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

2: 

3: \usepackage{color}

4: 

5: \newcommand{\vdag}{(v)^\dagger}

6: \newcommand{\myemail}{sokada@astro.isas.jaxa.jp}

7: \newcommand{{\ka}}{${\rm K\alpha}$}

8: 

9: \shorttitle{SS~Cyg observations}

10: \shortauthors{Okada, Nakamura and Ishida}

11: \begin{document}

12: 

13: \title{{\it Chandra} HETG line spectroscopy

14: of the Non-magnetic Cataclysmic Variable SS~Cyg}

15: 

16: 

17: \author{{\sc Shunsaku OKADA}\altaffilmark{1}, {\sc Ryoko

18: NAKAMURA\altaffilmark{1}}, and {\sc Manabu ISHIDA}}

19: \affil{Institute of Space and Astronautical Science, 3-1-1, Yoshinodai,

20: Sagamihara, Kanagawa 229-8150, Japan}

21: \email{sokada@astro.isas.jaxa.jp, nakamura@astro.isas.jaxa.jp,

22: ishida@astro.isas.jaxa.jp}

23: 

24: \altaffiltext{1}{Also at Department of Physics, Tokyo Institute of

25: Technology, 2-12-1 Ohokayama, Meguro, Tokyo 152-8551, Japan}

26: 

27: \begin{abstract}

28: 

29:  We present {\it Chandra} HETG observations of SS~Cygni in quiescence

30:  and outburst. The spectra are characterized by He-like and H-like {\ka}

31:  emission lines from O to Fe, as well as L-shell emission lines from

32:  Fe. In quiescence, the spectra are dominated by the H-like {\ka} lines,

33:  whereas in outburst the He-like lines are as intense as the H-like

34:  lines. In outburst, the H-like {\ka} lines from O to Si are broad, with

35:  widths of 4--14$\ {\rm eV}$ in Gaussian $\sigma$ (1800--2300$\ {\rm km\

36:  s^{-1}}$). The large line widths, together with line profiles, indicate

37:  that the line-emitting plasma is associated with the Keplerian disk and

38:  still retains the azimuthal bulk motion. In quiescence, the emission

39:  lines are narrower, with a Gaussian $\sigma$ of 1--3$\ {\rm eV}$

40:  (420--620$\ {\rm km\ s^{-1}}$). A slightly larger velocity for lighter

41:  elements suggests that the lines in quiescence are emitted from an

42:  ionizing plasma at the entrance of the boundary layer, where the bulk

43:  motion of the optically thick accretion disk is converted into heat due

44:  to friction.  Using the line intensity ratio of He-like and H-like {\ka}

45:  lines for each element, we have also investigated the temperature

46:  distribution in the boundary layer both in quiescence and outburst.

47:  The distribution of SS~Cyg is found to be consistent with other dwarf

48:  novae investigated systematically with {\it ASCA} data.

49: \end{abstract}

50: \keywords{accretion, accretion disks --- dwarf novae, cataclysmic variables

51: --- plasmas --- stars:individual (SS Cygni) --- X-rays: stars}

52: 

53: 

54: \section{Introduction}

55: 

56: A dwarf nova is a close binary system consisting of a weakly magnetized

57: white dwarf and a low-mass companion filling its Roche lobe and is

58: characterized by frequent and large-amplitude optical outbursts.

59: Mass accretion from the companion occurs via an optically thick disk extending down close to the white dwarf, and a so-called

60: boundary layer is formed between the white dwarf and the innermost

61: edge of the accretion disk, where the accreting matter is braked

62: due to strong friction and finally settles onto the white dwarf.

63: 

64: After a long debate since the 1970's, it is now widely accepted that the

65: optical outburst arises due to a thermoviscous instability in an outer

66: region of the accretion disk.  In quiescence, matter from the companion

67: is accumulated in the disk because the hydrogen is neutral, and hence

68: the disk viscosity is low.  As matter accumulates and the temperature

69: exceeds $\sim$10$^4$~K, the viscosity increases due to hydrogen

70: ionization, and matter rushes onto the white dwarf at a rate exceeding

71: the supply rate from the companion.  This latter state is the optical

72: outburst

73: \citep{1984PASP...96....5S,1993ApJ...419..318C,1996PASP..108...39O,2001NewAR..45..449L}.

74: In response to the phase transition of the outer disk, it is predicted

75: that the physical state of the boundary layer changes significantly.  In

76: quiescence, the boundary layer is optically thin, because the mass

77: accretion rate is low and the radiative cooling is inefficient.  The

78: temperature is as high as $10^8$~K, and the boundary layer is a hard

79: X-ray emitter.  In outburst, on the other hand, the boundary layer can

80: remain optically thick just as for the outer region of the disk, because

81: of efficient radiative cooling.  Since its temperature is

82: $\sim\!10^5$~K, it emits mainly in the {\it EUV} band

83: \citep{1977MNRAS.178..195P,1979MNRAS.187..777P,1981AcA....31..267T,1985ApJ...292..535P}.

84: The first observational suggestion to support this phase transition was

85: obtained from a series of {\it European X-ray Observatory Satellite}

86: ({\it EXOSAT}) observations \citep{1992MNRAS.257..633J} and was finally

87: established by \citet{2003MNRAS.345...49W}.

88: 

89: Although the phase transition of the boundary layer is better understood

90: than in the past, there still remain several unresolved issues.

91: Although the boundary layer is believed to be optically thick in

92: outburst, optically thin hard X-ray emission above 2~keV does not vanish

93: completely \citep{1994ApJ...436L..19N,2003MNRAS.345...49W}.  Unlike the

94: magnetic CVs in which the postshock accretion flow is essentially

95: one-dimensional, the temperature distribution of the boundary layer

96: plasma is not understood very well.  The temperature distribution is

97: probably tightly linked with the geometry of the boundary layer, for

98: which we have no more than a rough sketch.  In order to help resolve

99: these issues, we analyzed {\it Chandra} HETG data of SS~Cyg.  Since

100: SS~Cyg is the brightest dwarf nova in X-rays, it has been observed with

101: many X-ray astronomy satellites, such as {\it EXOSAT}, {\it Ginga}, the

102: {\it Rossi X-ray Timing Explorer} ({\it RXTE}), the {\it Advanced

103: Satellite for Cosmology and Astrophysics} ({\it ASCA}), {\it

104: Chandra}, and {\it Suzaku}

105: \citep{1992PASJ...44..537Y,1994ApJ...436L..19N,1997MNRAS.288..649D,2003MNRAS.345...49W,2004ApJ...610..422M,2005ASPC..330..355M,2007PThPS.169..178I}.

106: Since the {\it Chandra} HETG has an advantageous energy resolution, we

107: have attempted to approach these unresolved issues by means of

108: spectroscopy of He-like and H-like {\ka} emission lines.  We discuss the

109: temperature distribution and the geometrical structure of the boundary

110: layer based on our results.  In \S~\ref{sec:Observations} we introduce

111: the {\it Chandra} observations of SS~Cyg in outburst and quiescence and

112: present light curves and spectra. In \S~\ref{sec:DataAnalysis}, the

113: data analysis method and the results are presented. The results are

114: discussed in \S~\ref{sec:Discussion}, and the conclusions are summarized

115: in \S~\ref{sec:Conclusion}.

116: 

117: 

118: \section{Observations\label{sec:Observations}}

119: \subsection{{\it Chandra} data} 

120: 

121: The SS~Cyg data analyzed in this paper were retrieved from the {\it Chandra}

122: Data Archive\footnote{http://asc.harvard.edu/cda/} maintained by

123: the Smithonian Astrophysical Observatory.  The data in quiescence were obtained

124: between 10:28 UT on 2000 August 24 and 00:19 UT on August 25, with a

125: total good

126: time interval of 47.3 ks. The outburst data were obtained from 21:09 UT on 2000 September 14 to 14:15 UT on the following

127: day, with a total good exposure time of 59.5 ks. Both {\it Chandra}

128: observations employed the High Energy Transmission Grating (HETG)

129: \citep{2005PASP..117.1144C}, which allows us to carry out fine

130: spectroscopy with energy resolution better than CCD resolution by roughly an order

131: of magnitude. These data are analyzed also by

132: \citet{2003ApJ...586L..77M}, \citet{2005ASPC..330..355M}, and \citet{2006ApJ...642.1042R}. By referring to the light curve from the AAVSO home

133: page\footnote{http://www.aavso.org/}, we know that the visual magnitude

134: of SS~Cyg during the quiescence observation was in the usual range,

135: 11.5--12.5. It then went into a normal outburst around September

136: 10. Hence the {\it Chandra} outburst observation was performed $\sim$4--5

137: days after the onset of the outburst ($\sim$3 days after the peak with

138: $m_V \simeq 8.5$). The visual magnitude during the {\it Chandra}

139: observation was $m_V$ = 9.1--9.7.

140: 

141: %

142: \subsection{Light curves in quiescence and outburst}

143: %

144: The HETG consists of two independent grating instruments: the Medium

145: Energy Grating (MEG), which covers the 0.4--5.0 keV band with an energy

146: resolution $ E/\Delta E$ of $\sim$200 at 2.0 keV, and the High Energy

147: Grating (HEG), which covers the 0.8--10.0 keV band with $E/\Delta E

148: \simeq$ 200 at 6.0~keV. In analyzing the data, we used

149: the CIAO software (Version 3.2) and CALDB (Version 3.1.0) provided by

150: the {\it Chandra} data center. Figure~\ref{fig:LightCurve} shows the

151: HETG light curves of SS~Cyg in quiescence and in outburst. These light

152: curves were made by summing up the positive and negative first-order

153: photons in the above energy bands with a 512~s bin width using the

154: thread {\tt dmextract}. Backgrounds are not

155: subtracted. The count rate of the MEG was ${\rm 2.90\pm 0.47\ c\

156: s^{-2}}$ and ${\rm 1.00\pm 0.07\ c\ s^{-1}}$ in quiescence and outburst,

157: respectively. That of the HEG was, on the other hand, ${\rm 1.29\pm

158: 0.22\ c\ s^{-1}}$ and ${\rm 0.44\pm 0.04\ c\ s^{-1}}$ in quiescence and

159: outburst. The intensity of SS~Cyg in outburst is less than that in

160: quiescence by a factor of 3 on average and is less variable. The

161: weaker X-ray intensity during optical outburst is associated with the

162: optically thin-to-thick transition of the boundary layer when the mass

163: transfer rate in the outer disk exceeds a critical value

164: \citep{1977MNRAS.178..195P,1979MNRAS.187..777P,1981AcA....31..267T,1985ApJ...292..535P}.

165: Such events are detected by coordinated optical, EUV, and X-ray

166: observations \citep{2003MNRAS.345...49W}.

167: 

168: 

169: \subsection{HETG spectra in quiescence and outburst

170: \label{sec:HETGspecQandO}}

171: %

172: In Fig.~\ref{fig:Spectrum} we show spectra of SS~Cyg in quiescence and

173: outburst from the MEG and the HEG separately. 

174: The spectrum files (PHA files) were extracted from the level 1 data, and

175: the positive and negative orders were combined to enhance the

176: signal-to-noise ratio.  The MEG has a larger effective area in the lower

177: energy range and has clearly detected {\ka} emission lines from O, Ne,

178: Mg, and Si, while the HEG is advantageous at higher energies and has

179: detected {\ka} lines from Fe.  Simultaneous detection of the {\ka}

180: emission lines from these elements indicates the existence of optically

181: thin thermal plasma with a wide range of temperature distribution, both

182: in quiescence and outburst.  In quiescence, the spectra are dominated by

183: the H-like lines, whereas in outburst, the He-like lines are as strong as

184: the H-like lines. It is remarkable that, as demonstrated in

185: Fig.~\ref{fig:Spectrum}(c) and (d), the emission lines in outburst are

186: much broader than in quiescence. These characteristics

187: are also pointed out by \citet{2005ASPC..330..355M}. The profiles of

188: the H-like lines appear rectangular, rather than Gaussian-like.

189: 

190: \section{Data analysis\label{sec:DataAnalysis}}

191: %

192: %

193: As described in \S~1, we aim to elucidate the nature of the boundary

194: layer of SS~Cyg in terms of {\ka} emission lines from abundant metals.

195: We restrict ourselves to those from O up to Si, because the quantum

196: efficiency of both the MEG and HEG abruptly drops by a factor of $\sim$5

197: at the iridium M edge $\sim$2.1~keV (Fig.~\ref{fig:Spectrum}), above

198: which energy the line parameters are not very well constrained.  Since

199: the MEG has better sensitivity than the HEG in this energy band,

200: we concentrate on the MEG spectra, except for evaluation

201: of the H-like Si {\ka} line, for which we also utilize the HEG

202: data (\S~3.3 and \S~3.4).

203: 

204: \subsection{Continuum

205: \label{sec:continuum}}

206: %

207: In analyzing emission lines, we first attempted to evaluate the

208: continuum spectrum of the MEG in the 0.4--2.3 keV band.  We masked

209: energy bins, including the emission lines, and fitted the remaining

210: spectrum with a bremsstrahlung model undergoing a

211: uniform photoelectric absorption.  Prior to the fitting process, the

212: spectra were rebinned so that each energy bin includes at least 100

213: photons in quiescence and 50 photons in outburst, respectively.

214: The fits to the quiescence and outburst spectra were

215: marginally acceptable at the 90\% confidence level ($\chi^2$ [dof] of

216: 452.5 [441] and 300.7 [241], respectively), with temperatures of ${\rm

217: 14.5^{+4.8}_{-3.5}}$ and ${\rm 4.3^{+1.9}_{-0.7}\ keV}$ and with

218: $N_{\rm H}<0.3\times {\rm 10^{20}\ cm^{-2}}$ and ${\rm <1.1\times

219: 10^{20}\ cm^{-2}}$, respectively. The fits were improved significantly

220: when we adopted a partially covering absorber with $\chi^2$ (d.o.f.) of

221: 404.3 (440) and 259.1 (240).  The temperature of the

222: thermal bremsstrahlung in quiescence and outburst is ${\rm

223: 4.1^{+1.4}_{-0.9}\ keV}$ and ${\rm 1.6^{+1.5}_{-0.9}\ keV}$, which is

224: covered with an absorber with $N_H=3.9^{+1.4}_{-0.9}\times 10^{19}\ {\rm

225: cm^{-2}}$ and $4.9^{+1.5}_{-0.9}\times 10^{19}\ {\rm cm^{-2}}$ by ${\rm

226: 57^{+10}_{-9}}$\% and ${\rm 84^{+5}_{-7}}$\%, respectively. Below we adopt the latter as the continuum model.

227: 

228: \subsection{Emission lines

229: \label{sec:EvofLineParam}}

230: %

231: Given the continuum model, we retrieved all the energy channels and

232: tried to evaluate the line parameters. We fixed all the continuum

233: parameters at the best-fit values in \S~\ref{sec:continuum}, except for

234: the normalization, and added Gaussians to represent the H-like and

235: He-like {\ka} lines of O, Ne, Mg, and Si.  In fitting these Gaussians to

236: the data, all the parameters were allowed to vary.

237: Table~\ref{table:IonizationTemperature} summarizes the best-fit

238: parameters of the {\ka} lines thus obtained in quiescence and outburst,

239: together with the ionization temperature obtained from the He-like and

240: H-like line intensity ratio, which are plotted in

241: Fig.~\ref{fig:IonizationTemperature}.

242: 

243: It is clear that the obtained ionization temperatures of all ions in

244: quiescence are systematically higher than those in outburst. They show a

245: wide range depending on the element, from 0.3 to 2.0 keV in quiescence

246: and from 0.2 to 1.2 keV in outburst. The emission lines seen in the

247: 0.7--0.9$\ {\rm keV}$ bands are from L-shell Fe (mainly Ne-like),

248: indicating the existence of a plasma component with a temperature of

249: $\sim4\times 10^{6}\ {\rm K}$ \citep{1992ApJ...398..394A}. These facts

250: unambiguously indicate that the plasma in the boundary layer has a

251: significant temperature distribution. The differences in the line

252: intensity ratio distribution, as well as in the line profiles, probably

253: reflect a significant structural difference of the boundary layer

254: between quiescence and outburst.

255: 

256: 

257: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

258: \subsection{H-like line profile in outburst

259: \label{sec:HlineProfile}}

260: %

261: Unlike the He-like {\ka} line, which is composed of the resonance,

262: intercombination, and forbidden lines, as well as satellite lines, the

263: intrinsic structure of the H-like {\ka} line is much simpler,

264: consisting only of the following two resonance lines: 

265: $^2{\rm P}_{1/2}\rightarrow ^2\!{\rm S}_{1/2}$

266: and $^2{\rm P}_{3/2}\rightarrow ^2\!{\rm S}_{1/2}$.

267: Moreover, their energy separation is only 2.0~eV for Si ($E=2003.9$~eV

268: and 2005.9~eV, respectively)

269: and is even smaller for the lighter elements.

270: We can thus regard the H-like {\ka} lines of O, Ne, Mg, and Si

271: as consisting of a single component for the HETG resolution,

272: and their profile can be used to diagnose the geometry

273: and the dynamical motion of the line-emitting plasma.

274: 

275: From Table~\ref{table:IonizationTemperature}, it is seen that all the

276: H-like {\ka} lines are significantly broader in outburst than in

277: quiescence (see also Fig.~\ref{fig:Spectrum}). Their widths are in the

278: range 4--14$\ {\rm eV}$ in Gaussian $\sigma$, and corresponding

279: velocities are 1800--2300$\ {\rm km\ s^{-1}}$. Their central energies

280: are consistent with those in the laboratory.  Their profiles, fitted

281: with a broad Gaussian, are shown in Fig.~\ref{Fig:singlegauss}.  We then

282: fixed $E_{\rm c}$ at the laboratory values, which are defined as the

283: emissivity-weighted means of the Ly$\alpha$ doublet.

284: 

285: The abscissa axis covers an energy range of 10\% of each line's

286: rest-frame energy.  The best-fit parameters fitted with the broad

287: Gaussian are summarized in table~\ref{table:singlegauss}.  Although all

288: the fits are already acceptable at the 90\% confidence level, the

289: profile seems to change from a Gaussian for lighter elements to a

290: somewhat rectangular structure for heavier elements.

291: 

292: The rectangular line profile is reminiscent

293: of a homogeneously emitting thin spherical shell

294: that is radially in- or outflowing \citep{2002A&A...392..991H}.

295: Of these, an outflow associated with SS~Cyg in outburst

296: is detected as the disk wind by the {\it Chandra} LETG

297: \citep{2004ApJ...610..422M}.

298: The wind is detected as blueshifted absorption lines

299: with a velocity of 2500$\ {\rm km\ s^{-1}}$ due to intermediately ionized

300: O through Fe, overlaid on a blackbody emission with a temperature of

301: $250 \pm 50$kK. Based on this fact, we infer that the observed H-like

302: emission lines would also have been blueshifted, because the plasma below

303: the accretion disk is invisible to us. We thus consider the radial

304: outflow picture unlikely.

305: 

306: Compared with the outflow considered above, a radial inflow is more

307: likely if the innermost part of the boundary layer becomes optically

308: thin, as in quiescence.

309: The plasma in the boundary layer then should take the form of a cooling

310: flow for matter settling down onto the white dwarf.

311: The temperature distribution demonstrated in

312: \S~\ref{sec:EvofLineParam} is interpreted as a radial stratification in

313: the cooling flow.

314: The plasma emitting any given line can be treated as a geometrically

315: thin spherical shell, since the emissivity of any {\ka} line is so

316: sensitive to the plasma temperature.

317: 

318: In order to see whether the radial inflow interpretation is acceptable, we

319: have constructed a model representing a profile of an emission line

320: emanating from a geometrically thin uniform shell

321: falling onto a white dwarf.

322: The free parameters of the model are the radial velocity $v$,

323: the radius of the shell $r_{\rm sh}$,

324: the opening polar angle of the shell $\theta_{\rm o}$,

325: the inclination angle of the orbital plane $i$, and the

326: central energy of the emission line $E_{\rm c}$ (see

327: Fig.~\ref{fig:trapezmodel}a).

328: 

329: The white dwarf always hides the most blueshifted part of the shell,

330: and the hidden fraction is determined solely by $r_{\rm sh}/R_{\rm WD}$,

331: where $R_{\rm WD}$ is the white dwarf radius.

332: In the case that the shell does not expand

333: above the orbital plane enough to form a complete sphere,

334: we introduce the polar opening angle $\theta_{\rm o}$ 

335: within which the plasma is absent. 

336: Some examples of the resultant line profiles are shown 

337: in Fig.~\ref{fig:trapezmodel}(b).

338: Hereafter we refer to this model as the stratified shell model.

339: 

340: In fitting the model to the H-like {\ka} lines, we fixed $i$ at

341: 40$^\circ$, following the measurement of

342: \citet{1983PhDT........14S}. 

343: We also fixed $E_{\rm c}$ at the laboratory values. 

344: The best-fit parameters

345: of the lines with the stratified shell model are summarized in

346: table~\ref{table:trapez} and shown in Fig.~\ref{Fig:manyfits}.

347: As expected, the fitting to the H-like Si {\ka} line

348: is improved significantly, because the model can represent

349: the shoulders of the profile appropriately. The fitting to Mg is,

350: however, not improved significantly and that to Ne and O became poorer

351: than the Gaussian fit.

352: The radial velocity $v$ of all the elements is in the range

353: 3400--4500~km~s$^{-1}$ and does not show any sign of slowing down

354: from the Si shell (higher $T$) to the O shell (lower $T$).

355: 

356: Finally, we have tried the {\tt diskline} model in XSPEC

357: \citep{1996ASPC..101...17A} to see whether the profile

358: of the H-like {\ka} lines can be explained

359: by a bulk azimuthal motion of the plasma around the white dwarf.

360: The {\tt diskline} model is intended

361: to represent the emission line profile

362: from a geometrically thin Keplerian disk irradiated

363: by the central active galactic nucleus (AGN) via fluorescence, 

364: taking the gravitational redshift into account

365: \citep{1990Natur.344..747S, 1989MNRAS.238..729F}.

366: Hence, the parameters are the innermost radius of the disk

367: $R_{\rm in}$ in units of the Schwarzschild radius and the radial power $q$

368: of the emissivity $\varepsilon (r)$

369: in the form $\varepsilon \propto r^q$.

370: The results of the fits are

371: summarized in table~\ref{table:diskline} and displayed in

372: Fig.~\ref{Fig:manyfits}.

373: The resultant $\chi^2_\nu$ values of the fits to the O through Si lines

374: with the models introduced so far are shown in Fig.~\ref{Fig:chisq}.

375: 

376: The {\tt diskline} model can provide a comprehensive understanding

377: to a variety of the H-like {\ka} line profiles from O (Gaussian)

378: to Si (rectangular).

379: On the basis of this result, we conclude that the profiles of the H-like

380: {\ka} emission lines are best interpreted as due to azimuthal

381: bulk motion of the optically thin thermal plasma.

382: 

383: \subsection{Bulk motion of the plasma in outburst

384: \label{sec:PlaMotionOtb}}

385: 

386: As demonstrated in \S~\ref{sec:HlineProfile}, the nonrelativistic {\tt

387: diskline} model is successfully fitted to the H-like {\ka} lines from O

388: to Si in outburst.  This implies that the plasma has an azimuthal bulk

389: motion.  In order to evaluate this bulk motion, we hereafter fit the

390: profiles of the H-like {\ka} lines with two Gaussians that represent the

391: red- and blueshifted line components associated, respectively, with the

392: receding and approaching parts of the plasma.  The free parameters are

393: the average energy of the two line components $E_{\rm c}$, the energy

394: separation between them $E_{\rm sp}$, the energy widths $\sigma$, and

395: the normalizations.  Among these, $E_{\rm sp}$ is a measure of the

396: absolute radial location of the annulus from which the corresponding

397: H-like {\ka} line emanates, and $\sigma$ represents the geometrical

398: width of the annulus.  For simplicity, we set the $\sigma$ values, and

399: the normalizations of the red- and blueshifted components are

400: constrained to be common.

401: 

402: The results of the fits are summarized in

403: Table~\ref{table:doublegauss} and shown in Fig.~\ref{Fig:manyfits} and

404: \ref{Fig:doublegauss-results}, with $\sigma$ and $E_{\rm sp}$ being

405: normalized by $E_{\rm c}$.

406: In addition, we draw confidence contours

407: between $E_{\rm sp}$ and $\sigma$

408: in Fig.~\ref{Fig:doublegauss-correlation},

409: since they may couple with each other.

410: 

411: From Fig.~\ref{Fig:doublegauss-results} and

412: \ref{Fig:doublegauss-correlation}, $E_{\rm sp}$ is found to be smaller for

413: lighter elements, and the 90\% confidence contours of O and Si are

414: separated, while $\sigma$ does not show any difference among the

415: elements.  In Table~\ref{table:doublegauss}, the rotational velocity $v$

416: calculated from $E_{\rm sp}$ with $v = (E_{\rm

417: sp}/2E_{\rm c})\,c$ is also listed.  The rotational velocity of Si is

418: $\sim$1500~km~s$^{-1}$, whereas that of O is only $\sim$10~km~s$^{-1}$.

419: 

420: 

421: \section{Discussion\label{sec:Discussion}}

422: 

423: \subsection{Geometry of line-emitting plasma in outburst}

424: 

425: In \S~\ref{sec:HlineProfile} and \S~\ref{sec:PlaMotionOtb}, we have

426: demonstrated that the profiles of H-like {\ka} lines from O to Si can be

427: explained with the azimuthal bulk motion model (the {\tt

428: diskline} model), rather than the radial inflow model (the stratified

429: shell model) on the basis of fit statistics, and we have obtained the

430: bulk motion parameters with the double-Gaussian model.

431: Besides the statistics, the high radial velocities of

432: $v=$~3400--4500~km~s$^{-1}$ (table~\ref{table:trapez}) are problematic

433: for the stratified shell model.

434: Since the ionization temperature of O in outburst is only $\sim$200~eV,

435: its thermal velocity is as small as

436: $\langle v_{\rm th}\rangle =\sqrt{3kT/m_{\rm O}} \simeq$350~km~s$^{-1}$.

437: Hence the plasma flow is highly supersonic, thereby forming

438: a standing shock wave whose temperature

439: $kT_{\rm s} = (3/16)\mu m_{\rm H} v^2$ is as high as 14--24~keV.

440: However, no such high temperature component is detected.

441: The {\it Suzaku} observations in outburst, for example,

442: constrain the maximum temperature of the plasma to be

443: $kT_{\rm max}{\rm =8.4\pm 1.0~keV}$ \citep{2007PThPS.169..178I}.

444: 

445: These results indicate that the bulk motion in the optically thin

446: thermal plasma in outburst is azimuthal rather than radial.

447: Although the errors are somewhat large, the lighter

448: elements are likely to have smaller $E_{\rm sp}$

449: (Fig.~\ref{Fig:doublegauss-results} and

450: \ref{Fig:doublegauss-correlation}). It is natural to expect that both

451: the temperature and the azimuthal bulk velocity are gradually reduced as

452: the plasma approaches the white dwarf, because it eventually settles

453: down onto the white dwarf. The double-Gaussian model is consistent with

454: this expectation in that the lighter elements, which have lower

455: ionization temperature, have smaller $E_{\rm sp}$.

456: 

457: Since the boundary layer is believed to be optically thick in outburst,

458: it has been unclear where the observed optically thin thermal X-ray

459: emission emanates.

460: Our analysis indicates that the optically thin thermal plasma also shows a

461: bulk azimuthal motion whose velocity decreases as the plasma cools and

462: approaches the white dwarf.

463: We infer that the optically thin

464: thermal plasma exists over the optically thick disk like an accretion

465: disk corona in low-mass X-ray binaries, as sketched in

466: \citet{1985ApJ...292..535P}.

467: 

468: 

469: \subsection{Line emission site in quiescence}

470: %

471: In \S~\ref{sec:HETGspecQandO} we demonstrated that the emission lines

472: in the quiescent spectra (Fig.~\ref{fig:Spectrum}) 

473: are significantly narrower than those in outburst.

474: Nevertheless, the lines still have finite widths

475: in the range 1--3~eV in Gaussian $\sigma$ for the H-like {\ka} lines

476: (Table~\ref{table:IonizationTemperature}).

477: They are as large as 1 part per $\sim\!\!10^3$ of the line energies

478: $E_{\rm c}$, and if they are attributed to thermal motion, then the

479: following holds:

480: \[

481: \frac{\sigma}{E_{\rm c}} \;=\;

482:  \frac{1}{c}\sqrt{\frac13 \langle v_{\rm th}^2 \rangle}

483:  \;=\; \frac{1}{c}\sqrt{\frac{kT}{m}}

484: \]

485: where ${\langle v_{\rm th}^2 \rangle}^{1/2}$ is the root mean square

486: of the thermal velocity. Using the atomic mass number $A$, we obtain

487: \begin{equation}

488: kT \;=\; mc^2\,\left( \frac{\sigma}{E_{\rm c}}\right)^2

489:  \;=\; 75\,\left( \frac{A}{20} \right)

490:  \,\left( \frac{\sigma}{\mbox{2 eV}}\right)^2

491:  \,\left( \frac{E_{\rm c}}{\mbox{1 keV}}\right)^{-2}\;\;\mbox{keV},

492: \end{equation}

493: which is 64~keV for O and 51~keV for Si.

494: These temperatures are incompatible with the detection

495: of the H-like {\ka} emission lines,

496: because they are emitted from a plasma with a temperature of 

497: $\lesssim 1$~keV \citep{1985A&AS...62..197M}.

498: Hence, the line broadening should originate from a bulk motion whose geometry is so symmetric that the line

499: Doppler shift is observed as a line broadening, not as a line shift.

500: One may consider an outflow such as is observed in

501: the EUV to soft X-ray band in outburst \citep{2004ApJ...610..422M}.

502: However, the outflow is probably located farther out of

503: the optically thin boundary layer, and most of the receding parts of the

504: plasma are hidden by the accretion disk.

505: In fact, the soft X-ray absorption lines originating from the outflow

506: in outburst are blueshifted \citep{2004ApJ...610..422M}.

507: 

508: We consider the observed {\ka} emission lines from O to Si as emanating

509: from a transition region from the optically thick accretion disk to the

510: optically thin boundary layer,

511: where the Keplerian motion is gradually converted into thermal motion

512: due to strong friction.

513: The average line-of-sight velocity dispersion

514: estimated from the observation is

515: \begin{equation}

516:  \langle v \rangle \;=\; \frac{\sigma}{E_{\rm c}}\,c

517:  \;=\; 6.0\times 10^2\,\left( \frac{\sigma}{\mbox{2 eV}}\right)\,

518:  \left( \frac{E_{\rm c}}{\mbox{1 keV}} \right)^{-1}\;\;{\rm km\;s^{-1}},

519: \label{eq:ave_vel}

520: \end{equation}

521: which is 620$^{+130}_{-170}$,

522: 530$^{+100}_{-120}$, $460^{+220}_{-120}$,

523: and 420$\pm 140$~km~s$^{-1}$

524: for O, Ne, Mg, and Si, respectively.

525: This is roughly one-tenth of the Keplerian velocity on the surface of

526: the 1.2$M_\odot$ white dwarf ($\simeq$6300~km~s$^{-1}$).

527: Even considering the inclination angle

528: of 40$^\circ$

529: and the line-of-sight average of the azimuthal velocity,

530: $(2/\pi)\sin 40^\circ$,

531: we obtain the maximum possible line-of-sight velocity to be

532: $\langle v \rangle_{\rm max} = $2600~km~s$^{-1}$, 

533: which can cover the average velocity eq.~(\ref{eq:ave_vel})

534: evaluated from the observed line width.

535: 

536: As in outburst, the line width in quiescence can be attributed 

537: to azimuthal motion. Unlike in outburst, however, where the emission

538: lines can be interpreted as being emitted from a cooling plasma, we

539: believe those in quiescence are created in the heating process at the

540: entrance to the optically thin boundary layer.  In fact, although the

541: statistical significance is not very high, the bulk velocity in

542: quiescence estimated from the H-like {\ka} line widths seems to decrease

543: from O to Si.  It is likely that the plasma heating results

544: from the conversion of the energy of bulk motion.

545: 

546: 

547: \subsection{Temperature distribution }

548: %

549: In this section, we attempt to impose a constraint on the temperature

550: distribution of the plasma in outburst and quiescence in terms of the

551: intensity ratio of the He-like and H-like {\ka} emission lines.  We

552: simply assume that the differential emission measure $d(E\!M)$ is

553: proportional to the power of the plasma temperature.  In the following

554: analysis, we adopt the spectral model {\tt cemekl}, in which the

555: temperature distribution is taken into account as $d(E\!M)\propto

556: (T/T_{\rm max})^\alpha d(\log T)$ $\propto (T/T_{\rm

557: max})^{\alpha-1}dT$, where $T_{\rm max}$ is the maximum temperature of

558: the plasma.  Hence, the exponent $\alpha$ obtained by a fit with this model

559: is larger than the true weighting factor by 1

560: \citep{1997MNRAS.288..649D}.  In the framework of this model, the line

561: intensity ratio is uniquely determined once $\alpha$ and $T_{\rm max}$

562: are given.  Hence, we can conversely give a constraint on these two

563: parameters with the line intensity ratios of O, Ne, Mg, and Si that we have

564: obtained from the HETG data.  The allowed parameter regions delineated

565: by the 90\% errors of the intensity ratios

566: (Table~\ref{table:IonizationTemperature}) are shown in

567: Fig.~\ref{Fig:lineratio-coutour} with a different color for each element.

568: 

569: For all the elements, the contours become vertical, and $\alpha$

570: diverges as $T_{\rm max}$ decreases and approaches

571: a certain value.

572: This limiting $T_{\rm max}$ gives the temperature of the single-phase

573: plasma that can reproduce the observed intensity ratio.

574: On the contrary, the contour becomes horizontal

575: in the high-$T_{\rm max}$ limit.

576: Since the emissivity of atomic lines peaks at a finite temperature,

577: the line ratio becomes insensitive to $T_{\rm max}$

578: and is determined solely by $\alpha$ if $T_{\rm max}$

579: becomes too high compared with the emissivity peak temperature.

580: As shown in Fig.~\ref{Fig:lineratio-coutour},

581: there is a parameter region that can be shared

582: by all elements (except for O in outburst).

583: Taking into account the constraint $T_{\rm max} = 8.4\pm 1.0$~keV

584: for the outburst obtained from the {\it Suzaku} observation

585: \citep{2007PThPS.169..178I}, the common range of $\alpha$ 

586: is $-0.2$ to 0.3 in outburst.

587: On the other hand, the maximum temperature is $15.4^{+8.8}_{-5.4}$~keV

588: \citep{1997MNRAS.288..649D} in quiescence,

589: which is significantly higher than in outburst,

590: and the common range of $\alpha$ is found to be 0.6--1.2

591: in quiescence.

592: This significant difference is probably the result

593: of the geometrical difference of the optically thin boundary layer

594: between quiescence and outburst (for example,

595: see \citet{1985ApJ...292..535P}).

596: 

597: We remark that only the index $\alpha$ of O in outburst is not common with

598: the other elements. 

599: This is not unreasonable, in view of the shape of the cooling curve of

600: an optically thin thermal plasma.

601: Unlike the temperature region $T \gtrsim 2\times 10^7$~K,

602: the slope of the plasma cooling curve shows a complicated change as a

603: function of the temperature in the range $T \lesssim 10^7$~K

604: \citep{1993ApJ...418L..25G}.

605: Since $\alpha$ of a given element reflects the slope of the cooling

606: curve around the temperature at which the line emissivity of the

607: element shows the peak,

608: it is possible in principle that $\alpha$ is different for elements

609: whose emissivity peak appears in the range $T \lesssim 10^7$~K.

610: 

611: The index of the temperature distribution $\alpha$ was systematically

612: examined by \citet{2005MNRAS.357..626B} using the complete set of the

613: {\it ASCA} nonmagnetic CV data.

614: They found that $\alpha$ for the dwarf novae 

615: is in the range from 0.7 to 1.8 in quiescence (with a few exceptions)

616: and $-$0.1 to 0.3 in outburst.

617: The $\alpha$-values obtained from our analysis on SS~Cyg in quiescence

618: and outburst are both consistent with their results.

619: \citet{1997MNRAS.288..649D} carried out the same analysis using the

620: outburst and quiescence data of {\it Ginga} and {\it ASCA}.

621: Although $\alpha$ in the quiescence {\it Ginga} data is not

622: constrained very well ($2.9^{+\infty}_{-1.4}$),

623: they derived $\alpha = 0.43^{+0.41}_{-0.38}$ and $0.46^{+0.06}_{-0.05}$

624: for the normal and anomalous outbursts from the {\it Ginga} and {\it ASCA}

625: data, respectively.

626: The $\alpha$ of the anomalous outburst is outside the range of

627: \citet{2005MNRAS.357..626B} and that of ours.

628: The exponent $\alpha$ may possibly be different between the two types of

629: the outburst and may depend on the intensity of the reflection

630: component included in their modeling.

631: 

632: The temperature distribution in the postshock accretion column of a

633: magnetic cataclysmic variables (mCVs) can be approximated as $T(z)

634: \propto z^{2/5}$, where $z$ is the height from the white

635: dwarf surface if the cooling mechanism is dominated by free-free

636: emission \citep{2002apa..book.....F}.  In this case, the {\tt cevmekl}

637: power $\alpha \simeq +0.5$ \citep{1994MNRAS.266..367I},

638: which lies between that in the quiescence and the

639: outburst obtained from this work.

640: 

641: 

642: 

643: \section{Conclusion

644: \label{sec:Conclusion}}

645: %

646: We have investigated the optically thin boundary layer of SS~Cyg both in

647: quiescence and in outburst by carrying out high resolution line spectroscopy

648: with the {\it Chandra} HETG.

649: The spectra contain He-like and H-like {\ka} emission lines from O

650: to Fe, as well as L-shell emission lines from Fe,

651: both in quiescence and outburst.

652: However, the relative intensity of the He-like and H-like emission lines

653: of a given element is significantly different; in outburst they are

654: nearly equal, whereas in quiescence the H-like lines are much more intense

655: than the He-like line.

656: The ionization temperatures evaluated from the line

657: intensity ratio of O, Mg, Ne, and Si are in the range 0.3--2.0~keV and

658: 0.2--1.2~keV in quiescence and outburst, respectively.

659: 

660: The other remarkable difference of the line profiles between quiescence

661: and outburst is in the line width.  The H-like lines from O to Si in

662: outburst are much broader, with widths of 4--14~eV in

663: Gaussian $\sigma$ (1800--2300~km~s$^{-1}$), and especially that of Si has

664: a rectangular profile, rather than a Gaussian profile.  Among a few

665: possibilities, such as a homogeneously emitting thin spherical shell

666: that is radially in- or outflowing, an azimuthal bulk motion, which

667: can be mimicked by the {\tt diskline} model, most naturally explains the

668: observed line profile.  The boundary layer is, however, believed to be

669: optically thick all the way down to the white dwarf surface.  We thus

670: infer that the line-emitting plasma is located above the optically thick

671: disk, like an accretion disk corona commonly seen in low-mass X-ray

672: binaries.  Since lighter elements tend to have a smaller bulk velocity,

673: the observed line-emitting plasma in outburst is a cooling plasma.

674: 

675: Although less remarkable, the emission lines from O to Si in quiescence

676: also have a finite line width with a Gaussian $\sigma$ of 1--3~eV

677: (420--620~km~s$^{-1}$).  It is impossible to attribute the width to a

678: thermal motion, because the resultant temperature becomes as high as

679: $\gtrsim$50~keV.  Since no red- or blueshift is detected for the line

680: central energies, the radial in- or outflow is unlikely.  The most

681: natural origin of the bulk velocity is an azimuthal motion.  A slightly

682: larger velocity for lighter elements, contrary to outburst, suggests

683: that the lines in quiescence are emitted in the heating process at the

684: entrance to the boundary layer, where the bulk motion of the optically

685: thick accretion disk is converted to thermal energy due to friction.

686: 

687: We have also investigated the temperature distribution in the boundary

688: layer within a framework of the differential emission measure having a

689: power-law distribution of the temperature as $(T/T_{\rm max})^\alpha$.

690: From the line intensity ratio,

691: the index $\alpha$ is constrained in the range from $-$0.2 to 0.3 in

692: outburst and from 0.6 to 1.2 in quiescence. These values are

693: consistent with those obtained from the previous systematic work

694: of \citet{2005MNRAS.357..626B}. 

695: 

696: 

697: 

698: 

699: \begin{thebibliography}{10}

700: %

701: \bibitem[Arnaud(1996)]{1996ASPC..101...17A} Arnaud, K.~A.\ 1996, ASP 

702: Conf.~Ser.~101: Astronomical Data Analysis Software and Systems V, 101,17 

703: 

704: \bibitem[Arnaud \& Raymond(1992)]{1992ApJ...398..394A} Arnaud, M., \& 

705: Raymond, J.\ 1992, \apj, 398, 394 

706: 

707: \bibitem[Baskill et al.(2005)]{2005MNRAS.357..626B} Baskill, D.~S., 

708: Wheatley, P.~J., \& Osborne, J.~P.\ 2005, \mnras, 357, 626 

709: 

710: \bibitem[Canizares et al.(2005)]{2005PASP..117.1144C} Canizares, C.~R., et 

711: al.\ 2005, \pasp, 117, 1144 

712: 

713: \bibitem[Cannizzo(1993)]{1993ApJ...419..318C} Cannizzo, J.~K.\ 1993, \apj, 

714: 419, 318 

715: 

716: \bibitem[Done \& Osborne(1997)]{1997MNRAS.288..649D} Done, C., \& Osborne, 

717: J.~P.\ 1997, \mnras, 288, 649 

718: 

719: \bibitem[Fabian et al.(1989)]{1989MNRAS.238..729F} Fabian, A.~C., Rees, 

720: M.~J., Stella, L., \& White, N.~E.\ 1989, \mnras, 238, 729 

721: 

722: \bibitem[Frank et al.(2002)]{2002apa..book.....F} Frank, J., King, A., \& 

723: Raine, D.~J.\ 2002, Accretion Power in Astrophysics, by Juhan Frank and 

724: Andrew King and Derek Raine, pp.~398.~ISBN 0521620538.~Cambridge, UK: 

725: Cambridge University Press, February 2002.,  

726: 

727: \bibitem[Gehrels \& Williams(1993)]{1993ApJ...418L..25G} Gehrels, N., \& 

728: Williams, E.~D.\ 1993, \apjl, 418, L25 

729: 

730: \bibitem[Hynes et al.(2002)]{2002A&A...392..991H} Hynes, R.~I., et al.\ 

731: 2002, \aap, 392, 991 

732: 

733: \bibitem[Ishida et al.(2007)]{2007PThPS.169..178I} Ishida, M., Okada, S., 

734: Nakamura, R., Terada, Y., Hayashi, T., Mukai, K., \& Hamaguchi, K.\ 2007, 

735: Progress of Theoretical Physics Supplement, 169, 178 

736: 

737: \bibitem[Ishida et al.(1994)]{1994MNRAS.266..367I} Ishida, M., Makishima, 

738: K., Mukai, K., \& Masai, K.\ 1994, \mnras, 266, 367 

739: 

740: \bibitem[Jones \& Watson(1992)]{1992MNRAS.257..633J} Jones, M.~H., \& 

741: Watson, M.~G.\ 1992, \mnras, 257, 633 

742: 

743: \bibitem[Lasota(2001)]{2001NewAR..45..449L} Lasota, J.-P.\ 2001, New 

744: Astronomy Review, 45, 449 

745: 

746: \bibitem[Mauche et al.(2005)]{2005ASPC..330..355M} Mauche, C.~W., Wheatley, 

747: P.~J., Long, K.~S., Raymond, J.~C., \& Szkody, P.\ 2005, ASP 

748: Conf.~Ser.~330: The Astrophysics of Cataclysmic Variables and Related 

749: Objects , 330, 355 

750: 

751: \bibitem[Mauche(2004)]{2004ApJ...610..422M} Mauche, C.~W.\ 2004, \apj, 610, 

752: 422 

753: 

754: \bibitem[Mewe et al.(1985)]{1985A&AS...62..197M} Mewe, R., Gronenschild, 

755: E.~H.~B.~M., \& van den Oord, G.~H.~J.\ 1985, \aaps, 62, 197 

756: 

757: \bibitem[Mukai et al.(2003)]{2003ApJ...586L..77M} Mukai, K., Kinkhabwala, 

758: A., Peterson, J.~R., Kahn, S.~M., \& Paerels, F.\ 2003, \apjl, 586, L77 

759: 

760: \bibitem[Nousek et al.(1994)]{1994ApJ...436L..19N} Nousek, J.~A., Baluta, 

761: C.~J., Corbet, R.~H.~D., Mukai, K., Osborne, J.~P., \& Ishida, M.\ 1994, 

762: \apjl, 436, L19 

763: 

764: \bibitem[Osaki(1996)]{1996PASP..108...39O} Osaki, Y.\ 1996, \pasp, 108, 39 

765: 

766: \bibitem[Patterson \& Raymond(1985)]{1985ApJ...292..535P} Patterson, J., \& 

767: Raymond, J.~C.\ 1985, \apj, 292, 535 

768: 

769: \bibitem[Pringle \& Savonije(1979)]{1979MNRAS.187..777P} Pringle, J.~E., \& 

770: Savonije, G.~J.\ 1979, \mnras, 187, 777 

771: 

772: \bibitem[Pringle(1977)]{1977MNRAS.178..195P} Pringle, J.~E.\ 1977, \mnras, 

773: 178, 195 

774: 

775: \bibitem[Rana et al.(2006)]{2006ApJ...642.1042R} Rana, V.~R., Singh, K.~P., 

776: Schlegel, E.~M., \& Barrett, P.~E.\ 2006, \apj, 642, 1042 

777: 

778: \bibitem[Shafter(1983)]{1983PhDT........14S} Shafter, A.~W.\ 1983, 

779: Ph.D.~Thesis,

780: 

781: \bibitem[Smak(1984)]{1984PASP...96....5S} Smak, J.\ 1984, \pasp, 96, 5 

782: 

783: 

784: \bibitem[Stella(1990)]{1990Natur.344..747S} Stella, L.\ 1990, \nat, 344, 

785: 747 

786: 

787: \bibitem[Tylenda(1981)]{1981AcA....31..267T} Tylenda, R.\ 1981, Acta 

788: Astronomica, 31, 267 

789: 

790: \bibitem[Wheatley et al.(2003)]{2003MNRAS.345...49W} Wheatley, P.~J., 

791: Mauche, C.~W., \& Mattei, J.~A.\ 2003, \mnras, 345, 49 

792: 

793: \bibitem[Yoshida et al.(1992)]{1992PASJ...44..537Y} Yoshida, K., Inoue, H., 

794: \& Osaki, Y.\ 1992, \pasj, 44, 537 

795: 

796: \end{thebibliography}

797: 

798: \include{table}

799: \include{figure}

800: 

801: \end{document}

802: 

803: