

142: temperature $\sim10^6$ K \citep{dwek92}. Consequently, the existence of the

143: iron-bearing dust in the hot ISM could alter the chemical composition of the

144: hot gas, therefore, largely change the flux and the 

145: spectral shape of the hot gas radiation, affect the energy balance, 

146: change the cooling rate thus the lifetime of the hot gas, and eventually, 

147: adjust the star formation rate, in the whole ISM and affect the galactic

148: evolution in general.

149: 

150: The iron depletion level in the hot gas can be measured by modeling the iron 

151: emission lines or the absorption lines that the gas-phase iron ions imprint on 

152: the background point source spectrum, then comparing the iron abundance

153: with the solar values. Directly measuring the IR emission of the 

154: dust in such environments is difficult due to the confusion with  

155: foreground cold dust in the Galactic disk along the line of sight. 

156: A recent study of the very soft (0.25 keV) X-ray diffuse background emission 

157: in the vicinity of the Sun suggests that the gas phase iron is $\sim30\%$ 

158: of the solar value \citep{san01}. However, given that the Sun may reside in 

159: a privileged location (i.e., the Local Hot Bubble), this 

160: depletion level may not be typical for the general hot ISM. Furthermore,

161: the inferred emission measure is also sensitive to different adopted 

162: plasma emission models, and the less well known interaction between the solar 

163: wind ions with the local neutral ISM may further complicate the 

164: interpretations (e.g., Sanders \etal 2001;

165: Pepino \etal 2004; Hurwitz \etal 2005).

166: Absorption line studies, on the other hand, measuring the total column

167: density along the line of sight, are less affected by the different models and

168: the charge exchanges between the solar wind and the local neutral gas, and, 

169: therefore, should provide a reliable measurement of the iron abundance in 

170: the hot ISM.

171: 

172: In this Letter, we present the first detection of the \fexvii\ absorption line

173: at $\sim 15$ \AA\ from the hot ISM toward \source. 

174: Comparing the column density of \fexvii\ with that of \neix, and with those of 

175: \ovii\ and \oviii\ that we previously measured, we have derived a

176: relative abundance ratio of Fe/Ne and then inferred the iron depletion level

177: in the hot gas.

178: Throughout the Letter, we adopt the

179: solar abundances from \citet{and89}, 

180: \footnote{The ``real'' solar abundances are still under debate

181: (e.g., Asplund \etal 2005; Antia \& Basu 2006), so we still use

182: the old values.}

183: and quote the errors at 90\% confidence levels for single varying 

184: parameters until otherwise specified.

185: 

186: \begin{figure}

187: \centerline{

188: \includegraphics[width=0.47\textwidth]{f1.eps}}

189: \caption{Ionization fractions of oxygen, neon, and iron ions as a 

190:    function of

191:    temperature for a gas in the collisional ionization equilibrium 

192:    state \citep{arn92}. The fractions of neon and iron have been scaled down

193:    with respect to the relative solar abundance ratio to oxygen.

194:    The {\sl vertical line} indicates the best fit temperature 

195:    ($2.2\times10^6$ K) in the isothermal case 

196:    for the absorbing gas toward \source.

197: \label{fig:CIE} }

198: \end{figure}

199: 

200: 

201: 

202: \section{Source, observations, and the existing results \label{sec:source}}

203: 

204: \source\ is a bright low mass X-ray binary residing in the globular cluster

205: NGC 6624 (Galactic coordinates $l, b = 2\fdg79, 7\fdg91$) and its

206: distance has been determined as $7.6\pm0.4$ kpc \citep{kuu03},

207: meaning that it is very close to the Galactic center and is located 

208: $\sim 1$ kpc above the disk plane.

209: 

210: The {\sl Chandra X-Ray Observatory} has observed this source three times

211: with different instrumental configurations. 

212: The observation log, data reduction and analysis procedures

213: have been described in detail in Yao \& Wang (2006, hereafter Paper I),

214: and here, we summarize absorption line detections and

215: relevant absorption line diagnostic results.

216: 

217: We have detected highly ionized \ovii,

218: \oviii, and \neix\ K$\alpha$, and \ovii\ K$\beta$ absorption lines, 

219: which are produced in the hot ISM rather than in the circumstellar

220: gas associated with the binary system (Paper I; see also Futamoto \etal 2004;

221: Juett \etal 2006).

222: Modeling these lines with our absorption line model, 

223: {\sl absline} \citep{yao05}, we have constrained dispersion velocity of the

224: hot gas [$v_b=255(165, 369)\ {\rm km\ s^{-1}}$], 

225: column densities of \ovii, \oviii, and \neix. We have also obtained the

226: abundance ratio of Ne/O in the hot gas, which is consistent with the solar 

227: value. For a gas at temperature about $10^6-10^7$ K, the population of each 

228: abundant iron ion contained in the gas, e.g., \fexvi, \fexvii, and \fexviii, 

229: is distributed in a narrow temperature range (Fig.~\ref{fig:CIE}); therefore,

230: a well confined gas temperature or its distribution is crucial for inferring 

231: the total iron in the hot gas. The detection of multiple absorption lines

232: in this sight line enables us to obtain such a constraint.

233: For instance, if the intervening gas is in the collisional 

234: ionization equilibrium (CIE) state \citep{arn92} and isothermal, 

235: its temperature can be determined as $T=2.2\pm0.3\times10^6$ K (Paper I).

236: 

237: \section{\fexvii\ absorption line and iron abundance in the hot ISM}

238: 

239: We searched for the ionized iron absorption lines at the corresponding rest 

240: frame wavelengths in the wavelength range between 2 and 25\AA\

241: in the spectrum obtained

242: in Paper I. The \fexvi\ absorption lines are expected to be very

243: weak (the oscillation strength $f_{\rm ij} < 10^{-6}$), and are not 

244: considered further in this work. Table~\ref{tab:lines} lists the 

245: strong lines ($f_{\rm ij} > 0.5$) of ions \fexvii\ and \fexviii. We only

246: detected a significant \fexvii\ absorption line at 15.02\AA\ 

247: (Fig.~\ref{fig:counts}), and did not see any clear sign for the other lines 

248: listed in Table~\ref{tab:lines}. These detection results are not surprising. 

249: The constrained hot gas temperature favors the \fexvii\ population, and 

250: the transition of the 15.02 \AA\ line is strongest 

251: (Fig.~\ref{fig:CIE}; Table~\ref{tab:lines}). Therefore,

252: the \fexvii\ absorption line at $\sim15.02$ \AA\ is expected to be at least 

253: 4 times stronger (in terms of equivalent width [EW]) than the others.

254: 

255: \begin{deluxetable}{llcc}

256: \tablewidth{0pt}

257: \tablecaption{Strong Transition Lines of \fexvii\ and \fexviii

258: \label{tab:lines}}

259: \tablehead{

260:  Ion & Transition & $\lambda$ (\AA)& $f_{\rm ij}$ }

261: \startdata

262: \fexvii\ & $2s^22p^6(^1S)-2s^22p^54d(^1P^0)$ & 12.123 & 0.53 \\

263: \fexvii\ & $2s^22p^6(^1S)-2s^22p^53d(^1P^0)$ & 15.015 & 2.31 \\

264: \fexvii\ & $2s^22p^6(^1S)-2s^22p^53d(^3D^0)$ & 15.262 & 0.63 \\

265: \fexviii\ & $2s^22p^5(^2P^0)-2s^22p^4(^1S)3d(^2D)$ & 14.121 & 0.90 \\

266: \fexviii\ & $2s^22p^5(^2P^0)-2s^22p^4(^1D)3d(^2P)$ & 14.203 & 0.57 \\

267: \fexviii\ & $2s^22p^5(^2P^0)-2s^22p^43d(^2D)$ & 14.361 & 0.93 \\

268: \hline                   

269: \enddata                 

270: \end{deluxetable}

271: 

272: We use different models to characterize the \fexvii\ absorption 

273: line at 15.02\AA. The negative Gaussian model 

274: gives the line centroid % 826.1(825.6, 826.6) eV

275: as 15.008(14.999, 15.018) \AA\ or 132(-50, 314) ${\rm km\ s^{-1}}$, 

276: line width as

277: $\sigma<280\ {\rm km\ s^{-1}}$ or $v_b<396\ {\rm km\ s^{-1}}$, and its 

278: EW as 5.1(2.9, 7.3) mA. %0.28 (0.16, 0.40) eV;; 

279: 

280: \begin{figure}[hb!]

281: \centerline{

282: \includegraphics[width=0.47\textwidth]{f2.eps}}

283: \caption{\fexvii\ absorption line at 15.02\AA\ in the {\sl Chandra}

284:    MEG-LEG combined spectrum of \source, 

285:    modeled with the {\sl absline} model ({\sl histogram}). 

286:    The {\sl dip} at 15.26 \AA\ indicates the expected amount of absorptions

287:    for the second most significant \fexvii\ line (Table~\ref{tab:lines}). 

288:    The dotted line indicates the smooth continuum level around the line. 

289:    The model has

290:    been convolved with the instrumental response, and the binsize is 

291:    12.5 m\AA. The apparent bump on the blue-side of the line may be due to

292:    the imperfect calibration near the hot pixel. For the detail information

293:    of the co-added spectrum and the continuum fit, please refer to Paper I.

294:    \label{fig:counts} }

295: \end{figure}

296: 

297: 

298: We then fit the line with our {\sl absline} model, which adopts the 

299: Voigt function as line profile and allows a joint analysis 

300: of multiple absorption lines (see Yao \& Wang 2005; Paper I; 

301: Wang \etal 2005 for further discussion). The fit is as good as with

302: a Gaussian model, and the obtained line 

303: position is also identical. Connecting the $v_b$

304: with those of \ovii, \oviii, and \neix\ lines (since the non-thermal 

305: broadening dominates, we therefore ignore the tiny differences of the thermal

306: broadening in different elements), we obtain the column density of the

307: \fexvii\ as $\log[N_{\rm FeXVII}({\rm cm^{-2}})]=15.0(14.7, 15.2)$.

308: \footnote{

309: This column density, together with the dispersion velocity $v_b$ constrained in

310: jointly analyzing oxygen and neon lines (\S2),

311: can reproduce the EW measured with Gaussian model.}

312: %% 9.93574e+14  5.31129e+14  1.50176e+15

313: % possible useful measures: 

314: %The uncertainty of the iron energy (wavelength) is

315: %826.131(825.646, 826.587) eV, which is equivalent to 

316: %15.009(15.001, 15.018);; APEC: 15.014; 0.8258;; verner: 150.015

317: 

318: Following the procedure we established in Paper I, we probe the abundance ratio

319: of Fe/Ne in the hot gas. Since neon is a noble element, it is very unlikely 

320: depleted into dust grains. Therefore we take it as the reference element. 

321: In fact, we have obtained a Ne/O ratio that is consistent with the solar 

322: value (Paper I), and the following inferred Fe/Ne is essentially the same as 

323: Fe/O in units of solar value. Assuming that the absorbing gas is in a CIE state

324: and isothermal, we jointly analyze the \fexvii\ line with the \ovii, \oviii,

325: and \neix\ K$\alpha$ lines and the \ovii\ K$\beta$ line, requiring the common

326: absorbing gas to be of the same temperature. We fix the neon abundance at

327: the solar value, and let the abundances of oxygen and iron be free parameters.

328: In this way we constrain the hydrogen column density  to 

329: $N_{\rm H}=7.9(5.0, 10.2)\times10^{19}\ {\rm cm^{-2}}$, % 19.9(19.7, 20.01) 

330: the temperature to $T=2.2(1.9, 2.5)\times10^6$ K, 

331: and the abundance ratio to O/Ne = 0.9(0.5, 1.3) solar

332: for the absorbing hot gas, which are identical to those reported in Paper I. 

333: In addition, we obtain the abundance ratio of Fe/Ne as 0.8(0.4, 2.1) 

334: solar. %0.78(0.36, 2.08). 

335: Considering the dependence of the \fexvii\ population

336: on $T$ (Fig.~\ref{fig:CIE}), we calculate the confidence contours of Fe/Ne 

337: versus $T$, which is presented in Figure~\ref{fig:FeNe}a. 

338: 

339: %The other measures are: \\

340: %for single T: 2.2(1.9, 2.5)E6 K \\ %% 6.34(6.28, 6.40)

341: %NOVII: 2.1(1.4, 3.2)E16 \\ %% 16.33(16.50, 16.14)

342: %NOVIII: 2.4(3.6, 1.6)E16 \\

343: %bv: 255(165, 369) \\

344: %NFeXVII: 1.0(0.6, 1.5)E15 \\ %%15.0(14.73, 15.18)

345: %Fe/O: 0.9(0.4, 2.7) \\ 

346: 

347: Next, we investigate the effects on the inferred Fe/Ne ratio if the above

348: isothermal assumption of the intervening gas is relaxed.

349: Since the absorption samples almost the entire Galactic disk

350: from the Sun into the Galactic bulge, it is possible that the hot gas 

351: consists of different temperature components. Here we examine two 

352: simple temperature distributions. In each case, we first interpolate the 

353: ionization fractions at different temperatures, assuming the gas to be 

354: in CIE state, 

355: and then calculate the column density for each ion.

356: To get a better constraint, we also add the undetected K$\alpha$ 

357: \nex\ (12.134 \AA) absorption line in the joint fit. 

358: This line, except for the detected line absorptions from oxygen, neon, and

359: iron ions, is the next most expected one from a different ion to be observed

360: in the spectrum with high counting statistic because of its anticipated large

361: column density (Fig.~\ref{fig:CIE}) and large transition coefficient 

362: ($f_{\rm ij}=0.416$), and is particular useful for constraining the upper 

363: boundary to the gas temperature.

364: In the first case, we assume that the hot gas temperature distribution

365: follows a logarithmic Gaussian form, as a natural extension of the 

366: isothermal single temperature case, 

367: \begin{equation}

368: %\[ 

369: {\rm d}N_{\rm H}(T) \propto \exp\left[{\frac{-({\rm~log}T - {\rm~log}T_0)^2}{2(\sigma_{{\rm log}T})^2}}\right] {\rm d}{\rm log}(T), 

370: %\]

371: \end{equation}

372: where the mean temperature 

373: $T_{\rm 0}$ is equivalent to $T$ in the isothermal case, and 

374: $\sigma_{\log T}$ is the dispersion of $\log T_{\rm 0}$. Under this assumption,

375: we obtain $T_{\rm 0}=2.0(1.8, 2.4)\times10^6$ K and

376: $\sigma_{\log T}< 0.15$, and 

377: the abundance ratio Fe/Ne = 0.8(0.4, 2.3) solar. %0.84(0.41, 2.3).

378: In Paper I, we have demonstrated that since an isothermal absorbing 

379: plasma is adequate to describe the observation, the additional free parameter

380: $\sigma_{\log T}$ cannot be fully constrained in the spectral fitting, 

381: and there is an apparent 

382: correlation between $T_{\rm 0}$ and $\sigma_{\log T}$ (see Fig.~4 in Paper I)

383: due to the large value of $N_{\rm OVIII}$. But the inferred Fe/Ne is

384: insensitive to different $\sigma_{\log T}$ values, as illustrated

385: in the confidence contours of the Fe/Ne ratio versus $\sigma_{\log T}$ 

386: (see Fig.~\ref{fig:FeNe}b). 

387: %The other measures for Gaussian case are:

388: %$T:6.97(6.26, **)$ \\

389: %$\sigma_{\rm logT}: < 0.94$ \\

390: %Fe/0: 0.6(0.3, 1.2) \\

391: 

392: \begin{figure}

393: \centerline{

394: \includegraphics[width=0.47\textwidth]{f3.eps}}

395: \caption{The 68\%, 90\%, and 99\% confidence contours of (a) the abundance 

396:    ratio of Fe/Ne (in units of solar value) vs. gas temperature $T$, 

397:    (b) Fe/Ne vs. temperature 

398:    dispersion, and (c) Fe/Ne vs. power-law index $\gamma$ for the 

399:    isothermal case, log-Gaussian temperature distribution, and the 

400:    power-law temperature distribution, respectively.

401: \label{fig:FeNe} }

402: \end{figure}

403: 

404: 

405: In the second case, we  assume that the hot gas temperature

406: distribution follows a power-law (PL) form, 

407: \begin{equation} \label{equ:PL}

408:     dN_{\rm H}(T) = \frac{N_{\rm H0}(\gamma+1)}{T_{\rm 0}} (T/T_{\rm 0})^{\gamma} dT.

409: \end{equation}

410: This simple characterization of the temperature distribution

411: can be derived, for example, naturally from an exponential disk model 

412: \citep{yao06b}, where

413: $N_{\rm H0}$ is the total hydrogen column density along the sight line, 

414: $T_{\rm 0}$ is the Galactic mid-plane temperature, and the PL index 

415: $\gamma$ is the ratio of the temperature to density 

416: scale heights. Our joint analysis gives 

417: $N_{\rm H}=7.5(5.3, 10.0)\times10^{19}\ \rm{cm^{-2}}$, 

418: $\gamma>2$, and $T_{\rm 0} = 2.4(2.1, 3.4)\times10^6$ K.

419: Again, although the extra free parameter $\gamma$ can vary

420: in a large range, the constrained abundance ratio 

421: Fe/Ne=0.9(0.4, 2.0) solar %0.86(0.39, 1.95). 

422: is insensitive to $\gamma$.

423: Fig.~\ref{fig:FeNe}c shows the confidence 

424: contours of the Fe/Ne ratio versus the PL index $\gamma$.

425: % the other measures for PL case are:

426: %$\gamma>1.2$ \\

427: %T:2.35(2.04, 3.38)E6 \\

428: %Fe/O: 0.87(0.41, 2.11)\\

429: 

430: \section{Discussion}

431: 

432: We detect a significant \fexvii\ absorption line at $\sim15.02$ \AA\ 

433: in the {\sl Chandra} spectrum of 4U~1820--303. A

434: joint analysis of this line with the detected highly ionized oxygen and 

435: neon lines, all interstellar in origin and observed in

436: the same spectrum, gives the abundance ratio of Fe/Ne in the hot

437: ISM component,  which, although with large errors,

438: is consistent with the solar value. In addition, this result appears to be 

439: unaffected by the different gas temperature distributions adopted. 

440: We conclude that there is no evidence for substantial depletion of iron 

441: into dust grains in the hot ISM. This is in 

442: contrast to cooler phases of the ISM where iron is usually found to be

443: heavily depleted \citep{sem96, jue06}.

444: Grain cores containing iron oxides are generally rather resilient and it 

445: is quite difficult to liberate the iron from these cores \citep{sem96, fri03}. 

446: This solar value of Fe/Ne ratio, if confirmed, thus suggests that likely all 

447: of the dust in this

448: very hot ISM phase (T $\gsim 10^6$ K) has been destroyed by frequent 

449: and/or severe shocks during the dust grain processing in the ISM. 

450: 

451: Dust grains pre-existing in the ISM or formed in supernova ejecta can 

452: be destroyed by their generated forward and reverse shocks and 

453: subsequently, in heated hot gas.

454: By studying the iron abundance via the far-UV \feii\ absorption 

455: lines, \citet{sem96} find that while more than $99\%$ of iron is 

456: depleted into dust grains in the cold disk of the Galaxy,

457: the iron depletion is $\sim 80\%$ in warm clouds of the Galactic halo.

458: They attribute this difference to the dust grain 

459: disruption by the supernova (SN) shocks when circulating the grains between the 

460: Galactic disk and halo. If the hot ISM is believed to be heated from 

461: the cool ISM, it should have experienced shocks much more frequently 

462: and/or much more violently than the warm halo clouds. 

463: Therefore it is natural to expect that many more grains, even including

464: the resilient iron-rich cores, could have been destroyed in such harsh

465: environments. More recently, 

466: \citet{str04} have obtained an abundance ratio of Fe/O from

467: the diffuse extraplanar halo emission of many disk galaxies like our own,

468: which is $\sim40\%$ solar.

469: This result, although subject to different thermal plasma models adopted 

470: and the different thermal properties assumed for the emitting hot gas

471: in the data analysis, clearly rules out the depletion pattern found 

472: by \citet{sem96} in the cold and warm gas of the Galactic disk and halo,

473: further supporting the scenario that more iron has been released back

474: to gas phase.

475: 

476: % 2). More iron produced by type Ia SNe in the Galactic center

477: The above interpretation may not be entirely unique and in some cases is 

478: subject to systematic

479: effects introduced by line-of-sight variations and intrinsic X-ray source 

480: properties. Claims like an overabundance of heavier elements relative

481: to oxygen in neutral matter towards the Galactic center direction  

482: presumably caused by a significant contribution of Type Ia SNe in the 

483: Galactic bulge \citep{ueda05} would elevate the contribution of iron in 

484: the hot gas phase as well.

485: In this respect the result by \citet{jue06} that iron is significantly 

486: depleted in the cool phase in line of sight towards 4U~1820--303 and that Ne/O 

487: is close to the solar value is an important indicator 

488: that conditions are not so unusual.

489: Abnormal heavy element abundances have been observed in microquasars 

490: (e.g., Cyg~X-1: Schulz \etal 2001, Juett \etal 2004; 

491: GRS~1915+105: Lee \etal 2002, GX 339-4: Miller \etal 2004). 

492: In all these systems, the observed 

493: overabundance pattern of the heavy elements relative to 

494: oxygen can also be interpreted as the different photoionization structures 

495: of these elements in the intrinsic absorbing material \citep{sch02, lee02}.

496: \citet{fut04} and Juett \etal (2004, 2006) already argued 

497: successfully that this is likely not the case for \source.

498: 

499: %4). The Galactic bulge confusion. 

500: As a final note, we point out that the absorption path length toward

501: \source\ passes through the Galactic bulge region where the soft X-ray

502: background emission in the 0.75 keV band is greatly enhanced 

503: \citep{sno97}.  This enhancement indicates

504: that the emitting plasma is either of a high temperature or of a 

505: dense emitting region, or both. Recently, we have obtained an exponential 

506: scale height of $\sim2$ kpc and the mid-plane density of 

507: $\sim2.4\times10^{-3}\ {\rm cm^{-3}}$ for the hot gas toward Mrk~421 

508: (Galactic coordinates $l,b=179\fdg83, 65\fdg03$) in a joint-analysis 

509: of absorption and emission data at the same time \citep{yao06b}.

510: This characterization, if typical for the general hot ISM, only accounts 

511: for $\sim30\%$ of the observed 

512: absorption toward \source, meaning that a large portion of the absorption

513: originates from the Galactic bulge.

514: Therefore, the gas phase abundance ratio of Fe/Ne presented in this

515: letter could be biased because of the remarkable absorption contribution

516: from the bulge, where the metal abundance pattern may not be as the 

517: same as that in the overall ISM. 

518: Nevertheless, we present here a feasible way to infer the gas phase iron

519: abundance in the hot ISM that potentially affects our understanding of the 

520: cooling/heating process in the ISM in general. To obtain a global  

521: picture of the gas phase iron abundance in the hot gas, high quality 

522: absorption data along other sight lines that are away from the Galactic 

523: bulge region are therefore required.

524: 

525: 

526: \acknowledgments

527: We thank the referee for valuable suggestions that helped to improve

528: our presentation. We are also grateful to Claude Canizares and 

529: Aigen Li for useful discussions. 

530: This work is supported by NASA through the 

531: Smithsonian Astrophysical Observatory (SAO) contract SV3-73016 to

532: MIT for support of the {\sl Chandra} X-Ray Center, which is operated by the 

533: SAO for and on behalf of NASA under contract NAS 08-03060. Support from a 

534: {\sl Chandra} archival research grant AR6-7023X is also acknowledged.

535: 

536: 

537: \begin{thebibliography}{}

538: \bibitem[Anders \& Grevesse (1989)]{and89} Anders, E., \& Grevesse, N. 1989, Geochim. Cosmochim. Acta, 53, 197

539: \bibitem[Antia \& Basu (2006)]{ant06} Antia, H. M., \& Basu, S. 2006, \apj, 644, 1292

540: \bibitem[Arnaud \& Raymond (1992)]{arn92} Arnaud, M., Raymond, J. 1992, \apj, 398, 394

541: \bibitem[Asplund, Grevesse, \& Sauval (2005)]{asp05}Asplund, M., Grevesse, N., \& Sauval, J. 2005, ASPC, 336, 25

542: \bibitem[Draine \& Salpeter (1979)]{dra79} Draine, B. T., \& Salpeter, E. E. 1979, \apj, 231, 438

543: \bibitem[Dwek \& Arendt (1992)]{dwek92} Dwek, E., \& Arendt, R. G. 1992, \araa, 30, 11

544: \bibitem[Frisch \& Slavin (2003)]{fri03} Frisch, P. C., \& Slavin, J. D. 2003, \apj, 594, 844

545: \bibitem[Futamoto \etal (2004)]{fut04} Futamoto, K., Mitsuda, K., Takei, Y., Fujimoto, R., \& Yamasaki, N. Y. 2004, \apj, 605, 793

546: \bibitem[Hurwitz, Sasseen, \& Sirk (2005)]{hur05} Hurwitz, M., Sasseen, T. P., \& Sirk, M. M. 2005, \apj, 623, 911

547: \bibitem[Jenkins \& Wallerstein (1995)]{jen95} Jenkins, E. B., \& Wallerstein, G. 1995, \apj, 440, 227

548: \bibitem[Jones \etal (1994)]{jon94} Jones, A. P., Tielens, A. M., Hollenbach, D. J., McKee, C. F. 1994, \apj, 433, 797

549: \bibitem[Juett \etal (2004)]{jue04} Juett, A., Schulz, N., Chakrabarty, D. 2004, \apj, 612, 308

550: \bibitem[Juett \etal (2006)]{jue06} Juett, A., Schulz, N., Chakrabarty, D., \etal 2006, \apj, 648, 1066

551: \bibitem[Kuulkers \etal (2003)]{kuu03} Kuulkers, E., den Hartog, P. R., in't Zand, J. J. M., \etal 2003, A\&A, 399, 663

552: \bibitem[Lee \etal (2002)]{lee02} Lee, J., Reynolds, C., Remillard, R., \etal 2002, \apj, 567, 1102

553: %%\bibitem[Matsushita \etal (2003)]{mat03} Matsushita, K., Finoguenov, A., \& B\"ohringer, H. 2003, A\&A, 401, 443

554: %%\bibitem[McCammon \etal (2002)]{mcc02} McCammon, D., Almy, R., Apodaca, E., \etal 2002, \apj, 576, 188

555: \bibitem[Miller \etal (2004)]{mil04} Miller, J. M., Fabian, A. C., Reynolds, C. S., \etal 2004, \apj, 606, 131 

556: \bibitem[Pepino \etal (2004)]{pep04} Pepino, R., Kharchenko, V., Dalgarno, A., \& Lallement, R. 2004, \apj, 617, 1347

557: \bibitem[Savage \& Sembach (1996)]{sav96} Savage, B. D., \& Sembach, K. R. 1996, \araa, 34, 279

558: \bibitem[Sanders \etal (2001)]{san01} Sanders, W. T., Edgar, R. J., Kraushaar, W. L., \etal 2001, \apj, 554, 694

559: \bibitem[Schulz \etal (2001)]{sch01} Schulz, N., Chakrabarty, D., Marshall, H. \etal 2001, \apj, 563, 941

560: \bibitem[Schulz \etal (2002)]{sch02} Schulz, N., Cui, W., Canizares, C., \etal 2002, \apj, 565, 1141

561: \bibitem[Sembach \& Savage (1996)]{sem96} Sembach, K. R., \& Savage, B. D., 1996, \apj, 457, 211

562: \bibitem[Snowden \etal (1997)]{sno97} Snowden, S. L., Egger, R., Freyberg, M. J., \etal 1997, \apj, 485. 125

563: \bibitem[Sofia \etal (1994)]{sof94} Sofia, U., J., Cardelli, J. A., \& Savage, B. D. 1994, \apj, 430, 650

564: \bibitem[Strickland \etal (2004)]{str04} Strickland, D. K., Heckman, T. M., Colbert, E. J., \etal 2004, \apjs, 151, 193

565: \bibitem[Ueda \etal (2005)]{ueda05} Ueda, Y., Mitsuda, K., Murakami, H., \& Matsushita, K. 2005, \apj, 620, 274

566: %%\bibitem[Wilms \etal (2000)]{wil00} Wilms, J., Allen, A., \& McCray, R. 2000, \apj, 542, 914

567: \bibitem[Wang \etal (2005)]{wang05} Wang, Q. D., Yao, Y., Tripp, T. M., \etal 2005, \apj, 635, 386

568: \bibitem[Yao \& Wang (2005)]{yao05} Yao, Y., \& Wang, Q. D. 2005, \apj, 624, 751 

569: \bibitem[Yao \& Wang (2006)]{yao06} ---. 2006, \apj, 641, 930 (Paper I)

570: \bibitem[Yao \& Wang (2007)]{yao06b} ---. 2007, \apj, submitted

571: %%\bibitem[Young (2005)]{young05} Young, P. R. 2005, A\&A submitted, astro-ph/0510264

572: 

573: \end{thebibliography}

574: 

575: \end{document}

576: