1: \documentclass[twocolumn,twoside]{article}

2: \usepackage{graphicx,aasab}

3: \usepackage[super,sort&compress]{natbib}

4: \pagestyle{myheadings}

5: 

6: %%% [Nature Manuscript number B04636]

7: %%% [Title: maximum of 3 lines of 30 characters per line, including spaces.]

8: %%% [Abstract: maximum of 180 words.]

9: %%% [Text: maximum of 2500 words.]

10: %%% [Caption: maximum of 300 words.]

11: 

12: \renewcommand{\baselinestretch}{0.949}

13: \renewcommand{\thefootnote}{\fnsymbol{footnote}}

14: 

15: \setlength{\paperheight}{11.0in}

16: \setlength{\topmargin}{-0.75in}

17: \setlength{\headheight}{0.25in}

18: \setlength{\headsep}{0.25in}

19: \setlength{\topskip}{0.0in}

20: \setlength{\textheight}{9.75in}

21: \setlength{\footskip}{0.5in}

22: 

23: \setlength{\paperwidth}{8.5in}

24: \setlength{\textwidth}{7.0in}

25: \setlength{\oddsidemargin}{-0.25in}

26: \setlength{\evensidemargin}{-0.25in}

27: \setlength{\marginparwidth}{0pt}

28: \setlength{\marginparsep}{0pt}

29: 

30: \setlength{\parskip}{0ex}

31: \setlength{\parindent}{1em}

32: 

33: \newcommand{\bibfont}{\scriptsize}

34: \setlength{\bibsep}{0ex}

35: \bibpunct[,]{(}{)}{,}{s}{ }{,}

36: 

37: \newcommand{\sgrastar}{Sgr~A\mbox{$^*$}}

38: 

39: \begin{document}

40: 

41: \markright{Nature, Vol.~413, pp.~45--48 (2001)}

42: 

43: \begin{flushleft}

44: {\raggedright\Large\bf\noindent Rapid X-ray flaring from the direction

45: of the supermassive black hole at the Galactic Centre}

46: \end{flushleft}

47: 

48: \begin{flushleft}

49: {\bf\noindent F.~K.~Baganoff$^\star$, M.~W.~Bautz$^\star$,

50: W.~N.~Brandt$\dagger$, G.~Chartas$\dagger$, E.~D.~Feigelson$\dagger$,

51: G.~P.~Garmire$\dagger$, Y.~Maeda$\dagger\ddagger$, M.~Morris\S,

52: G.~R.~Ricker$^\star$, L.~K.~Townsley$\dagger$ \& F.~Walter$\parallel$}

53: \end{flushleft}

54: 

55: \begin{flushleft}

56: \footnotesize\noindent

57: \emph{$^\star$ Center for Space Research, Massachusetts Institute of

58:   Technology, Cambridge, MA 02139-4307, USA}\\

59: \emph{$\dagger$ Department of Astronomy and Astrophysics, Pennsylvania

60:   State University, University Park, PA 16802-6305, USA}\\

61: \emph{$\ddagger$ Institute of Space and Astronautical Science, 3-1-1

62:   Yoshinodai, Sagamihara, 229-8501, Japan}\\

63: \emph{\S Department of Physics and Astronomy, University of California

64:   at Los Angeles, Los Angeles, CA 90095-1562, USA}\\

65: \emph{$\parallel$ Department of Astronomy, California Institute of

66:   Technology, Pasadena, CA 91125, USA}

67: \end{flushleft}

68: 

69: \medskip

70: 

71: {\small

72: 

73: {\bf\noindent Most galactic nuclei are now believed to harbour

74: supermassive black holes\nolinebreak\cite{Richstone98}.  Studies of

75: stellar motions in the central few light-years of our Milky Way Galaxy

76: indicate the presence of a dark object with a mass of

77: $\mathbf{\approx2.6 \times 10^6}$ solar masses (refs 2, 3).  This

78: object is spatially coincident with Sagittarius~A$^*$ (\sgrastar), the

79: unique compact radio source located at the dynamical centre of our

80: Galaxy.  By analogy with distant quasars and nearby active galactic

81: nuclei (AGN), \sgrastar\ is thought to be powered by the gravitational

82: potential energy released by matter as it accretes onto a supermassive

83: black hole\nolinebreak\cite{Lynden-Bell71, Melia01c}.  However,

84: \sgrastar\ is much fainter than expected in all wavebands, especially

85: in X-rays, casting some doubt on this model.  Recently, we reported

86: the first strong evidence of X-ray emission from \sgrastar\ (ref.~6).

87: Here we report the discovery of rapid X-ray flaring from the direction

88: of \sgrastar.  These data provide compelling evidence that the X-ray

89: emission is coming from accretion onto a supermassive black hole at

90: the Galactic Centre, and the nature of the variations provides strong

91: constraints on the astrophysical processes near the event horizon of

92: the black hole.}

93: 

94: Our view of \sgrastar\ in the optical and ultraviolet wavebands is

95: blocked by the large visual extinction, $A_{\rm V} \approx

96: 30$~magnitudes\nolinebreak\cite{Morris96}, caused by dust and gas

97: along the line of sight.  \sgrastar\ has not been detected in the

98: infrared due to its faintness and to the bright infrared background

99: from stars and clouds of dust\nolinebreak\cite{Menten97}.  Detection

100: of X-rays from \sgrastar\ is therefore essential to constrain the

101: spectrum at energies above the radio-to-submillimetre band and to test

102: the supermassive-black-hole accretion-flow

103: paradigm\nolinebreak\cite{Melia01c}.

104: 

105: We first observed the Galactic Centre on 21 September 1999 with the

106: imaging array of the Advanced CCD Imaging Spectrometer (ACIS-I) aboard

107: the \emph{Chandra X-ray Observatory}\nolinebreak\cite{Weisskopf96} and

108: discovered an X-ray source coincident within $0\farcs35 \pm 0\farcs26$

109: ($1\sigma$) of the radio source\nolinebreak\cite{Baganoff01}.  The

110: luminosity in 1999 was very weak, $L_{\rm X} \approx 2 \times 10^{33}$

111: erg s$^{-1}$ in the 2--10~keV band, after correction for the inferred

112: neutral hydrogen absorption column $N_{\rm H} \approx 1 \times

113: 10^{23}$ cm$^{-2}$.  This is far fainter than previous X-ray

114: observatories could detect\nolinebreak\cite{Baganoff01}.

115: 

116: We observed the Galactic Centre a second time with

117: \emph{Chandra}/ACIS-I from 26 October 2000 22:29 through 27 October

118: 2000 08:19 (UT), during which time we saw a source at the position of

119: \sgrastar\ brighten dramatically for a period of $\approx10$~ks.

120: Figure~\ref{fig:surfplot} shows surface plots for both epochs of the

121: 2--8~keV counts integrated over time from a $20\arcsec \times

122: 20\arcsec$ region centred on the radio position of \sgrastar.  The

123: modest peak of the integrated counts at \sgrastar\ in 1999 increased

124: by a factor of $\approx7$ in 2000, despite the 12\% shorter exposure.

125: The peak integrated counts of the fainter features in the field show

126: no evidence of strong variability, demonstrating that the flaring at

127: \sgrastar\ is intrinsic to the source.

128: 

129: \begin{figure}[!htb]

130: \begin{center}

131: \includegraphics[width=0.484\textwidth]{fig1.eps}

132: \caption{\footnotesize Surface plots of the 2--8~keV counts within a

133: $20\arcsec \times 20\arcsec$ field centred on \sgrastar\ at two

134: epochs.  The data were taken with \emph{Chandra}/ACIS-I on ({\bf a})

135: 21 September 1999 and ({\bf b}) 26--27 October 2000.  The effective

136: exposure times were 40.3~ks and 35.4~ks, respectively.  The spatial

137: resolution is 0\farcs5 per pixel.  An angle of 1\arcsec\ on the sky

138: subtends a projected distance $\approx0.04$~pc at the galactocentric

139: distance of 8.0~kpc (ref.~30).  The peak integrated counts per pixel

140: at the position of \sgrastar\ increased by a factor of $\approx7$ from

141: the first epoch to the second, despite the slightly smaller exposure

142: time ($\approx12\%$) in the second epoch.  The low-level peak a few

143: arcseconds to the southwest of \sgrastar\ is the infrared source

144: IRS~13, while the ridge of emission to the northwest is from a string

145: of unresolved point sources.  The fainter features in the field are

146: reasonably consistent between the two epochs, considering the limited

147: Poisson statistics and the fact that these stellar sources may

148: themselves be variable; this consistency shows that the strong

149: variations at \sgrastar\ are intrinsic to the source.}

150: \label{fig:surfplot}

151: \end{center}

152: \end{figure}

153: 

154: Figure~\ref{fig:ltc} shows light curves of the photon arrival times

155: from the direction of \sgrastar\ during the observation in 2000.

156: Panels (a) and (b) show hard-band (4.5--8~keV) and soft-band

157: (2--4.5~keV) light curves constructed from counts within an angular

158: radius of 1\farcs5 of \sgrastar.  Both bands exhibit roughly constant,

159: low-level emission for the first $\approx14$~ks, followed by an

160: $\approx6$-ks period of enhanced emission beginning with a 500-s event

161: ($4.4\sigma$ significance using 150-s bins).  At 20~ks, there occurs a

162: large-relative-amplitude flare or flares lasting $\approx10$~ks, and

163: finally a return to the low state for the remaining $\approx6$~ks.

164: About 26~ks into the observation, the hard-band light curve drops

165: abruptly by a factor of $\approx5$ within a span of $\approx600$~s and

166: then partially recovers within a period of $\approx1.2$~ks.  The

167: soft-band light curve shows a similar feature, but it appears to lag

168: the hard-band event by a few hundred seconds and is less sharply

169: defined.

170: 

171: \begin{figure}[!htb]

172: \begin{center}

173: \includegraphics[width=0.484\textwidth]{fig2.eps}

174: \caption{\footnotesize Light curves of the photon arrival times and

175: band ratios from the direction of \sgrastar\ on 26--27 October 2000.

176: Panels ({\bf a}) hard-band (HB = 4.5--8~keV) counts, ({\bf b})

177: soft-band (SB = 2--4.5~keV) counts, ({\bf c}) hard/soft band ratio

178: within a circle of radius 1\farcs5, and ({\bf d}) hard/soft band ratio

179: within an annulus of inner and outer radii 0\farcs5 and 2\farcs5,

180: respectively.  The x-axis shows the time offset from the start of the

181: observation at 26 October 2000 22:29 (UT).  The data are shown with

182: $1\sigma$ error bars.  The single 2.8-hour period of flaring activity

183: which we have detected so far during a total of 21 hours of

184: observations yields a (poorly determined) duty cycle of $\sim1/8$.

185: Our continuing observations of this source with \emph{Chandra} will

186: permit us to refine this value.  During the quiescent intervals at the

187: beginning and end of this observation, the mean count rate in the

188: 2--8~keV band was $(6.4 \pm 0.6) \times 10^{-3}$ counts s$^{-1}$,

189: consistent with the count rate we measured in 1999, which was $(5.4

190: \pm 0.4) \times 10^{-3}$ counts s$^{-1}$.  The detected count rate at

191: the peak of the flare was $0.16 \pm 0.01$ counts s$^{-1}$ within a

192: 1\farcs5-radius extraction circle, but this is $\approx30$\% less than

193: the true incident count rate due to pile-up of X-rays in the detector

194: during the 3.2-s integration time per CCD read out.}

195: \label{fig:ltc}

196: \end{center}

197: \end{figure}

198: 

199: The band-ratio time series in panel (c), defined as the ratio of

200: hard-band counts to soft-band counts, suggests the spectrum hardened

201: during the flare.  The difference between the band ratio measured at

202: the peak of the flare and the average of the band ratios during the

203: quiescent periods at the beginning and end of the observation is $0.63

204: \pm 0.21$ (i.e., $3\sigma$).  The peak-flare band ratio in panel (c)

205: is affected to some extent by the effects of pile-up (see

206: Figure~\ref{fig:ltc} caption), which would tend to harden the

207: spectrum; however, the band ratios in panel (d), which were computed

208: using the non-piled-up data extracted from the wings of the point

209: spread function, also show evidence for spectral hardening with

210: $2.7\sigma$ significance.  The sizes of dust-scattering halos in the

211: Galactic Centre are typically $\ga1\arcmin$ (ref.~10), so

212: dust-scattered X-rays from the source contribute a negligible fraction

213: of the emission within the source extraction region that we used;

214: hence it cannot account for the spectral variations.  We therefore

215: conclude that the spectral hardening during the flare is likely to be

216: real.

217: 

218: \markboth{Baganoff et~al.---Discovery of rapid X-ray flaring from the

219: direction of \sgrastar}{Baganoff et~al.---Discovery of rapid X-ray

220: flaring from the direction of \sgrastar}

221: 

222: The quiescent-state spectra in 1999 and 2000 and the peak

223: flaring-state spectrum in 2000 are shown in Figure~\ref{fig:spectra}.

224: We fit each spectrum individually using a single power-law model with

225: corrections for the effects of photoelectric absorption and dust

226: scattering\nolinebreak\cite{Predehl95}.  The best-fit values and 90\%

227: confidence limits for the parameters of each fit are presented in the

228: first three lines of Table~\ref{tab:fits}.  The column densities for

229: the three spectra are consistent, within the uncertainties, as are the

230: photon indices of both quiescent-state spectra.  Next, we fit a double

231: power-law model to the three spectra simultaneously, using a single

232: photon index for both quiescent spectra, a second photon index for the

233: flaring spectrum, and a single column density for all three spectra.

234: The best-fit models for each spectrum from the simultaneous fits are

235: shown as solid lines in Figure~\ref{fig:spectra}; the parameter values

236: are given in the last line of Table~\ref{tab:fits}.  Using these

237: values, we derive an absorption-corrected 2--10~keV luminosity of

238: $L_{\rm X} = (2.2_{-0.3}^{+0.4}) \times 10^{33}$ erg s$^{-1}$ for the

239: quiescent-state emission and $L_{\rm X} = (1.0 \pm 0.1) \times

240: 10^{35}$ erg s$^{-1}$ for the peak of the flaring-state emission, or

241: $\approx45$ times the quiescent-state luminosity.  We note that

242: previous X-ray observatories did not have the sensitivity to detect

243: such a short-duration, low-luminosity flare in the Galactic Centre

244: \nolinebreak\cite{Baganoff01}.  The best-fit photon index $\Gamma_{\rm

245: f} = 1.3_{-0.6}^{+0.5}$ ($N(E) \propto E^{-\Gamma}$) for the

246: flaring-state spectrum is slightly flatter than, but consistent with,

247: systems thought to contain supermassive black

248: holes\nolinebreak\cite{Mushotzky93}.

249: 

250: \begin{figure}[!htb]

251: \begin{center}

252: \includegraphics[width=0.484\textwidth]{fig3.eps}

253: \caption{\footnotesize X-ray spectra of the \emph{Chandra} source at

254: the position of \sgrastar.  The data are shown as crosses with

255: vertical bars indicating the $1\sigma$ errors in the count rate and

256: horizontal bars the energy range of each bin.  The events have been

257: grouped to yield 10 counts per bin.  The counts in the 1999 (black)

258: and 2000 (red) quiescent-state spectra were extracted using a source

259: radius of 1\farcs5.  A non-piled-up, peak flaring-state spectrum

260: (green) was extracted from the wings of the point spread function,

261: using an annulus with inner and outer radii of 0\farcs5 and 2\farcs5,

262: during the time interval 23.7--26.3~ks after the start of the 2000

263: observation.  The solid lines are the best-fit models for each

264: spectrum, obtained by fitting an absorbed, dust-scattered, double

265: power-law model to the three spectra simultaneously.  The best-fit

266: values and 90\% confidence intervals for the model parameters are

267: given in the last line of Table~\ref{tab:fits}.  The spectra are

268: well-fitted using a single photon index for both quiescent spectra, a

269: second photon index for the flaring spectrum, and a single column

270: density for all three spectra.  The column density from the

271: simultaneous fits corresponds to a visual extinction $A_{\rm V} =

272: 29.6_{-6.1}^{+5.0}$~magnitudes, which agrees well with

273: infrared-derived estimates of $A_{\rm V} \approx

274: 30$~magnitudes\protect\nolinebreak\cite{Morris96}; we thus find no

275: evidence in our X-ray data of excess gas and dust localized around the

276: supermassive black hole.  This places an important constraint on the

277: maximum contribution to the infrared spectrum of \sgrastar\ produced

278: by local dust reprocessing of higher energy photons from the accretion

279: flow.  The spectral models used in ref.~6 did not account for dust

280: scattering; hence a higher column density was needed to reproduce the

281: low-energy cut-off via photoelectric absorption alone.  We note there

282: is no sign of an iron K$\alpha$ emission line in the flaring-state

283: spectrum.}

284: \label{fig:spectra}

285: \end{center}

286: \end{figure}

287: 

288: \begin{table}[!hb]

289: \begin{center}

290: \footnotesize

291: \caption{Spectral Fits}

292: \label{tab:fits}

293: \begin{tabular}[0.484\textwidth]{lcccc}

294: \multicolumn{5}{l}{} \\

295: \hline

296: \multicolumn{5}{l}{} \\

297: \multicolumn{1}{c}{Spectrum} &

298: \multicolumn{1}{c}{$N_{\rm H}$} &

299: \multicolumn{1}{c}{$\Gamma_{\rm q}$} &

300: \multicolumn{1}{c}{$\Gamma_{\rm f}$} &

301: \multicolumn{1}{c}{$\chi^2/\nu$} \\

302: \multicolumn{5}{l}{} \\

303: \hline

304: \multicolumn{5}{l}{} \\

305:  1999 Quiescent  & $5.8_{-1.4}^{+1.3}$ & $2.5_{-0.7}^{+0.8}$ & \nodata             & \phd19/22 \\[1ex]

306:  2000 Quiescent  & $5.0_{-2.3}^{+1.9}$ & $1.8_{-0.9}^{+0.7}$ & \nodata             &    7.6/12 \\[1ex]

307:  2000 Flaring    & $4.6_{-1.6}^{+2.0}$ & \nodata             & $1.0_{-0.7}^{+0.8}$ & \phd12/17 \\[1ex]

308:  All             & $5.3_{-1.1}^{+0.9}$ & $2.2_{-0.7}^{+0.5}$ & $1.3_{-0.6}^{+0.5}$ & \phd45/55 \\

309: %%% 1999 Quiescent &  5.8 (4.4--7.1)     &  2.5 (1.8--3.3)     & \nodata             &    19.1/22 \\

310: %%% 2000 Quiescent &  5.0 (2.7--6.9)     &  1.8 (0.9--2.5)     & \nodata             & \phn7.6/12 \\

311: %%% 2000 Flaring   &  4.6 (3.0--6.6)     & \nodata             &  1.0 (0.3--1.8)     &    12.4/17 \\

312: %%% All            &  5.3 (4.2--6.2)     &  2.2 (1.5--2.7)     &  1.3 (0.7--1.8)     &    45.2/55 \\

313: \multicolumn{5}{l}{} \\

314: \hline

315: \end{tabular}

316: \end{center}

317: \parbox{0.484\textwidth}{

318: \scriptsize

319: \renewcommand{\baselinestretch}{0.7}

320: \noindent Best-fit parameter values and 90\% confidence intervals for

321: power-law models, corrected for photoelectric absorption and dust

322: scattering\protect\nolinebreak\cite{Predehl95}.  $N_{\rm H}$ is the

323: neutral hydrogen absorption column in units of $10^{22}$ H~atoms

324: cm$^{-2}$.  $\Gamma_{\rm q}$ and $\Gamma_{\rm f}$ are the

325: photon-number indices of the quiescent-state and peak flaring-state

326: spectra ($N(E) \propto E^{-\Gamma}$).  $\chi^2$ is the value of the

327: fit statistic for the best-fit model and $\nu$ is the number of

328: degrees of freedom in the fit.  The parameter values for the spectrum

329: marked ``All'' were derived by fitting an absorbed, dust-scattered,

330: double power-law model to the three spectra simultaneosly (see text

331: and Figure~\ref{fig:spectra}).}

332: \end{table}

333: 

334: If we view the outburst as a single event, the few-hundred-second

335: rise/fall timescales and the 10-ks duration are consistent with the

336: light-crossing and dynamical timescales for the inner $\la10$

337: Schwarzchild radii ($R_{\rm s} \equiv 2GM/c^2$) of the accretion flow

338: around a black hole of $2.6 \times 10^{6}$ solar masses; here $R_{\rm

339: s}$ is the radius of the black-hole event horizon (i.e., the boundary

340: at which the escape velocity equals the speed of light), $G$ is the

341: gravitational constant, $M$ is the mass of the black hole, and $c$ is

342: the speed of light.  While we cannot strictly rule out an unrelated

343: contaminating source as the origin of the flare (e.g., an X-ray

344: binary, for which little is known about such short-timescale,

345: low-luminosity events as we have detected; W.~Lewin, private

346: communication), this explanation seems unlikely since the

347: characteristic angular scales of the young and old stellar clusters

348: around \sgrastar\ are 5--20\arcsec\ (ref.~7), whereas the flaring

349: source lies within 1/3\arcsec\ of the radio position.  These clusters

350: contain up to a million-solar-masses worth of stars and stellar

351: remnants\nolinebreak\cite{Genzel00}; hence it is rather improbable

352: that there would be only one very unusual stellar X-ray source in the

353: image and that it would be fortuitously superposed on \sgrastar.

354: Furthermore, it is not clear that X-ray binaries can be easily formed

355: or long endure near \sgrastar, given the high velocity dispersion and

356: high spatial density of the stars in its deep gravitational potential

357: well\nolinebreak\cite{Davies98, Baganoff01}.

358: 

359: Strong, variable X-ray emission is a characteristic property of AGN;

360: factors of $\sim2$--3 variations on timescales ranging from minutes to

361: years are typical for radio-quiet AGN\nolinebreak\cite{Mushotzky93}.

362: Moderate- to high-luminosity AGN (i.e., Seyfert galaxies and quasars)

363: show a general trend of increasing variability with decreasing

364: luminosity\nolinebreak\cite{Nandra97}.  However, this trend does not

365: extend to low-luminosity active galactic nuclei (LLAGN), which show

366: little or no significant variability on timescales less than a

367: day\nolinebreak\cite{Ptak98}.  Assuming the X-ray flare is from

368: \sgrastar, it is remarkable that this source---generally thought to be

369: the nearest and least luminous example of accretion onto a central

370: supermassive black hole---has shown a factor of $\approx45$ variation

371: that is an order of magnitude more rapid than the fastest observed

372: variation of similar relative amplitude by a radio-quiet AGN of any

373: luminosity class\nolinebreak\cite{Ulrich97}.  We note that flares of

374: similar luminosity would be undetectable by \emph{Chandra} in the

375: nucleus of even the nearest spiral galaxy, M31.  LLAGN emit $L_{\rm X}

376: \ga 10^{38}$ erg s$^{-1}$ (ref.~14), so it should be kept in mind

377: that the astrophysics of accretion onto even the LLAGN may differ

378: substantially from that of \sgrastar.  This makes \sgrastar\ a

379: uniquely valuable source for testing the theory of accretion onto

380: supermassive black holes in galactic nuclei.

381: 

382: The faintness of \sgrastar\ at all wavelengths requires that the

383: supermassive black hole be in an extremely quiet phase, either because

384: the accretion rate is very low, or because the accretion flow is

385: radiatively inefficient, or both\nolinebreak\cite{Melia01c}.  A

386: variety of theoretical scenarios, usually based on advective accretion

387: models\nolinebreak\cite{Narayan98, Quataert99, Ball01, Blandford99},

388: jet-disk models\nolinebreak\cite{Falcke00}, or Bondi-Hoyle

389: models\nolinebreak\cite{Melia94, Melia01a}, have developed this idea.

390: An important prediction of the advective accretion models is that the

391: X-ray spectrum in the \emph{Chandra} energy band should be dominated

392: by thermal bremsstrahlung emission from hot gas in the outer regions

393: of the accretion flow ($R \ga 10^3\ R_{\rm s}$), but a region this

394: large could not produce the rapid, large-relative-amplitude variations

395: we have seen.  Thus, the properties of the X-ray flare are

396: inconsistent with the advective accretion flow models.  The low

397: luminosity and short timescales of this event are also inconsistent

398: with tidal disruption of a star by a central supermassive black

399: hole\nolinebreak\cite{Rees88}.

400: 

401: In all models, the radio-to-submillimetre spectrum of \sgrastar\ is

402: cyclo-synchrotron emission from a combination of sub-relativistic and

403: relativistic electrons (and perhaps positrons) spiralling around

404: magnetic field lines either in a jet or in a static region within the

405: inner $10\ R_{\rm s}$ of the accretion flow.  The electron Lorentz

406: factor inferred from the radio spectrum of \sgrastar\ is $\gamma_e

407: \sim 10^2$.  If the X-ray flare were produced via direct synchrotron

408: emission, then the emitting electrons would need $\gamma_e \ga 10^5$.

409: For the 10--100 G magnetic field strengths predicted by the models,

410: the cooling time of the particles would be $\sim1$--100~s.  Thus, the

411: $\approx10$-ks duration of the flare would require repeated injection

412: of energy to the electrons.  On the other hand, if the X-rays were

413: produced via up-scattering of the submillimetre photons off of the

414: relativistic electrons, a process called synchrotron

415: self-Comptonization (SSC), then $\gamma_e \sim 10^2$--$10^3$ would be

416: required, and the cooling time would be of order hours, which is

417: consistent with the duration of the flare.  The rapid turn-off of the

418: X-ray emission might then be attributed to the dilution of both photon

419: and electron densities in an expanding plasma.

420: 

421: The X-ray spectra of radio-quiet quasars and AGN are thought to be

422: produced by thermal Comptonization of infrared-to-ultraviolet seed

423: photons from a cold, optically thick, geometrically thin accretion

424: disk by hot electrons in a patchy corona above the

425: disk\nolinebreak\cite{Haardt97, Nandra97}.  The X-ray spectra of these

426: sources generally soften as they brighten\nolinebreak\cite{Ulrich97}.

427: In contrast, the extremely low luminosity of \sgrastar\ precludes the

428: presence of a standard, optically thick accretion

429: disk\nolinebreak\cite{Melia01c}; hence, the dominant source of seed

430: photons would be the millimetre-to-submillimetre synchrotron photons.

431: 

432: The energy released by an instability in the mass accretion rate or by

433: a magnetic reconnection event near the black hole would shock

434: accelerate the electrons, causing the synchrotron spectrum to

435: intensify and to extend farther into the submillimetre band.

436: Consequently, the Compton up-scattered X-ray emission would harden as

437: the X-ray intensity increased, exactly as observed.  We note that the

438: millimetre-band spectrum of \sgrastar\ has been observed to harden

439: during one 3-week flare\nolinebreak\cite{Tsuboi99} and one 3-day

440: flare\nolinebreak\cite{Wright93}, as would be required by the current

441: SSC models for \sgrastar\ (refs 20, 22).

442: 

443: To test the SSC models, we measured the flux density of \sgrastar\ at

444: a wavelength of 3~mm with the Millimeter Array at the Owens Valley

445: Radio Observatory, simultaneous with part of the 2000 \emph{Chandra}

446: measurement.  Unfortunately, the available observing window

447: (20:10--02:30 UT) preceded the X-ray flare (04:03--06:50 UT) by a few

448: hours.  The observed flux density of \sgrastar\ was $2.05 \pm 0.3$~Jy,

449: consistent with previously reported

450: measures\nolinebreak\cite{Serabyn97, Falcke98}.  Recently, a 106-day

451: quasi-periodicity has been reported in the centimetre band from an

452: analysis of 20 years of data taken with the Very Large Array

453: (VLA)\nolinebreak\cite{Zhao01}.  A weekly VLA monitoring program

454: detected an $\approx30$\% increase in the radio flux density of

455: \sgrastar\ beginning around 24 October 2000 and peaking on 5 November

456: 2000.  This increase was seen at 2~cm, 1.3~cm, and 7~mm (R.~McGary,

457: J.-H.~Zhao, W.~M.~Goss, and G.~C.~Bower, private communication).  The

458: timing of the X-ray flare and the rise in the radio flux density of

459: \sgrastar\ suggests there is a connection between the two events,

460: providing additional indirect support for the association of the X-ray

461: flare with \sgrastar\ and further strengthening the case that it was

462: produced via either the SSC or direct synchrotron processes.

463: Definitive evidence for these ideas will require detection of

464: correlated variations in the radio-to-submillimetre and X-ray

465: wavebands through future coordinated monitoring projects.

466: 

467: }

468: 

469: %%% [Bibliography: maximum of 30.]

470: 

471: \begin{thebibliography}{}

472: 

473: \bibitem{Richstone98} Richstone, D.\ \emph{et~al.}  Supermassive black

474: holes and the evolution of galaxies.  \emph{Nature} {\bf 395}, A14--19

475: (1998).

476: 

477: \bibitem{Genzel00} Genzel, R., Pichon, C., Eckart, A., Gerhard, O.~E.\

478: \& Ott, T.\ Stellar dynamics in the Galactic Centre: proper motions

479: and anisotropy.  \emph{Mon.\ Not.\ R.\ Astron.\ Soc.}  {\bf 317},

480: 348-374 (2000).

481: 

482: \bibitem{Ghez00} Ghez, A.~M., Morris, M., Becklin, E.~E., Tanner, A.,

483: \& Kremenek, T.\ The accelerations of stars orbiting the Milky Way's

484: central black hole.  \emph{Nature} {\bf 407}, 349--351 (2000).

485: 

486: \bibitem{Lynden-Bell71} Lynden-Bell, D.\ \& Rees, M.~J.\ On quasars,

487: dust and the Galactic Centre. \emph{Mon.\ Not.\ R.\ Astron.\ Soc.}

488: {\bf 152}, 461--475 (1971).

489: 

490: \bibitem{Melia01c} Melia, F.\ \& Falcke, H.\ The supermassive black

491: hole at the Galactic Center. \emph{Annu.\ Rev.\ Astron.\ Astrophys.}

492: {\bf 39}, (in the press).

493: 

494: \bibitem{Baganoff01} Baganoff, F.~K.\ \emph{et~al.} Chandra X-ray

495: spectroscopic imaging of \sgrastar\ and the central parsec of the

496: Galaxy. \emph{Astrophys.~J.} (submitted); also preprint

497: astro-ph/0102151 at $\langle$xxx.lanl.gov$\rangle$ (2001).

498: 

499: \bibitem{Morris96} Morris, M.\ \& Serabyn, E.\ The galactic center

500: environment.  \emph{Annu.\ Rev.\ Astron.\ Astrophys.} {\bf 34},

501: 645--702 (1996).

502: 

503: %%% \bibitem{Mezger96} Mezger, P.~G., Duschl, W.~J.\ \& Zylka, R.\ The

504: %%% Galactic Center: a laboratory for AGN?  \emph{Astron.\ Astrophys.\

505: %%% Rev.} {\bf 7}, 289--388 (1996).

506: 

507: \bibitem{Menten97} Menten, K.~M., Reid, M.~J., Eckart, A.\ \& Genzel,

508: R.\ The position of Sagittarius~A$^*$: accurate alignment of the radio

509: and infrared reference frames at the Galactic Center.

510: \emph{Astrophys.~J.} {\bf 475}, L111--L114 (1997).

511: 

512: \bibitem{Weisskopf96} Weisskopf, M.~C., O'Dell, S.~L.\ \& van

513: Speybroeck, L.~P.\ Advanced X-Ray Astrophysics Facility (AXAF).

514: \emph{Proc.\ SPIE} {\bf 2805}, 2--7 (1996).

515: 

516: \bibitem{Predehl95} Predehl, P.\ \& Schmitt, J.~H.~M.~M.\ X-raying the

517: interstellar medium: ROSAT observations of dust scattering halos.

518: \emph{Astron. Astrophys.} {\bf 293}, 889--905 (1995).

519: 

520: \bibitem{Mushotzky93} Mushotzky, R.~F., Done, C.\ \& Pounds, K.~A.\

521: X-ray spectra and time variability of active galactic nuclei.

522: \emph{Annu.\ Rev.\ Astron.\ Astrophys.} {\bf 31}, 717-761 (1993).

523: 

524: \bibitem{Davies98} Davies, M.~B., Blackwell, R., Bailey, V.~C.\ \&

525: Sigurdsson, S.\ The destructive effects of binary encounters on red

526: giants in the Galactic Centre.  \emph{Mon.\ Not.\ R.\ Astron.\ Soc.}

527: {\bf 301}, 745--753 (1998).

528: 

529: \bibitem{Nandra97} Nandra, K., George, I.~M., Mushotzky, R.~F.,

530: Turner, T.~J.\ \& Yaqoob, T.\ ASCA observations of Seyfert~1

531: galaxies. I.~data analysis, imaging, and timing.  \emph{Astrophys.~J.}

532: {\bf 476}, 70--82 (1997).

533: 

534: \bibitem{Ptak98} Ptak, A., Yaqoob, T., Mushotzky, R., Serlemitsos, P.\

535: \& Griffiths, R.\ X-ray variability as a probe of advection-dominated

536: accretion in low-luminosity active galactic nuclei.

537: \emph{Astrophys.~J.} {\bf 501}, L37--L40 (1998).

538: 

539: \bibitem{Ulrich97} Ulrich, M.-H., Maraschi, L.\ \& Urry, C.~M.\

540: Variability of active galactic nuclei.  \emph{Annu.\ Rev.\ Astron.\

541: Astrophys.} {\bf 35}, 445--502 (1997).

542: 

543: \bibitem{Narayan98} Narayan, R., Mahadevan, R., Grindlay, J.~E.,

544: Popham, R.~G.\ \& Gammie, C.\ Advection-dominated accretion model of

545: Sagittarius~A$^*$: evidence for a black hole at the Galactic center.

546: \emph{Astrophys.~J.} {\bf 492}, 554--568 (1998).

547: 

548: \bibitem{Quataert99} Quataert, E.\ \& Narayan, R.\ Spectral models of

549: advection-dominated accretion flows with winds.  \emph{Astrophys.~J.}

550: {\bf 520}, 298-315 (1999).

551: 

552: \bibitem{Ball01} Ball, G.~H., Narayan, R.\ \& Quataert, E.\ Spectral

553: models of convection-dominated accretion flows.  \emph{Astrophys.~J.}

554: {\bf 552}, 221--226 (2001).

555: 

556: \bibitem{Blandford99} Blandford, R.~D.\ \& Begelman, M.~C.\ On the

557: fate of gas accreting at a low rate on to a black hole.  \emph{Mon.\

558: Not.\ R.\ Astron.\ Soc.} {\bf 303}, L1--L5 (1999).

559: 

560: \bibitem{Falcke00} Falcke, H.\ \& Markoff, S.\ The jet model for

561: Sgr~A$^*$: radio and X-ray spectrum.  \emph{Astron.\ Astrophys.}  {\bf

562: 362}, 113--118 (2000).

563: 

564: \bibitem{Melia94} Melia, F.\ An accreting black hole model for

565: Sagittarius~A$^*$. II: A detailed study.  \emph{Astrophys.~J.} {\bf

566: 426}, 577--585 (1994).

567: 

568: \bibitem{Melia01a} Melia, F., Liu, S.\ \& Coker, R.\ A magnetic dynamo

569: origin for the submillimeter excess in Sagittarius~A$^*$.

570: \emph{Astrophys.~J.} {\bf 553}, 146--157 (2001).

571: 

572: \bibitem{Rees88} Rees, M.~J.\ Tidal disruption of stars by black holes

573: of $10^6$--$10^8$ solar masses in nearby galaxies.  \emph{Nature} {\bf

574: 333}, 523--528 (1988).

575: 

576: \bibitem{Haardt97} Haardt, F., Maraschi, L.\ \& Ghisellini, G.\ X-ray

577: variability and correlations in the two-phase disk-corona model for

578: Seyfert galaxies.  \emph{Astrophys.~J.} {\bf 476}, 620--631 (1997).

579: 

580: \bibitem{Tsuboi99} Tsuboi, M., Miyazaki, A.\ \& Tsutsumi, T.\ Flare of

581: Sgr~A$^*$ at short millimeter wavelengths.  in \emph{The Central

582: Parsecs of the Galaxy} (ed.\ Falcke, H.\ \emph{et~al.}) {\bf 186},

583: 105--112 (ASP Conf.\ Ser., Astron.\ Soc.\ Pac., San Francisco, 1999).

584: 

585: \bibitem{Wright93} Wright, M.~C.~H.\ \& Backer, D.~C.\ Flux density of

586: Sagittarius~A at $\lambda = 3$ millimeters.  \emph{Astrophys.~J.}

587: {\bf 417}, 560--564 (1993).

588: 

589: \bibitem{Serabyn97} Serabyn, E.\ \emph{et~al.} High frequency

590: measurements of the spectrum of SGR A$^*$.  \emph{Astrophys.~J.} {\bf

591: 490}, L77--L81 (1997).

592: 

593: \bibitem{Falcke98} Falcke, H.\ \emph{et~al.} The simultaneous spectrum

594: of Sagittarius~A$^*$ from 20~centimeters to 1~millimeter and the

595: nature of the millimeter excess.  \emph{Astrophys.~J.} {\bf 499},

596: 731--734 (1998).

597: 

598: \bibitem{Zhao01} Zhao, J.-H., Bower, G.~C.\ \& Goss, W.~M.\ Radio

599: variability of Sagittarius~A$^*$---a 106 day cycle.

600: \emph{Astrophys.~J.} {\bf 547}, L29--L32 (2001)

601: 

602: \bibitem{Reid93} Reid, M.~J.\ The distance to the center of the

603: Galaxy.  \emph{Annu.\ Rev.\ Astron.\ Astrophys.} {\bf 31}, 345--372

604: (1993).

605: 

606: \end{thebibliography}

607: 

608: \medskip

609: 

610: {\small\bf\noindent Acknowledgements}

611: 

612: \smallskip

613: 

614: {\footnotesize\noindent We thank M.~Begelman for useful comments.

615: This work has been supported by a grant from NASA.

616: 

617: \smallskip

618: 

619: \noindent Correspondence and requests for material should be addressed

620: to F.K.B. (e-mail: fkb@space.mit.edu).}

621: 

622: \end{document}

623: 

624: