1: %\documentclass{emulateapj}

2: %\documentclass{aastex}

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

4: \shortauthors{K. Itoh, M. Ishida, and H. Kunieda}

5: \shorttitle{Density Diagnostics of the Plasma in AE Aqr}

6: 

7: \newcommand{{\aaq}}{AE~Aqr}

8: \newcommand{{\ka}}{K$\alpha$}

9: \newcommand{{\ha}}{H$\alpha$}

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

11: 

12: \begin{document}

13: 

14: \title{Density Diagnostics of the Hot Plasma in AE Aquarii with XMM-Newton}

15: 

16: \author{Kei~{\sc Itoh}\altaffilmark{1}, Manabu~{\sc Ishida}\altaffilmark{2}, 

17: and Hideyo~{\sc Kunieda}\altaffilmark{1}}

18: \email{keito@astro.isas.jaxa.jp, ishida@phys.metro-u.ac.jp,

19: kunieda@astro.isas.jaxa.jp}

20: 

21: % Notice that each of these authors has alternate affiliations, which

22: % are identified by the \altaffilmark after each name.  The actual alternate

23: % affiliation information is typeset in footnotes at the bottom of the

24: % first page, and the text itself is specified in \altaffiltext commands.

25: % There is a separate \altaffiltext for each alternate affiliation

26: % indicated above.

27: 

28: \altaffiltext{1}{Institute of Space and Astronautical

29: Science, 3-1-1 Yoshinodai, Sagamihara, Kanagawa 229-8510, Japan}

30: \altaffiltext{2}{Department of Physics, Tokyo Metropolitan

31: University, 1-1 Minami-Osawa, Hachioji, Tokyo 192-0397, Japan}

32: 

33: \begin{abstract}

34: High resolution spectroscopy of AE~Aquarii with the {\xmm} RGS

35: has enabled us to measure the electron number density

36: of the X-ray-emitting hot plasma to be $\sim10^{11}$~cm$^{-3}$

37: by means of intensity ratios of the He-like triplet

38: of Nitrogen and Oxygen.

39: Incorporating with the emission measure evaluated by the EPIC cameras,

40: we have also found a linear scale of the plasma 

41: to be $\simeq 5\times 10^{10}$~cm.

42: Both these values, obtained model-independently, 

43: are incompatible with a standard post-shock accretion column 

44: of a magnetized white dwarf,

45: but are naturally interpreted as the plasma being formed

46: through interaction between an accretion flow and the magnetosphere

47: of the white dwarf.

48: Our results provide another piece of evidence 

49: of the magnetic propeller effect being at work in {\aaq}.

50: 

51: \end{abstract}

52: 

53: \keywords{binaries: close --- novae, cataclysmic variables --- stars:

54: individual (AE Aquarii) --- plasmas --- X-rays: stars}

55: %\keywords{globular clusters,peanut clusters,bosons,bozos}

56: 

57: \section{Introduction}

58: {\aaq} is a close binary system composed of a magnetized white dwarf 

59: rotating at a period of $33.08$~s \citep{1979ApJ...234..978P}

60: and a Roche-lobe filling K3IV secondary star orbiting at a period of 9.88~h

61: \citep{1993ApJ...410L..39W}.

62: The masses of the primary and the secondary are reported to be

63: $M_1 = 0.79\pm 0.16M_\odot$ and $M_2 = 0.50\pm 0.10M_\odot$

64: \citep{1996MNRAS.282..182C}, 

65: and the orbital semi-major axis length is $a = 1.8\times 10^{11}$~cm.

66: 

67: {\aaq} has been known as one of the most enigmatic 

68: magnetic Cataclysmic Variables (mCVs) on various aspects, 

69: including large optical flares and flickering \citep{1979ApJ...234..978P},

70: large radio flares \citep{1996ApJ...461.1016B},

71: TeV $\gamma$-ray emissions \citep{1994ApJ...434..292M} and so on.

72: Although the orbital period of this long predicts existence

73: of an accretion disk,

74: an optical spectrum of {\aaq} does not seem as such, 

75: but only shows a broad single-peaked {\ha} emission

76: whose centroid velocity was found to lag behind the white dwarf orbit

77: by some 70$^\circ$ \citep{1993ApJ...410L..39W}.

78: Hard X-ray emissions from magnetic CVs (mCVs) 

79: originate from the post-shock plasma

80: that takes place close to the white dwarf surface in the accretion column.

81: Although the plasma temperature is a few tens of keV in general

82: \citep{1999ApJS..120..277E,1995cava.conf...93I},

83: that of {\aaq} is measured to be as low as $\sim 3$~keV

84: \citep{1999mcv..work..343E,1999ApJ...525..399C}.

85: All these results, together with discovery of a steady spin-down of the

86: white dwarf at a rate $\dot P = 5.64\times 10^{-14}$~s~s$^{-1}$

87: \citep{1994MNRAS.267..577D}, lead \citet{1997MNRAS.286..436W} 

88: to draw a picture

89: that the accreting matter from the secondary trapped at the magnetosphere

90: does not accrete onto the white dwarf 

91: but is expelled from the binary owing to magnetic torque 

92: \citep{1998MNRAS.298..285W,2001A&A...374.1030I}.

93: This is so-called the magnetic propeller effect.

94: 

95: In this paper, we present new observational results

96: that strongly indicate the magnetic propeller being at work in {\aaq},

97: by means of a model-independent density diagnostics with the He-like triplet

98: of Nitrogen and Oxygen fully resolved by the {\xmm} RGS for the first time.

99: 

100: \section{Observations}

101: 

102: {\aaq} was observed with {\xmm} \citep{2001A&A...365L...1J}

103: on 2001 November 7--8.

104: The observation log is given in Table \ref{tbl-1}.

105: %

106: \placetable{tbl-1}

107: %

108: The entire observation consists of two parts.

109: During the former $\sim$10~ks, 

110: both the EPIC pn \citep{2001A&A...365L..18S} and 

111: MOS \citep{2001A&A...365L..27T} had been switched off.

112: The RGS \citep{2001A&A...365L...7D}, on the other hand, 

113: had been normally operated throughout the observation.

114: As a result, 27~ks data are available for the RGS and

115: 17~ks for the EPIC pn/MOS.

116: In Fig.~\ref{fig-1} shown are the light curves of pn, MOS1, and RGS1.

117: %

118: \placefigure{fig-1}

119: %

120: 

121: For the data reduction, we use the SAS version 6.0.0.

122: We adopt circular apertures of $54.\!\!''7$ and $50.\!\!''6$ in radius 

123: as the source photon integration region for pn and MOS, respectively.

124: The background photons are accumulated from concentric annuli,

125: being set out of the source regions, with three times larger outer radii.

126: Source photons are somewhat piled up in the pn data

127: especially during the flare. 

128: We thus have excluded the central circular region with a diameter of 16$''$

129: from the analysis of pn data.

130: 

131: \section{Data Analysis}

132: 

133: \subsection{The EPIC Spectra}

134: 

135: The time-averaged energy spectra extracted separately from the MOS and

136: pn data are shown in Fig.~\ref{fig-2}.

137: %

138: \placefigure{fig-2}

139: %

140: A number of K$\alpha$ emission lines of H-like

141: and/or He-like Ne, Mg, Si, S, and Fe can be recognized.

142: Coexistence of these lines indicate that the X-ray

143: emission originates from a multi-temperature optically thin thermal plasma

144: \citep{1999ApJ...525..399C,1999mcv..work..343E}.

145: We thus have adopted a multi-temperature {\sc vmekal} model 

146: \citep{1995Legacy.....6...16M} to fit these spectra.

147: We have added a new temperature component one by one until the addition

148: of a new component does not improve the fit significantly

149: on the basis of F-test.

150: To this end, we have arrived at a four temperature {\sc vmekal} model

151: with common elemental abundances undergoing photoelectric absorption

152: with a common hydrogen column density.

153: The best-fit parameters are listed in Table \ref{tbl-2}.

154: %

155: \placetable{tbl-2}

156: %

157: Emission measures of the four continuum components

158: with the temperatures of 0.14, 0.59, 1.4, and 4.6~keV are

159: $1.3$, $3.6$, $2.7$, and $5.3\times 10^{53}\mathrm{cm}^{-3}$,

160: and $1.29\times 10^{54}\mathrm{cm}^{-3}$ in total,

161: for the assumed distance of 100~pc \citep{1993ApJ...406..229W}.

162: The highest temperature $kT = 4.6$~keV is considerably lower 

163: than any other mCVs.

164: Also, the fluorescent neutral iron K$\alpha$ emission line at 6.40~keV

165: ubiquitous among mCVs \citep{1999ApJS..120..277E} is absent.

166: The upper limit of its equivalent width is obtained to be $< 88$~eV.

167: Thanks to a large effective area of {\xmm}, 

168: the abundances of the elements from N to Ni are obtained.

169: They are generally sub-solar \citep{1989GeCoA..53..197A}, except for N

170: which is more than three times the solar value.

171: 

172: \subsection{Line Intensities from RGS Spectra} 

173: 

174: In Fig.~\ref{fig-3} we have shown the RGS spectra.

175: %

176: \placefigure{fig-3}

177: %

178: The K$\alpha$ emission lines from Nitrogen through Silicon

179: in the H-like and/or He-like ionization states can easily be recognized.

180: Figure~\ref{fig-4} (a)-(c) are blow-up of energy bands of 

181: the He-like triplets of N, O, and Ne.

182: %

183: \placefigure{fig-4}

184: %

185: In the next column shown are the spectra predicted 

186: by the best-fit four temperature {\sc vmekal} model (table~\ref{tbl-2}),

187: being convolved by the RGS1 1st order response function.

188: Note that the {\sc vmekal} model represents emission spectrum

189: from a thermal plasma in the low density coronal limit.

190: One can easily see, for N and O, that the intensity of the

191: intercombination increases from the coronal limit

192: by consumption of the forbidden.

193: This behavior is interpreted as a high density effect;

194: if the electron density exceeds a certain critical value inherent in

195: each element, one of the two electrons excited to the upper level 

196: of the forbidden line ($^3\mathrm{S}_1$) is further pumped by another impact

197: of a free electron up to the higher level $^3\mathrm{P}_{2,\,1}$,

198: and is then relaxed by radiating the intercombination line.

199: The relative intensities of the intercombination and forbidden lines

200: can therefore be utilized as a density diagnostics

201: \citep{1969MNRAS.145..241G,1981ApJ...249..821P}.

202: 

203: In order to evaluate the electron number density $n_\mathrm{e}$,

204: we begin with evaluating intensities of the He-like triplets.

205: For this, we utilize the four temperature {\sc vmekal} model

206: that provides the best-fit to the EPIC pn and MOS spectra.

207: We have fixed the hydrogen column density and the four temperature 

208: at the values in table~\ref{tbl-2}

209: and the relative normalizations of the four continuum components.

210: The abundances are also fixed at the best-fit values, except for an element

211: to be used as a density diagnostic, 

212: for which the abundance is set equal to zero, and instead,

213: four Gaussians are added, 

214: representing the He-like triplet and the Ly$\alpha$ line.

215: The line intensities thus obtained for N, O, and Ne are summarized in

216: table~\ref{tbl-3}, and the best-fit results are displayed in Fig.~\ref{fig-4}

217: (a)-(c) as the histograms.

218: %

219: \placetable{tbl-3}

220: %

221: The ionization temperature $kT_i$

222: calculated from the intensity ratio between the Ly$\alpha$ and $r$

223: are obtained to be 0.16, 0.30, and 0.34~keV for N, O, and Ne,

224: respectively.

225: 

226: Given the line intensities of the triplets,

227: we have carried out density diagnostics by means of the intensity ratio

228: $f/(r+i)$.

229: In Fig.~\ref{fig-4} (g)-(i) shown are theoretical curves

230: of the ratio $f/(r+i)$ versus the electron density,

231: which is calculated by means of the plasma code {\sc spex} 

232: \citep{1996uxsa.coll..411K} at the ionization temperature $kT_i$

233: of each element.

234: In each panel we have also drawn a range of the intensity ratio

235: allowed from the data and the resultant density range as a box.

236: The electron densities are obtained

237: to be $1.4\times 10^{10}-1.3\times 10^{11}\mathrm{cm}^{-2}$,

238: $4.0\times 10^{10}-6.8\times 10^{11}\mathrm{cm}^{-2}$,

239: and $< 9.3\times 10^{12}\mathrm{cm}^{-2}$ for N, O, and Ne, respectively.

240: As a crude approximation, the electron number density

241: of the plasma is $n_{\rm e} \simeq 10^{11}{\rm cm}^{3}$.

242: 

243: \section{Discussion and Conclusion}

244: 

245: It is of great importance to note that the resultant density

246: $n_{\rm e} \simeq 10^{11}{\rm cm}^{3}$ is smaller 

247: by several orders of magnitude 

248: than the conventional estimate in the post-shock accretion column of mCVs: 

249: $n_{\rm e}\simeq 10^{16}{\rm cm}^{-3}$ \citep{2002apa..book.....F}.

250: Moreover, combined with the emission measure 

251: $EM = 1.3\times 10^{54}{\rm cm}^{-3}$ obtained with the EPIC pn/MOS (\S~3.1),

252: the linear scale of the plasma is evaluated to be 

253: $\ell_\mathrm{p} = (EM/n_{\rm e}^2)^{1/3} \simeq 5\times 10^{10}$~cm,

254: which is much larger than a radius of a typical white dwarf

255: and rather close to the orbital scale.

256: We thus conclude that the optically thin hot plasma in {\aaq} does not

257: accrete onto the white dwarf but rather spreads over the orbit of the binary.

258: Note that this conclusion is derived simply

259: on the basis of widely approved atomic physics, and hence, 

260: is free from any specific model.

261: 

262: It is known, however, that photo-excitation due to UV radiation can

263: also pump $^3\mathrm{S}_1$ electron up to $^3{\rm P}_{2,\,1}$,

264: thereby affecting the density diagnostics.

265: Although there is no evidence of accretion disk in {\aaq} 

266: \citep{1993ApJ...410L..39W} and the secondary star is a late type K3 subgiant,

267: the white dwarf can be a source of the photo-excitation, 

268: the emission spectrum of which is reported to be a blackbody

269: with a temperature of $T_{\rm r} = 26,000$~K \citep{1999mcv..work..357W}.

270: The photo-excitation rate of such a white dwarf 

271: from the initial state $i =\; ^3\mathrm{S}_1$

272: to the final state $j =\; ^3{\rm P}_{2,\,1}$ is

273: %

274: \begin{equation}

275: \Gamma_{ij}(T_{\rm r}) \;=\; \frac{\pi e^2}{m_{\rm e}c}\,f_{ij}\cdot W\cdot

276:     \frac{\pi B_\nu (T_{\rm r})}{h\nu}

277: \end{equation}

278: \citep{2001A&A...376.1113P,2002pcvr.conf..113M} 

279: where $f_{ij}$ is the effective oscillator strength of photo-excitation

280: which is 0.03574 \citep{1999A&AS..135..347N},

281: and $W$ is a geometrical dilution factor to be taken here as

282: $(R_{wd}/\ell_{\rm p})^2$ where $R_{wd}$ is the radius of the white dwarf.

283: On the other hand, 

284: the collisional excitation rate to be compared with $\Gamma_{ij}$ is

285: $n_{\rm e} q_{ij}$, where,

286: %

287: \begin{equation}

288: q_{ij}(T_{\rm e})\;=\;

289:   \frac{8.63\times 10^{-6}}{\omega_i\,T_{\rm e}^{1/2}}\,

290:     e^{-E_{ij}/kT_{\rm e}}\gamma_{ij}(T_{\rm e})\;\;\;{\rm [\,cm^3\;s^{-1}\,]}

291: \label{eq:colrate}

292: \end{equation}

293: is the rate coefficient of the electron impact excitation

294: \citep{1978A&A....65...99M}.

295: In this equation, $\omega_i (= 3)$ is the statistical weight of the

296: lower level, $E_{ij}$ is the energy difference between the two levels,

297: and $\gamma_{ij}$ is the Maxwellian-averaged collision strength

298: which is tabulated as a function of $T_{\rm e}$ in \citet{1981ApJ...246.1031P}.

299: Adopting $T_i$ listed in table~\ref{tbl-3} as $T_{\rm e}$,

300: we compared eqs.~(1) and (2) and obtained 

301: $\Gamma_{ij} \simeq 0.07\,n_{\rm e} q_{ij}$ for both N and O.

302: Hence, the photo-excitation effect can be neglected

303: in the first order approximation.

304: 

305: We remark that the density obtained in \S~3.2 should be treated 

306: as an upper limit, in the case that the UV radiation is stronger 

307: than the estimation above.

308: Even if so, the conclusion above need not to be changed because

309: a lower density, and hence, a larger geometrical scale are required

310: for the plasma.

311: 

312: The low density and the large scale of the plasma naturally lead us to invoke

313: the magnetic propeller model, in which blobby accreting matter 

314: originally following a ballistic trajectory from the inner Lagrangian point

315: is gradually penetrated by the magnetic field of

316: the white dwarf, and is finally blown out of the binary

317: due to interaction with the rapid rotating white dwarf magnetosphere

318: \citep{1997MNRAS.286..436W}.

319: The magnetic propeller model is advantageous in explaining 

320: various characteristics of {\aaq}, such as 

321: the spin down of the white dwarf, the velocity modulation of H$\alpha$ line.

322: The low plasma temperature compared with other mCVs

323: can be attributed to a halfway release of the gravitational potential energy

324: of accreting matter down to the magnetosphere.

325: We believe that

326: the density diagnostics presented here add another piece of

327: evidence to support the magnetic propeller effect being at work in {\aaq}.

328: 

329: \citet{1998MNRAS.298..285W} estimated that the velocity of the out-flowing

330: plasma can be as high as several hundred km at most.

331: If so, a Doppler shift of the N and O lines may be detected through

332: phase-resolved spectral analysis.

333: We thus have made RGS spectra of the first and second flares separately,

334: during the time interval of 3,000-8,500~s and 19,000-26,000~s 

335: in Fig.~\ref{fig-1}, respectively,

336: and evaluated the central energy of Oxygen Ly$\alpha$.

337: The result is negative with the line central energies of

338: $653.4\pm 0.5$~eV and $653.4\pm 0.4$~eV, respectively,

339: which are fully consistent with the laboratory value.

340: We have further split the data of the second flare into evenly segregated

341: three pieces, but the central energy distributes in the range

342: 653.2-653.4~eV.

343: Note, however, that we have obtained a finite value 

344: for a 1-$\sigma$ line width of $2.4\pm 0.6$~eV and $1.7\pm 0.6$~eV 

345: for the first and second flares, respectively, 

346: which are larger than the natural width ($\sim 0.001$~eV)

347: or the thermal velocity of Oxygen ion ($\sim 70$~km~s$^{-1}$ or

348: $\sim 0.2$~eV estimated from $kT_i = 0.3$~keV).

349: Since the ionization temperature of Oxygen is lower than the highest

350: temperature of the plasma (table~\ref{tbl-2}) by an order of magnitude, 

351: and also from the fact $\ell_{\rm p} \simeq 5\times 10^{10}$~cm, 

352: the Oxygen Ly$\alpha$ line emanates from a region

353: far out of the magnetosphere where the velocity collimation is

354: already dissolved.

355: 

356: In order to finally confirm the magnetic propeller effect being 

357: at work in {\aaq},

358: it is important to detect a bulk velocity shift of an emission line

359: expected from the plasma flow in the vicinity of the magnetosphere.

360: Since the highest temperature of $\sim 5$~keV, 

361: we expect this can be done by the iron K$\alpha$ line

362: with the {\it Astro-E2} XRS.

363: 

364: \acknowledgments

365: 

366: \include{references}

367: \include{table}

368: \include{figure}

369: 

370: \end{document}

371: