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

2: \documentclass{emulateapj}

3: \newcommand{\gtappeq}{\raisebox{-0.6ex}{$\,\stackrel

4: {\raisebox{-.2ex}{$\textstyle >$}}{\sim}\,$}}

5: \newcommand{\deltE}{\Delta\kern-1ptE}

6: \newcommand{\lam}{$\lambda$}

7: \def\ion#1#2{#1\,{\sc #2}}

8: 

9: \input epsf

10: 

11: \usepackage{longtable}

12: \usepackage{amssymb}

13: \usepackage{graphicx}

14: \usepackage{epsfig}

15: \usepackage{epsf}

16: 

17: \slugcomment{To be submitted to ApJ}

18: \shorttitle{FeK$\alpha$ Fluorescence from II~Peg}

19: \shortauthors{Ercolano et al.}

20: \begin{document}

21: \title{Fe~K$\alpha$ and hydrodynamic loop model diagnostics for a large flare on II~Peg}

22: \author{Barbara Ercolano$^{1}$, Jeremy J.~Drake$^2$, Fabio Reale$^{3,4}$, 

23: Paola Testa$^2$ and Jon M.~Miller$^5$}

24: \affil{$^1$Institute of Astronomy, University of Cambridge,\\

25: Madingley Rd, Cambridge, CB3 OHA, UK}

26: \affil{$^2$Harvard-Smithsonian Center for Astrophysics MS-3,\\

27: 60 Garden Street, Cambridge, MA 02138, USA}

28: \affil{$^3$Dipartimento di Scienze Fisiche ed Astronomiche, 

29: Sezione di Astronomia, \\ 

30: Universit`a di Palermo, 

31: Piazza del Parlamento 1, 

32: 90134 Palermo, Italy}

33: \affil{$^4$INAF - Osservatorio Astronomico di Palermo, 

34: Piazza del Parlamento 1, \\

35: 90134 Palermo, Italy}

36: \affil{$^5$Department of Astronomy, University of Michigan, 

37: 500 Church Street,\\ 

38: Ann Arbor, MI 48109-1042}

39: 

40: \begin{abstract}

41: The observation by the {\it Swift} X-ray Telescope of the Fe

42: K$\alpha_1,\alpha_2$ doublet during a large flare on the RS CVn binary

43: system II~Peg represents one of only two firm detections to date of

44: photospheric Fe~K$\alpha$ from a star 

45: other than our Sun.  We present

46: models of the Fe~K$\alpha$ equivalent widths reported in the

47: literature for the II~Peg observations and show that they are most

48: probably due to fluorescence following inner shell photoionisation of

49: quasi-neutral Fe by the flare X-rays. Our models constrain the maximum

50: height of flare the to 0.15~R$_*$ assuming solar abundances for the

51: photospheric material, and 0.1~R$_*$ and 0.06~R$_*$ assuming depleted

52: photospheric abundances ([M/H]~=~-0.2 and [M/H]~=~-0.4, respectively).

53: Accounting for an extended loop geometry has the effect of increasing the

54: estimated flare heights by a factor of $\sim$3. 

55: These predictions are consistent with those derived using results of

56: flaring loop models, which are also used to estimate

57: the flaring loop properties and energetics. From loop models we

58: estimate a flare loop height of 0.13~R$_*$, plasma density of

59: $\sim 4\times10^{12}$~cm$^{-3}$ and emitting volume of

60: $\sim 6\times10^{30}$~cm$^{3}$. Our estimates for the flare dimensions and

61: density allow us to estimate the conductive energy losses to

62: $E_{cond} \leq 2\times10^{36}$~erg, consistent with upper limits

63: previously obtained in the literature.  Finally, we estimate 

64: the average energy output of

65: this large flare to be $\sim10^{33}$~erg~sec$^{-1}$, or

66: 1/10th of the stellar bolometric luminosity.

67: \end{abstract}

68: 

69: \keywords{X-rays: stars --- stars: coronae --- stars: individual

70:   (II~Pegasi) --- stars: flare}

71: 

72: \section{Introduction}\label{s:intro}

73: 

74: \begin{figure*}

75: \begin{minipage}{17.cm}

76: \epsscale{0.32}\plotone{f1a.eps}

77: \epsscale{0.32}\plotone{f1b.eps}

78: \epsscale{0.32}\plotone{f1c.eps}

79: %\epsscale{0.3}\plotone{f1d.eps}

80: %\epsscale{0.3}\plotone{f1e.eps}

81: %\epsscale{0.31}\plotone{f1f.eps}

82: %\epsscale{0.3}\plotone{f1g.eps}

83: %\epsscale{0.3}\plotone{f1h.eps}

84: %\epsscale{0.3}\plotone{f1i.eps}

85: %\epsscale{0.3}\plotone{f1j.eps}

86: %\epsscale{0.31}\plotone{f1k.eps}

87: %\epsscale{0.3}\plotone{f1l.eps}

88: %\epsscale{0.3}\plotone{f1m.eps}

89: %\epsscale{0.3}\plotone{f1n.eps}

90: %\epsscale{0.3}\plotone{f1o.eps}

91: 

92: \end{minipage}

93: \caption[]{Example Fe~K$\alpha$ equivalent widths as a function of flare

94: height and inclination angle for three different photospheric metallicities 

95: for the 1a segment of the flare, as

96: defined by O07. The observed value and its error is indicated by the horizontal

97: black dashed line and teh shaded area. Left: Models calculated at solar abundances

98: (Grevesse \& Sauval, 1998); Central: Models calculated with

99: [M/H]~=~-0.2; Right: Models calculated with [M/H]~=~-0.4.}

100: \label{f:f1}

101: \end{figure*}

102: 

103: 

104: A fraction of the X-rays emitted during stellar flares or by hot

105: stellar coronae are directed downwards towards the stellar

106: photosphere. Here they interact with the photospheric gas and are

107: reprocessed through scattering and photoionisation events. The

108: downward cascade following inner shell photoionisation of

109: quasi-neutral gas in the stellar photosphere produces characteristic

110: fluorescent emission from astrophysically abundant species. While the

111: reprocessed X-ray spectra reach the observer at low flux levels

112: compared to that of the flares/coronae, fluorescent emission lines,

113: can be significantly stronger and can be detected against the spectra

114: of the flare/corona itself.

115: 

116: The intensity or equivalent width of the fluorescent lines detected at

117: Earth depends uniquely on the details of the fluorescing spectrum, the

118: photospheric abundance of the photoionised species and the geometry of

119: the system (flare height and inclination angle; e.g. see fig. 2 of

120: Testa et al., 2008), making them potentially powerful diagnostics. A number of

121: theoretical studies exist which provide a framework for the

122: interpretation of the strong iron fluorescent emission in the solar

123: context (e.g. Tomblin, 1972; Bai, 1979), and more recently for

124: arbitrarily photoionised slabs (Kallman et al., 2004) and stellar

125: photospheres (Drake, Ercolano and Swartz, 2008, DES08). Drake \&

126: Ercolano (2008a,b) have also recently showed that fluorescent emission

127: from neon and oxygen should also be detectable in the solar spectrum

128: with current and future instrumentation. These may provide an

129: independent measure of the photospheric solar oxygen and neon

130: abundance, elements that are central to the solution of the ``solar

131: oxygen crisis'' (e.g. Ayres et al.\ 2006, Socas-Navarro 2007, Basu \&

132: Antia 2008).

133: 

134: Observationally, the $2s-1p$~6.4~KeV Fe K$\alpha$ doublet from low

135: ionisation stages of iron has often been detected in solar spectra

136: (e.g. Neupert et al., 1967; Doschek et al., 1971; Fedelman et al.,

137: 1980; Tanaka et al., 1984; Parmar et al. 1984; Zarro et al., 1992),

138: from disks around pre-main sequence stars (e.g. Tsujimoto et al.,

139: 2005; Favata et al., 2005), and from the photospheres of the single

140: G-type giant HR~9024 (Testa et al., 2007) and of the RS CVn binary

141: system II~Peg (Osten et al., 2007, hereafter O07).

142: 

143: The detection of the 6.4~keV fluorescent iron line in the {\it Swift}

144: X-ray Telescope spectrum of II~Peg taken during a large flare (O07) is

145: particularly interesting as it represents the first of only two

146: detections (with HR~9024, Testa et al., 2007) of photospheric

147: fluorescence emission in stars other than the Sun. O07, however,

148: assigned the excitation mechanism to electron impact ionization of

149: photospheric Fe, rather than photoionisation. In this paper we show by

150: means of Monte Carlo fluorescence modeling that the Fe~K$\alpha$ 

151: equivalent widths

152: reported by O07 are perfectly consistent with photoionisation induced

153: fluorescence. We also show that our predictions of the flaring loop scale

154: height agree with those from flaring loop models (Reale, 2007) and

155: discuss other implications on flaring plasma density, volume and

156: energetics. 

157: 

158: In Section~2 we briefly describe our Monte Carlo model, and in

159: Section~3 we present and discuss our results. Section~4 deals with the

160: flaring loop properties and energetics from hydrodynamic models. Our

161: final conclusions are given in Section~5.

162: 

163: 

164: \section{Monte Carlo Fluorescence Calculations}

165: 

166: The 3D Monte Carlo calculations presented here were performed using a

167: modified version of the photoionisation and dust radiative transfer

168: code, MOCASSIN (Ercolano et al. 2003, 2005, 2008). The code uses a a

169: stochastic approach to the radiation transfer allowing it to deal with

170: problems of arbitrary geometry, while treating both the primary and

171: secondary components of the radiation field self-consistently. The

172: modified version used in this work, which can deal with the production

173: and transfer of fluorescent radiation, is further described by DES08.

174: 

175: We performed calculations to fit all Fe~K$\alpha$ detections reported

176: by O07 for the 5 different orbits (flare intervals) of the {\it Swift}

177: X-ray telescope.  Spectra representing the flaring source of X-ray

178: irradiation for input into the photospheric fluorescence modeling were

179: computed for each of these flare intervals.  Spectra were

180: computed using the CHIANTI database version 5.2 (Dere et al. 1997,

181: Landi et al. 2006) for the energy range 6.2-30~keV, using the

182: temperatures and emission measures found by O07 from two-temperature

183: model fitting and listed in Table~3 of that article.  The upper limit

184: to the energy range was chosen to be sufficiently high that any

185: contribution to the observed fluorescent Fe~K line from higher energy

186: photons in these thermal spectra should be less than 1\%.

187: 

188: Photospheric metal abundances in II~Peg have been studied by Ottmann

189: et al. (1998) and Berdyugina et al. (1998).  Both are based on high

190: resolution, high signal-to-noise spectroscopy, and there is little to

191: choose between the analyses.  The former obtained abundances

192: [Fe/H]~$=-0.2$, [Mg/H]~$=-0.15$, and [Si/H]~$=-0.15$, expressed relative

193: to the solar composition in conventional spectroscopic logarithmic

194: bracket notation, and each with uncertainty of $\pm 0.1$.  As a

195: verification of their method, they successfully recovered the

196: currently accepted solar atmospheric parameters from a spectrum of

197: moonlight.  Berdyugina et al.\ (1998) obtained [M/H]~$=-0.4\pm 0.1$,

198: and verified their approach against the detailed analysis of the K0

199: giant Pollux by Drake \& Smith (1991).  It is therefore reasonable to

200: infer that II~Peg is mildly metal-poor by 0.1-0.4 dex relative to the

201: Sun.  We have investigated the effects of this slight metal depletion

202: on Fe~K$\alpha$ equivalent widths by running three sets of models: (i)

203: with solar abundances (Grevesse \& Sauval, 1998), (ii) with

204: [M/H]~$=-0.2$ and (iii) with [M/H]$=-0.4$.

205: 

206: Fe~K$\alpha$ equivalent widths also strongly depend on the geometry of

207: the system, i.e. height of the flare, $h$, and inclination angle,

208: $\theta$. In particular Fe~K$\alpha$ efficiencies (and therefore

209: equivalent widths) depend on the function $f(h,\theta)$, which has

210: been numerically determined for the case of a single flare

211: illuminating a spherical photosphere by DES08.  Assuming that the

212: flare can be approximated by a point source above the stellar surface,

213: we have performed 

214: calculations at $h~=~0., 0.1, 0.2, 0.3, 0.5$ and $\theta = 0.$ and

215: used the analytical form of $f(\theta)$ given by DES08 in order to 

216: estimate the inclination angle dependence.

217: %in order to produce the geometry diagnostic diagrams presented in

218: %Figure~1 and further discussed in the next Section.

219: We note that in order to scale

220: $\theta = 0$ equivalent widths to an arbitrary value of $\theta$

221: using the $f(\theta)$ functions one has to first scale the latter such

222: that $f(0)~=~1$.

223: 

224: \section{Results from Monte Carlo Fluorescence Modeling}

225: 

226: \subsection{Assuming a point source flare}

227: 

228: Representative flare geometry diagnostic diagrams illustrating the

229: Fe~K$\alpha$ line equivalent width as a function of $\theta$ for

230: different flare heights are presented in Figure~1 for the three values

231: of [M/H] (0.0, $-0.2$ and $-0.4$) for Orbit segment 1a.  Also

232: indicated in these figures for comparison is the observed equivalent

233: width (dashed line) and uncertainties (shaded area).  

234: The combination of predicted and observed equivalent widths

235: can be combined so as to determine the inferred flare height implied

236: by the observations as a function of the angle $\theta$.  For the grid

237: of calculations comprising the five different flare segments and three

238: values of metallicity, we have recast the results into this form in

239: which the flare point source height implied by the observed equivalent

240: width is illustrated as a function of $\theta$ in Figure~2.

241: 

242: At solar metallicities, Fe~K$\alpha$

243: equivalent widths reported for Orbits 1a, 1b, 1c and 2b are well-fit

244: by models with flare height, $h$, up to $\sim$0.15~R$_*$ for low

245: ($\la$20$^o$) inclination angles; larger values of $\theta$ would

246: necessarily demand a lower $h$. Lower metallicity models can also

247: successfully reproduce the observations of Orbits 1a, 1b, 1c and

248: 2b.  Models with [M/H]$=-0.2$ (central column) indicate a maximum flare

249: height of $\sim0.1$~R$_*$, while models with [M/H]$=-0.4$ indicate a

250: maximum flare height of $\sim$0.06~R$_*$. We note that the lowest

251: metallicity models fall slightly short of reproducing the equivalent

252: width of the Fe~K$\alpha$ line measured during Orbit 2b. The

253: discrepancy is small enough to be easily explained by observational

254: uncertainties in the measured equivalent width, which appears

255: somewhat larger compared with the values determined for the other 

256: time intervals by O07.  

257: 

258: Finally, the observations during Orbit 2a yield much lower values of

259: Fe~K$\alpha$ equivalent widths (by a factor of $\sim$3), than for

260: Orbits 1a, 1b, 1c and 2b. This was already pointed out by O07,

261: although there was no obvious explanation for the difference.  

262: According to our model the

263: Fe~K$\alpha$ equivalent width observed during Orbit 2a implies maximum

264: flare heights between 0.5 (for [M/H]~=~0) and 0.45~R$_*$ (for

265: [M/H]~=~-0.4) with an uncertainty of a factor of $\sim 2$.  These 

266: values are even slightly discrepant with the those deduced

267: from the other flare intervals.  Intrigued by this discrepancy, we

268: have independently verified from the {\it Swift} data that the line is

269: indeed weak during this segment, though we obtain a slightly larger 

270: 95\%\ confidence upper limit of $EW < 43$~eV than found by O07.

271: 

272: \subsection{Effects of extended loop geometry on Fe~K$\alpha$

273:   fluorescence} 

274: 

275: The above fluorescence calculations and interpretation assume that the

276: flare can be approximated by a point source of X-rays.  In the case of

277: large flares similar to those seen on the Sun, with peak temperatures

278: of order $10^7$~K, the ``hard'' X-ray photons above the 7.11 keV Fe~K

279: ionisation threshold will be emitted almost uniquely by the hotter

280: loop apex and the point source approximation should be accurate.

281: However, Testa et al. (2008) have shown that the extreme flare 

282: observed on the giant HR~9024 that reached temperatures of $\sim

283: 10^8$~K emits ionising radiation over its entire length, 

284: such that the equivalent point-source flare height for Fe~K$\alpha$ 

285: fluorescence was $h/3$, where $h$ is the actual loop height.  

286: 

287: The II~Peg flare analysed here is similar to the HR~9024 one in terms

288: of plasma temperature, and we can expect a similar scaling of the

289: point-source height and true flare height.  The corresponding heights

290: for the [M/H]~$=-0.2$ and $-0.4$ models are $h\sim 0.3$ and

291: $0.15$~R$_*$, respectively, taking segments 1a, 1b and 1c as

292: representative. 

293: 

294: 

295: \section{Flaring loop properties and energetics}

296: 

297: \begin{figure}

298: \epsscale{0.92}\plotone{f2a.eps} \\

299: \epsscale{0.92}\plotone{f2b.eps} \\

300: \epsscale{0.92}\plotone{f2c.eps}

301: \caption[]{Inferred flare height vs astrocentric angle for the different

302: flare segments computed for the different photospheric metallicities 

303: considered here: [M/H]=0.0 (top); [M/H]$=-0.2$ (middle); and [M/H]$=-0.4$ 

304: (bottom).  The observed X-ray spectral properties and Fe~K$\alpha$ 

305: equivalent widths for 

306: segments 1b and 1c are essentially identical and have been averaged.  

307: The different segments are colour-coded, with upper and lower limits 

308: given by dashed and dotted curves, respectively. }

309: \label{f:h_ang}

310: \end{figure}

311: 

312: The allowed ranges of flaring loop scale height derived from the

313: strength of the Fe~K line in the previous section can be compared with

314: predictions from detailed hydrodynamic flare models.  Here, we use the

315: scaling laws for loop length, cross-sectional area and plasma density

316: derived from hydrodynamic simulations by Reale (2007) in application to the

317: light curve and plasma emission measure and temperature found from

318: the {\it Swift} observations by O07.  

319: 

320: Relevant quantities required by the Reale (2007) relations are the

321: flare rise time $t_{M,3}$ (in units of $10^3$s), the maximum plasma

322: temperature $T_{0,7}$ (in units of $10^7$~K), the peak emission measure

323: $\Phi$ (cm$^{-3}$), and the $1/e$ decay time $\tau^\prime_s$ (s).

324: We estimate the flare rise and decay times from Fig.~1 in O07

325: al.\ (2007), and take the flare temperature and emission measure data

326: from their Table 2.  Since we are interested in the peak flare

327: temperature, we adopt the highest temperature obtained from their

328: 3-temperature best-fit models.  The final set of adopted parameters is

329: %\begin{align}

330: \begin{eqnarray}

331: t_{M,3} & \approx & 2 \\ 

332: T_{0,7} & \approx & 30 \\

333: \Phi & \approx & 10^{56} \\

334: \tau^\prime_s & \approx & 9000 

335: %\end{align}

336: \end{eqnarray}

337: Here, we have assumed that the maximum plasma temperature is the value

338: observed at the time of maximum density, and that this time corresponds to the

339: peak in the observed flare emission.  Since the flaring loop is a

340: closed volume, the latter assumption is straightforward. However, as noted by

341: Reale (2007), the maximum flare temperature can be reached some time after 

342: the peak density is reached; we allow for this in the analysis below. 

343: 

344: \subsection{Loop length}

345: 

346: The flaring loop half-length can be estimated from the from rise time 

347: and the maximum plasma temperature, as given in Eqn.~(12) of Reale (2007):

348: \begin{equation}

349: L_9 \approx 3 ~ \psi^2 T_{0,7}^{1/2} t_{M,3}

350: \label{eq:lris}

351: \end{equation}

352: where $L_9$ is the loop half-length in units of $10^9$~cm.  The

353: parameter $\psi$ is the ratio of the maximum plasma temperature to that at 

354: density maximum,

355: \[

356: \psi = \frac{T_0}{T_{M}}.

357: \]

358: From Fig.~2 of O07, it can be seen that the rise

359: phase flattens toward the emission maximum; this suggests the flaring

360: loop is close to equilibrium at the maximum and that, therefore, $\psi

361: \approx 1$ and the assumption of maximum plasma temperature at the

362: time of maximum density holds.  To allow for deviations from this

363: assumption, we also quote results for $\psi \approx 1.3$ (see examples

364: given in Reale 2007).  For $T_{0,7} \approx 30$, $t_{M,3} \approx 2$,

365: $\psi \approx 1$

366: \[

367: L_9 \approx 30

368: \]

369: For $\psi \approx 1.3$

370: 

371: \[

372: L_9 \approx 70

373: \]

374: As a reasonable average value, we adopt $L_9 \approx 50 \pm 20$.

375: Adopting a radius for II~Peg of $3.4R_\odot$ (Berdyugina et al.\

376: 1998), this loop half-length corresponds to a loop {\em height} of

377: $\sim0.13R_\star$.  This compares very favourably with the values we have

378: estimated from the Fe~K fluorescence line in the previous section,

379: provided the flare were located toward the centre of the visible

380: stellar disk and not toward the limb.

381: 

382: As a verification that this loop length value is reasonable, we can

383: also obtain an estimate of the {\em upper limit} to $L_9$

384: from the decay timescale.  If the decay is entirely driven by cooling,

385: with no further heat input, the observed decay timescale can be

386: equated with the thermodynamic decay time,

387: \begin{equation}

388: \tau^\prime_s = \phi \tau_s

389: \label{eq:taunew}

390: \end{equation}

391: where $\phi \approx 1.3$ and 

392: \begin{equation}

393: \tau_{s} = 3.7 \times 10^{-4} \frac{L}{\sqrt{T_0}}

394: = 120 \frac{L_9}{\sqrt{T_{0,7}}}

395: \label{eq:tserio}

396: \end{equation}

397: For $\tau^\prime_s \approx 9000$~s, we obtain an upper limit for

398: the loop length:

399: \[

400: L_9 \leq 320. 

401: \]

402: While much larger, this upper limit is 

403: consistent with the value obtained from the rise phase above.

404: 

405: Huenemoerder et al.\ (2001) observed a large flare on II~Peg during a

406: 45~ks observation with the {\it Chandra} High Energy Transmission

407: Grating Spectrometer (HETGS), during which the count rate was observed

408: to rise to 2.5 times its quiescent value and the peak flare

409: temperature reached $\log T \sim 7.6$.  While much less intense than

410: the {\it Swift} flare discussed here, the decay timescale of $\sim

411: 65$~ks was somewhat longer.  Huendmoerder et al.\ (2001) interpreted

412: the light curve in terms of both single loop and two-ribbon type

413: flares and inferred a loop height in the range 0.05--0.25~R$_\star$

414: for the former, depending on the plasma density assumed.  This height

415: then scales inversely with the cube of the number of loops

416: involved---e.g.\ smaller by a factor of ~5 for 100 loop strands.

417: While this estimate must be considered very approximate, the

418: similarity with the loop height derived here for the {\it Swift} flare

419: suggests that both events occurred in similar coronal structures that

420: might be typical of the outer atmosphere of II~Peg.

421: 

422: \subsection{Plasma Density, Volume and Loop Cross-sectional Area}

423: 

424: From Reale (2007), the maximum possible density equilibrium value at

425: the loop apex in units of 

426: $10^{10}$~cm$^{-3}$ is given by 

427: \[

428: n_{0,10} = 13 \frac{T_{0,7}^2}{ L_9}

429: \]

430: While this is an upper limit, the true value should not be lower by 

431: more than a factor of 2.  For $T_{0,7} \approx 30$ and $L_9 \approx 50$, the 

432: density is then 

433: \[

434: n_{0,10} \leq 360

435: \]

436: 

437: A reasonable value to adopt is $n_{0,10} \approx 200$. This value is compatible with density diagnostics from line ratios of

438: hot He-like triplets, such as Mg~{\sc xi}, observed during flaring and

439: quiescence on II~Peg (Huenemoerder et al.\ 2001; Testa, Drake \& Peres

440: 2004). Similarly high densities have also been reported by G{\"u}del et al. (2002) during an X-ray flare in 

441: Proxima Centauri.

442: From Reale (2007), the volume is:

443: \begin{equation}

444: V \approx \frac{\Phi}{n_{avg}^2}

445: \label{eq:vol}

446: \end{equation}

447: where 

448: \begin{equation}

449: n_{avg} = n_M \frac{T_M}{T_{avg}}

450: \label{eq:navg}

451: \end{equation}

452: Taking $T_{avg} \approx 150$ MK from the {\em single temperature} model 

453: fit of O07 (their Table 2), we obtain

454: \[

455: n_{avg} \approx 4 \times 10^{12} ~~~~~ \rm cm^{-3}

456: \]

457: and 

458: \[

459: V \approx 6 \times 10^{30} ~~~~~ \rm cm^3

460: \]

461: 

462: The cross-sectional area is:

463: \begin{equation}

464: A \approx \frac{V}{2 ~ L}

465: \label{eq:area}

466: \end{equation}

467: For $L_9 \approx 50$, $A \approx 6 \times 10^{19}$ cm$^2$. For a

468: single loop with circular cross-section we obtain a radius in units of

469: $10^9$~cm of $r_9 \approx 4$, which is slightly less than 1/10 the

470: loop half-length and very similar to the values found for solar

471: flaring loops (e.g. Cheng et al.\ 1980, Golub et al.\ 1980, Peres et al.\ 

472: 1987).

473: 

474: \subsection{Conductive energy losses}

475: 

476: One of the most striking aspects of the {\it Swift} II~Peg flare is

477: the potentially large flare energy budget.  Using the classical

478: Spitzer formulism, O07 found an upper limit to the conductive

479: losses of $5\times 10^{43}$~erg, but were unable to constrain this

480: further without estimates of the flare dimensions and density.  The

481: bolometric luminosity of II~Peg is $\log L_{bol}=34.2$ (Marino et al.\

482: 1999), and conductive losses as high as $10^{43}$~erg during a single,

483: compact flare would represent an astounding concentration of energy.

484: 

485: We can refine the estimate of the conductive energy losses

486: using Eqn.~(5) in O07 and the parameters derived above

487: for the flaring loop:

488: \[

489: E_{cond} = \frac{\kappa T^{7/2} \Phi \Delta t}{L^2 n_e^2},

490: \]

491: where, for Orbit 1, $T \approx 300$ MK (an upper limit for the flare as

492: a whole), $\Phi \approx 10^{56}$, $\kappa \approx 10^{-6}$ (in

493: c.g.s. units), $\Delta t \approx 2$ ks, $L_9 \approx 50$, $n_e \approx

494: n_{avg} \approx 4 \times 10^{12}$ cm$^{-3}$, we obtain

495: \[

496: E_{cond} \leq 2 \times 10^{36} ~~~ \rm erg

497: \]

498: which is similar to the total radiated energy of $6\times 10^{36}$~erg

499: in the 0.01-200 keV energy band estimated by O07.

500: The duration of the flare was of order $10^4$s, and the radiative and

501: conductive losses therefore represent an energy output of $\sim

502: 10^{33}$~erg~s$^{-1}$, or about 1/10 the stellar bolometric

503: luminosity.

504: 

505: 

506: \section{Conclusions}

507: 

508: We have presented a set of Monte Carlo calculations able to reproduce

509: the Fe~K$\alpha$ equivalent widths from the {\it Swift} X-ray

510: Telescope observations of II~Peg during a 'superflare' (Osten et al.,

511: 2007, O07). Our models show that the data are consistent with the

512: Fe~K$\alpha$ emission being produced by fluorescence following K-shell

513: photoionisation of quasi-neutral iron in the stellar photosphere.

514: This contrasts with the interpretation of O07, who favoured

515: collisional ionisation by non-thermal electrons to produce the

516: Fe~K$\alpha$ emission. They argued that the normal incidence flare

517: X-ray penetration depth required in the photosphere to obtain the

518: observed equivalent widths is similar to the $\tau=1$ Compton

519: scattering depth, and consequently any fluorescent photons produced

520: would not easily escape.  However, this assessment was based on a

521: simple analytical formula appropriate for optically thin cases in

522: which only a small fraction of the incident X-ray flux undergoes

523: photoabsorption or scattering (e.g. Liedahl, 1999; Krolik \& Kallman,

524: 1987).  The semi-infinite photospheric case lies outside the range of

525: applicability of this formula.  Furthermore, incident angles on the

526: photosphere range from $\sim 0$--$90^\circ$ and path lengths for

527: escape can therefore be much smaller than gas penetration depths (by a

528: factor equal to the inverse cosine of these angles).  Our stochastic

529: treatment includes Compton scattering and shows that the observed

530: equivalent widths can be produced by fluorescence.  We also note that

531: the impact excitation mechanism is a very low efficiency process and

532: requires a large amount of energy in the form of accelerated

533: electrons.  This was already noted by, e.g., Parmar (1984) in

534: reference to solar Fe fluorescence and by Ballantyne (2003), and was

535: discussed in reference to a large flare on II~Peg by Testa et

536: al. (2008).

537: 

538: The derived flare loop heights are $h$~=~0.15~R$_*$, assuming solar

539: photospheric abundances, $h$~=~0.1~R$_*$ assuming photospheric

540: abundances depleted by [M/H]~=~-0.2 and $h$~=~0.06~R$_*$ assuming

541: [M/H]~=~-0.4. These value are in good agreement with the predictions

542: from our alternative analysis based on hydrodynamic models

543: (e.g. Reale, 2007) which yield $h \approx 0.13$~R$_*$. We estimate

544: flaring loop properties and energetics using scaling laws based on the

545: hydrodynamic loop modeling of Reale (2007) and obtain a plasma

546: density of $4\times10^{12}$~cm$^{-3}$ and volume of

547: $6\times10^{30}$~cm$^{3}$. Using the derived values for the flare

548: dimensions and densities we obtain a more stringent upper limit for

549: the conductive energy loss, $E_{cond}$, than previously possible,

550: setting $E_{cond} \leq 2\times10^{36}$~erg. Considering the duration

551: of the flare and the radiative and conductive losses, the energy

552: output of the flare is estimated at $\sim$10$^{33}$~erg sec$^{-1}$,

553: which is approximately one tenth of the stellar bolometric luminosity.

554: 

555: While we cannot rule out a contribution to the observed Fe~K$\alpha$

556: flux from electron impact with nonthermal electrons, as proposed by

557: O07, such contribution is not necessary to explain the available data.

558: 

559: \section*{Acknowledgments}

560: 

561: We thank the anonymous referee and the editor Eric Feigelson for helpful comments that added to the clarity of the paper and the interpretation of the results. JJD was supported by the Chandra X-ray Center NASA contract NAS8-39073 during the course of this research. The simulations were run on the Cosmos 

562: (SGI altix 4700) supercomputer at DAMTP in Cambridge. Cosmos is a 

563: UK-CCC facility which is supported by HEFCE and STFC.

564: 

565: 

566: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

567: 

568: 

569: \subsection*{References}

570: 

571: \begin{description}

572: \item[] Ayres, T.~R., Plymate, 

573: C., \& Keller, C.~U.\ 2006, \apjs, 165, 618 

574: 

575: \item[] Basu, S., \& Antia, H.~M.\ 2008, \physrep, 457, 217 

576: 

577: \item[] Socas-Navarro, H., \& Norton, A.~A.\ 2007, \apjl, 660, L153 

578: 

579: \item[] Berdyugina, S.~V., Jankov, S., Ilyin, I., Tuominen, I., \& 

580:   Fekel, F. C. 1998, A\&A, 334, 863

581: 

582: \item[] Cheng, C.-C.,

583: Tandberg-Hanssen, E., \& Smith, J.~B., Jr.\ 1980, Sol.Phys., 67, 259

584: 

585: \item[] Dere, K.~P., Landi, E., Mason, H.~E., Monsignori Fossi, B.~C., 

586: \& Young, P.~R.\ 1997, A\&AS, 125, 149

587: 

588: \bibitem[Drake et al.(2008)]{2008ApJ...678..385D} Drake, J.~J., Ercolano, 

589: B., \& Swartz, D.~A.\ 2008, \apj, 678, 385, DES08

590: 

591: \bibitem[Drake 

592: \& Ercolano(2007)]{2007ApJ...665L.175D} Drake, J.~J., \& Ercolano, B.\ 2007,

593: \apjl, 665, L175 

594: 

595: \bibitem[Drake 

596: \& Smith(1991)]{1991MNRAS.250...89D} Drake, J.~J., \& Smith, G.\ 1991, \mnras,

597: 250, 89 

598: 

599: \item[] Ercolano, B.,

600: Barlow, M.~J., Storey, P.~J., \& Liu, X.-W.\ 2003a, \mnras, 340, 1136 

601: 

602: \item[] Ercolano, B., Young, 

603: P.~R., Drake, J.~J., 

604: \& Raymond, J.~C.\ 2007, ArXiv e-prints, 710, arXiv:0710.2103 

605: 

606: \item[] Golub, L., Maxson, C., 

607: Rosner, R., Vaiana, G.~S., \& Serio, S.\ 1980, ApJ, 238, 343

608: 

609: \item[] G{\"u}del, M., 

610: Audard, M., Skinner, S.~L., \& Horvath, M.~I.\ 2002, \apjl, 580, L73 

611: 

612: 

613: \item[] Huenemoerder, 

614: D.~P., Canizares, C.~R., \& Schulz, N.~S.\ 2001, \apj, 559, 1135 

615: 

616: \item[] Krolik, J.~H., \& Kallman, T.~R.\ 1987, \apjl, 320, L5 

617: 

618: \item[] Landi, E., Del Zanna, G., 

619: Young, P.~R., Dere, K.~P., Mason, H.~E., \& Landini, M.\ 2006, ApJS, 162, 261 

620: 

621: \item[] Liedahl, D.~A.\ 1999, X-Ray Spectroscopy in Astrophysics, 520, 189 

622: 

623: \item[] Marino, G., Rodon\'o, M., Leto, G., \& Cutispoto, G. 1999, A\&A, 

624: 352, 189

625: 

626: \item[] Osten, R.~A., Drake, S., Tueller, J., Cummings, J., Perri, M.,

627: Moretti, A., \& Covino, S.\ 2007, ApJ, 654, 1052

628: 

629: \item[] Peres, G., Reale, F., 

630: Serio, S., \& Pallavicini, R.\ 1987, ApJ, 312, 895 

631: 

632: \item[] Reale, F.\ 2007, A\&A, 471, 271

633: 

634: \bibitem[Testa et al.(2004)]{2004ApJ...617..508T} Testa, P., Drake, J.~J., 

635: \& Peres, G.\ 2004, \apj, 617, 508 

636: 

637: \item[] Testa, P., Drake, J.~J., Ercolano, B., Reale, F., Huenemoerder, D.~P., Affer, L., Micela, G., \& Garcia-Alvarez, D.\ 2008, \apjl, 675, L97 

638: 

639: 

640: 

641: \end{description}

642: 

643: 

644: 

645: 

646: 

647: \end{document}

648: 

649: