1: \documentclass{aastex}

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

3: \usepackage{graphicx}

4: 

5: \usepackage{emulateapj5}

6: \usepackage{times}

7: 

8: \makeatletter

9: \newenvironment{inlinetable}{%

10: \def\@captype{table}%

11: \noindent\begin{minipage}{0.999\linewidth}\begin{center}\footnotesize}

12: {\end{center}\end{minipage}\smallskip}

13: 

14: \newenvironment{inlinefigure}{%

15: \def\@captype{figure}%

16: \noindent\begin{minipage}{0.999\linewidth}\begin{center}}

17: {\end{center}\end{minipage}\smallskip}

18: \makeatother

19: 

20: 

21: 

22: %\usepackage{mathptmx}

23: 

24: % \usepackage{mathtime}

25: % \usepackage[parens,letters,alpha]{fixmathtimebelleek}

26: % \usepackage{timesexpert}

27: % \def\altaffilmark#1{\mbox{\hskip0.05em\normalfont\fontshape{fn}\selectfont#1}}

28: 

29: 

30: \def\keV{ke\kern-0.05emV}

31: \newcommand{\chandra}{\emph{Chandra}}

32: \newcommand{\chandraintitle}{{\itshape\footnotesize CHANDRA}\/}

33: \newcommand{\ROSAT}{\textsl{ROSAT}}

34: \newcommand{\NGC}{NGC\,}

35: \renewcommand{\arcsec}{\ensuremath{''}}

36: \renewcommand{\mp}{m_{\!p}}

37: 

38: %\def\baselinestretch{1.3}

39: 

40: \def\gsim{\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$>$}}}}

41: \def\lsim{\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$<$}}}}

42: \def\asca       {{\em ASCA}\/}

43: \def\chandra    {{\em Chandra}\/}

44: \def\rxj1720    {{ RXJ1720.1+2638}\/}

45: \def\ginga      {{\em Ginga}\/}

46: \def\xmm        {{\em XMM}\/}

47: \def\rosat      {{\em ROSAT}\/}

48: \def\am         {$^\prime$}

49: \def\as         {$^{\prime\prime}$}

50: \def\cmsq       {~cm$^{-2}$}

51: \def\kms        {~km$\;$s$^{-1}$}

52: \def\kmsmpc     {~km$\;$s$^{-1}\,$Mpc$^{-1}$}

53: \def\msunyr     {$M_{\odot}\;$yr$^{-1}$}

54: \def\degd       {$^{\circ}\!$}

55: \def\second     {{\prime\prime}}

56: \def\ms1455     {{MS 1455.0+2232}}

57: 

58: 

59: 

60: \begin{document}

61: 

62: \submitted{2002, ApJL, 569, 31}

63: 

64: \title{\chandra ~ Observation of a $300$~kpc Hydrodynamic  

65: Instability in the Intergalactic Medium of the Merging Cluster of Galaxies A3667}

66: 

67: 

68: \author{Pasquale\ Mazzotta\altaffilmark{1,2}, Roberto\ Fusco-Femiano

69: \altaffilmark{3},

70:  Alexey\ Vikhlinin \altaffilmark{1}}

71: \altaffiltext{1}{Harvard-Smithsonian Center for Astrophysics, 60 Garden St.,

72: Cambridge, MA 02138; mazzotta@cfa.harvard.edu}

73: \altaffiltext{2}{Department of Physics, University of Durham,

74: South Road, Durham DH1 3LE}

75: \altaffiltext{3}{Istituto Astrofisica Spaziale, Area CNR Tor Vergata,

76: via del Fosso del Cavaliere, 00133 Roma (Italy)}

77: 

78: \begin{abstract}

79:   

80:   We present results from the combination of two \chandra ~ pointings of the

81:   central region of the cluster of galaxies A3667. From the data analysis of

82:   the first pointing Vikhlinin et al. reported the discovery of a

83:   prominent cold front which is interpreted as the boundary of a cool gas

84:   cloud moving through the hotter ambient gas. 

85: Vikhlinin et al. discussed the

86:     role of the magnetic fields in maintaining the apparent dynamical

87:     stability of the cold front over a wide sector at the forward edge of the

88:   moving cloud

89:  and suppressing

90:  transport processes across the front.  

91:   In this Letter, we identify two new 

92:   features in the X-ray image of A3667: i) a 300 kpc arc-like

93:   filamentary X-ray excess extending from the cold gas cloud border into the

94:   hotter ambient gas; ii) a similar arc-like filamentary X-ray depression

95:   that develops inside the gas cloud.  Both features are located beyond the

96:   sector identified by the cold front and are oriented in a direction

97:   perpendicular to the direction of motion. The temperature map suggests

98:   that the temperature of the filamentary excess is consistent with that

99:   inside the gas cloud while the temperature of the depression is consistent

100:   with that of the ambient gas.  We suggest that the observed features

101:   represent the first evidence for the development of a large scale

102:   hydrodynamic instability in the cluster atmosphere resulting from a major

103:   merger.  

104:   This result confirms previous claims for the

105:   presence of a moving  cold gas cloud into the 

106:   hotter ambient gas.

107:   Moreover it shows that, although the gas mixing

108:   is suppressed at the leading edge of the subcluster due to 

109:   its magnetic structure, strong turbulent mixing 

110:   occurs at larger angles to the direction of motion.

111:   We show that this mixing process may favor the deposition of a 

112:   nonnegligible quantity of thermal energy right in the cluster 

113:   center,  affecting the development of the central cooling flow.

114:  

115: 

116: 

117: 

118: \end{abstract}

119: 

120: \keywords{galaxies: clusters: general --- galaxies: clusters: individual

121:   (A3667) --- magnetic fields --- shock waves --- intergalactic medium --- X-rays: galaxies: cluster --- cooling flows --- instabilities --- MHD --- turbulence}

122: 

123: 

124: 

125: \section{Introduction}

126: 

127: The central region of A3667, a nearby, hot merging cluster (Markevitch,

128:   Sarazin, \& Vikhlinin 1999), was observed for the first time by \chandra

129: ~ in Sept 1999.  

130: The analysis of this

131:   first observation by Vikhlinin, Markevitch \& Murray (2001a,b) reveals

132: the presence of a prominent 500~kpc-long density

133:   discontinuity (``cold front'') in the cluster atmosphere. 

134: Vikhlinin et al.\ show that: 

135: i) the density discontinuity is the

136: contact surface between the moving 

137: cloud of 4~keV gas

138: and the hotter ambient gas; ii) the speed of the moving

139: cloud is slightly supersonic.

140:   

141: An interesting aspect raised by 

142: Vikhlinin et al. is that 

143: the cold front

144: must develop hydrodynamical instabilities. 

145: These instabilities would destroy the front on very short time

146: scales. 

147: However, the \chandra ~ image shows that

148: the front is stable in a wide, $\varphi =\pm 30^\circ$ sector (where $\varphi$ 

149: is the angle with respect to the direction of motion). Moreover,

150: the front width is smaller than the Coulomb mean free path, which indicates

151: that the transport and mixing processes are suppressed (see also Ettori \&

152: Fabian 2000). To reconcile this observational evidence with the

153: expected hydrodynamical instability of the cold front, 

154: Vikhlinin et al. (2001b) put forward the

155: following scenario. As the cloud moves through the ambient gas, the

156: magnetic field lines, which are frozen in the intracluster medium, are

157: stretched by the tangential plasma motions. This results in the formation of

158: a layer with the ordered magnetic field parallel to the front surface. The

159: magnetic field intensity in such a layer is sufficient to suppress the

160: hydrodynamical instabilities within the $30^\circ$ sector, and to prevent gas

161: mixing and transport processes.

162: 

163: 

164: 

165: \begin{inlinefigure}

166: \centerline{\includegraphics[width=0.95\linewidth]{f1.eps}}

167: \caption{Count rate image in the $0.7-4$~keV band binned by $4^\second$.

168: The arrows indicate the prominent X-ray features: the cold front discussed by Vikhlinin et al. (2001a,b), a filamentary arc-like surface brightness excess 

169: extending toward the east, and a filamentary arc-like surface brightness depression extending toward the west.}   

170: \label{fig:image}

171: \end{inlinefigure}

172: 

173: 

174: In this letter we present the results from the joint analysis of two

175:  \chandra ~ observations of A3667. We identify two 

176: filamentary structures forming at the

177: border of the merging subclump: one extends toward the outskirts, 

178: the other toward the cluster center.

179:  Both structures lie well beyond the  hydrodynamically stable region

180: identified by Vikhlinin et al. (2001a).  

181: We speculate that these filaments correspond to the well-developed K-H

182: instability, and discuss their implications for the proposed dynamical model

183: of A3667. 

184: 

185: 

186: The physical size of the structures under

187:   discussion is computed assuming $H_0=50$~km~s$^{-1}$~kpc$^{-1}$ 

188: ($1$~arcsec$\approx 1.46$~kpc). Unless

189: specified differently, all the errors are at $90\%$ confidence level for one

190: interesting parameter.

191: 

192: 

193: 

194: \begin{inlinefigure}

195: \centerline{\includegraphics[width=0.95\linewidth]{f2.eps}}

196: \caption{Adaptively smoothed \chandra ~ image. 

197: The four lines identify the External Sector (ES) and the Internal Sector (IS)

198: used to derive the surface brightness and temperature profiles reported in Fig.~\ref{fig:p_dep}. The position angles are from -15\degd ~ to 15\degd ~ and from -15\degd ~ to 40\degd ~ for IS and ES, respectively 

199: (angles are measured from North through East). 

200: Regions 4 and 2 identify the X-ray excess and depression, respectively. Regions 1 and 3 are two 

201: circular $r=200$~kpc regions 

202: centered

203: at the center of curvature

204:  of each filamentary structure. 

205: The actual temperatures of the regions, indicated by cardinal numbers 

206: 1 to 4 are, reported in 

207: Fig.~\ref{fig:t_reg}.}

208: \label{fig:panda}

209: \end{inlinefigure}

210: 

211: \section{Data Reduction}\label{par:data} 

212: 

213: 

214: A3667 was observed twice by \chandra ~ with the ACIS-I detector on

215:   Sept 22, 1999 and on Sept 9, 2000 for useful exposure times of 45~ks and

216:   50~ks, respectively. We used the data analysis procedures already

217:   described elsewhere (Markevitch et al.\ 2000,

218:   Vikhlinin et al.\ 2001a, Markevitch \& Vikhlinin 2001, and

219:   Mazzotta et al.\ 2001 ). We only note that, since the observations have been

220:   performed at different detector temperatures, the spectral response and

221:   background data have been generated individually for each observation and

222:   then appropriately combined.

223:   

224:   Spectral analysis has been performed in the 1--9~keV energy band in PI

225:   channels. 

226:   The spectral fitting has been

227:   performed assuming an absorption fixed at the Galactic value

228:   ($N_H=4.1\times 10^{20}$~cm$^{-2}$), and a fixed plasma metallicity,

229:   $Z=0.3\ Z_\odot$. Because of the hard energy band we used, the derived

230:   plasma temperatures are insensitive to the precise value of either $N_H$

231:   or $Z$. The temperature map has been produced with a fixed Gaussian

232:   smoothing of $\sigma=40^\second$ (see Vikhlinin et al.\ 2001a for the details

233:   of the temperature map computation).

234: 

235: 

236: 

237: 

238: 

239: \section{X-ray image and Temperature map}

240: 

241: The combined

242:   background subtracted and vignetting corrected 

243:   \chandra~ image in the 0.7-4~keV energy band is

244: shown in Fig.~\ref{fig:image}.

245: We extracted the image in the 0.7-4~keV band

246: to minimize the fractional contribution of the cosmic background and 

247: thereby to maximize the signal-to-noise ratio.  

248: The strongest X-ray feature present in

249: Fig.~\ref{fig:image} is the sharp surface brightness edge (``the cold

250: front'') to the South-East extensively 

251: discussed by Vikhlinin et al. (2001a,b). To the North of the cold front

252: the cluster image shows two new interesting X-ray features:

253: 

254: i) a filamentary X-ray excess extending toward the east to the chip boundary.

255:   The filament has an arc-like shape with a curvature

256: radius of $\approx 200$~kpc, is $\approx 300$~kpc long and $\approx 90$~kpc

257: wide;

258: 

259: ii) a filamentary X-ray depression that develops toward the west inside the

260: cluster center. Also this feature has an arc-like

261: shape with a curvature radius of $\approx 200$~kpc. It appears to be

262: $\approx 200$~kpc long and $\approx 80$~kpc wide;

263: 

264: \begin{inlinefigure}

265: \centerline{\includegraphics[width=0.95\linewidth]{f3.eps}}

266: \caption{Temperature map  with overlaid ACIS-I X-ray

267: surface brightness contours (spaced by a factor of $\sqrt{2}$) in the

268: 0.7-4~keV energy band after adaptive smoothing.   The black cut-out regions identify

269:   the point sources that were masked out.

270: The regions indicated by cardinal numbers 1 to 4 are  the same shown in 

271: Fig.~\ref{fig:t_reg}.}

272: \label{fig:tmap}

273: \end{inlinefigure}

274: 

275: \begin{inlinefigure}

276: \centerline{\includegraphics[width=0.95\linewidth]{f4.eps}}

277: \caption{Projected emission-weighted temperatures in the

278:  corresponding regions shown in  Fig.~\ref{fig:panda} and  Fig.~\ref{fig:tmap}

279: ($90\%$ confidence level error bars).}

280: \label{fig:t_reg}

281: \end{inlinefigure}

282: 

283: 

284: A close-in view of the filamentary X-ray features is shown in the 

285: adaptively smoothed image in Fig.~\ref{fig:panda}. We select two ellipsoidal regions indicated by the

286: cardinal numbers 4 and 2 to identify the X-ray excess and depression,

287: respectively.  To derive the surface brightness and temperature

288: profiles (see \S~\ref{par:prof}, below) for each feature, we identified two

289: sectors originating near the curvature center of the corresponding feature

290: and containing the feature itself.  We refer to the sector for the excess

291: and the depression as the External Sector (ES) and the Internal Sector (IS),

292: respectively.  Finally with the cardinal  numbers 1 and 3 we identify

293: two circular $r=200$~kpc regions centered at the origin of the ES and IS,

294: respectively.

295:  

296: In Fig.~\ref{fig:tmap} we report the temperature map with  overlaid X-ray

297: surface brightness contours. 

298: The temperature map 

299: suggests that, while the X-ray excess (region 4) corresponds 

300: to a local decrease in the projected temperature,

301: the  temperature of the X-ray depression (region 2) 

302: is higher than in the nearby regions. 

303:   To quantify the statistical significance of the features in

304:   the temperature map, we fitted the spectra from the four

305: selected regions with a single temperature model.  

306: The spectra from region 2 and 4 have 5155 and 3641 net photons 

307: plus  575 and 705 background photons, respectively.

308: The best-fit temperatures reported in

309:   Fig.~\ref{fig:t_reg} show

310: that while the projected temperature of the gas from the 

311: filamentary excess is consistent with the temperature of 

312: the cold subcluster, the temperature of the 

313:   filamentary depression is consistent with the temperature of the

314: ambient cluster gas.

315: 

316: 

317: 

318: Finally, we note that, in the

319: overlapping region, 

320: the Chandra temperature map qualitatively agrees  

321: with the coarser ASCA temperature map shown by 

322: Markevitch et al. (1999). 

323: 

324: \begin{inlinefigure}

325: \centerline{\includegraphics[width=0.95\linewidth]{f5.eps}}

326: \caption{X-ray surface brightness (upper panels) and temperature  (lower panels)

327: profiles from the ES (left column) and IS (right column), 

328: defined in Fig.~\ref{fig:panda}. 

329: The dashed line in the left column indicates the relative maximum in the 

330: ES X-ray surface brightness profile, while 

331: the dashed line in the right column indicates the relative minimum in the 

332: IS X-ray surface brightness profile.

333: The x-axis    indicates the distance in kpc from the surface brightness profile stationary point 

334: of the corresponding sector.

335: Error bars are at $68\%$ confidence level.}

336: \label{fig:p_dep}

337: \end{inlinefigure}

338: 

339: 

340: \subsection{Surface Brightness and Temperature Profiles}\label{par:prof}

341: 

342: To verify the statistical significance of the observed features we extracted

343: both the surface brightness and the projected emission-weighted temperature

344: profiles.  The profiles are shown in the left and the right columns of

345: Fig.~\ref{fig:p_dep} for the ES and IS, respectively.  The profiles confirm

346: that the observed features are statistically significant.  The ES profiles

347: show that the filamentary excess produces a significant enhancement of the

348: surface brightness profile and corresponds to a decrement in the temperature

349: profile.  On the other hand, the IS profiles shows that the filamentary

350: depression produces a significant decrement of the surface brightness

351: profile corresponding to an increment in the temperature profile.  This

352: suggests that the excess is a colder, denser filamentary gas structure

353: embedded in the more diffuse and hotter external cluster atmosphere while

354: the depression is a hotter, rarefied filament of gas embedded in the denser

355: and colder cluster core. 

356: 

357: 

358: 

359: 

360: 

361: 

362: \section{Discussion}\label{par:discussion}

363: 

364: 

365: We have presented 

366: evidence for the existence of two adjacent

367: filamentary arc-like structures in the X-ray image of A3667: an X-ray excess

368: extending toward the outskirts and an X-ray depression extending toward the

369: cluster center.  Both features are present in the soft band as well as in

370: the hard band. This indicates that they are produced by significant gas

371: density deviations with respect to the surrounding gas.  In particular, the

372: former indicates a denser filamentary gas structure embedded in the more

373: diffuse external cluster atmosphere while the latter indicates the presence

374: of a rarefied filament of gas embedded in the denser cluster core.  

375:  The

376: temperature analysis clearly shows that the dense external filament

377: is colder and the rarefied

378: internal filament is hotter than the ambient gas,

379: respectively.  We show also that, while the projected temperature of the gas

380: from the external filament is consistent with the temperature of the central

381: region, the temperature of the internal filament is consistent with the

382: temperature of the *ambient* external cluster gas (see Fig.~\ref{fig:t_reg}).

383: The observed features suggest that some dense cold gas from the cluster core

384: is ``stripped out'' into the hotter cluster atmosphere as well as some

385: rarefied hot gas from larger radii penetrates 

386: into the cluster core.  This observation, thus, represents good evidence for

387: ongoing turbulent gas mixing in the cluster atmosphere.

388: 

389: 

390: In  previous work, Vikhlinin et al. (2001,a,b) 

391: found that the gas in the  $r\approx 300$~kpc 

392: central, colder region of the cluster moves from North-West to

393: South-East into the rarefied hotter cluster atmosphere,

394: and the cold front is the the contact surface between

395: the two 

396: gases.  They 

397: determined the motion speed from the pressure jump at the

398: front and found the near-sonic velocity $M\equiv v/v_s \gsim 1$,

399: where $v_s$ is the speed of sound in the hot

400: ambient gas.

401: During the gas cloud motion, the surrounding  gas flows around the cold cloud.

402: In this situation, the interface between the tangentially moving gas layers

403: must develop both Rayleigh-Taylor (R-T) and Kelvin-Helmholtz (K-H)

404: instabilities (see Inogamov, 1999 for a review).

405: 

406: 

407: 

408: 

409: The R-T instability develops at the interface  between the two fluids when the

410: rarefied fluid accelerates the denser one.  If, as it is reasonable to

411: assume, the gas cloud moves together with its own dark matter halo 

412: (Vikhlinin \& Markevitch 2002, in preparation)

413: the gas cloud is stabilized against the R-T instability by 

414: gravity, so that the development of R-T

415: instability is quite unlikely.  In contrast the K-H instability is related

416: to the shearing motion at the boundary between two fluids and is expected to

417: develop along the lateral boundaries of the cloud. 

418: The wavevector  of the fastest

419: growing mode, $\lambda$, is parallel to the flow and its growing time $\tau$ is given by

420: solving the dispersion equation (see e.g. Miles 1958).

421: Vikhlinin et al.\ (2001b) computed the K-H instability growth time for the

422: flow near the cold front in A3667. 

423: As the only perturbations relevant for our discussion are the ones that 

424: grow on time scales shorter than the cluster core

425: passage time, $t_{cross}=L/v_{cross}$ (here $L$ is the cluster size and

426: $v_{cross}$ is the motion speed), we derived the ratio between 

427: these two times\footnote{This equation reproduces eq.(4)

428:   from Vikhlinin et al.\ (2001b), corrected for an algebraic error which

429:   resulted in overestimation of $\tau$ by a factor of $4\pi^2$. Note that

430:   the correct equation does not affect any results presented in Vikhlinin et

431:   al., and indeed, strengthens their arguments.}:

432: \begin{equation}

433:   \frac{t_{cross}}{\tau}=3.3 {L \over \lambda} \sin \varphi,

434:   \label{eq:1}

435: \end{equation}

436: where $\varphi$ is the angle between the perturbation and the leading edge

437: of the moving cloud.  

438: 

439: An important aspect already discussed by Vikhlinin et al.\ (2001b) is that, 

440: in the absence of stabilizing factors and assuming $L\approx 1$~Mpc,

441:  the growth time for all  perturbations on scales $\lsim 10$~kpc 

442: is much shorter

443:  than the cluster passage time  even at very small angles 

444:  ($\varphi \gsim 10-50$~arcmin).

445:  This means that  there should develop a turbulent layer which

446:  would smear the cold front by at least $\sim 10$~kpc at angles 

447:  $\varphi \gsim 10-50$~arcmin.  

448:  The Chandra data, however,  exclude such a smearing within 

449:  the $\pm30^\circ$ sector of the leading

450:  edge of the cold front. 

451:  The expected smearing appears evident only at angles 

452:  $\varphi > 30^\circ$ (see e.g. Fig.~\ref{fig:image}). 

453:  Vikhlinin et al.\ propose that the ambient magnetic field is amplified

454:  in a narrow

455:  boundary region between the two moving gas layers.

456:  The amplification is the result of the stretching of the magnetic field lines

457:  along the front by tangential plasma motions.

458:  The surface tension of the amplified magnetic 

459:  field acts to stabilize the development of the K-H instability in the 

460:  $\pm30$\degd ~ sector. At larger angles, however, the magnetic field 

461:  surface tension becomes insufficient to stabilize the front because 

462:  of the higher flow speed.  Thus, outside the $\pm30^\circ$

463:  sector, the development of the K-H instability is unaffected by the magnetic

464:  field, and therefore the perturbation growth time is given by

465:  eq.~\ref{eq:1}.

466:  Eq.~\ref{eq:1} gives us also two important clues to understand the presence

467:  of the observed

468: features on the side of the colder moving subcluster:

469: 

470: i) small scale perturbations

471: grow faster than the larger scale ones.  This means that small scale

472: perturbations develop immediately, when the interface is still a

473: discontinuity. Their growth, however, widens the interface, which becomes a

474: turbulent layer of finite width.  At this

475: point the perturbations on scales smaller than the evolving width of the

476: front are damped (they becomes unobservable as individual

477: structures) while perturbations on a larger scale start to

478: grow (thus they may appear as distinct  structures) 

479: (see e.g. Esch 1957). As this process

480: continues, it is expected that the remaining observable wavelengths are the

481: ones with a growth time comparable to the cloud crossing time, namely

482: $t_{cross}/\tau\sim 1-10$;

483: 

484: ii) for a fixed wavelength the growth time is shorter at larger angles

485: $\varphi$ reaching its minimum at $\varphi=90$\degd .  This is a direct

486: consequence of the fact that the speed of the external fluid increases with

487: $\varphi$ being maximum at $\varphi=90$\degd .

488:   

489: In this picture the most natural explanation for the newly discovered

490: filamentary structures in A3667 is that they are the result

491: of the development of K-H instabilities.  

492: The structure lies, in fact, at $\varphi \approx 90$\degd ~

493: which,  because of point  

494: ii) above, is indeed a privileged point for the growth

495: of the perturbations.

496: Furthermore, if we assume that $\lambda=300$~kpc and

497: $L=1$~Mpc, the growing time for this perturbation 

498: is such that $t_{cross}/\tau\approx 10$. This means

499: that the perturbation is just now entering the strong non-linear regime,

500:  consistently with what we observe.

501: 

502:  

503: We notice that  the observed instability act to:

504: i) deposit low density hot gas right in the cluster center; 

505: ii) remove denser colder gas from the cluster center. 

506: If efficient, these two processes 

507: may contribute to inhibit (or, if already present, to destroy)

508: the development of a central cooling flow.

509: We calculated the  mass rate of 

510: hot gas which is being deposited right 

511: in  the cluster center.

512: For the scope of this letter we use a very simple  model and  

513: we assume that the external hot gas flows toward the center 

514: in a cylinder-like filament with a speed $v$ close to the speed

515: of the subclump ($v\approx M v_c\approx 1400{\rm km~s}^{-1}$).

516: Moreover, we assume that the gas density $n_0$ in the filament is equal to 

517: the density outside the front and that the radius

518: of the filament is $r=40$~kpc. We Find:

519: \begin{equation}

520: \dot M=108 

521: \Big(\frac{r}{40\rm{kpc}}\Big)^2\Big(\frac{v}{1400\rm{km/s}}\Big)\Big(\frac{n_0}{8\times 10^{-4}/\rm{cm}^{3}}\Big)\frac{\rm{M}_\odot}{\rm{yr}}

522:   \label{eq:2}

523: \end{equation}

524: From Eq.~\ref{eq:2} we see that the  

525: instability  transports hot gas in the cluster center at a rate which is comparable to the  

526: mass deposition rate of many cooling flow clusters.

527:  It is clear, than, that such a mechanism may play an important role 

528: in the development and/or evolution of a central cooling flow.

529: 

530:  

531: \section{Conclusion}

532: 

533: We presented the combination of two \chandra ~ observations of the central

534: region of A3667.  We showed that the cluster

535: hosts two arc-like adjacent filamentary structures: one,

536: extending from the colder subcluster toward the cluster

537: outskirts, appears as a dense structure embedded in the less dense cluster

538: atmosphere, and the other, extending inside the

539:   subcluster, appears as a rarefied structure

540: embedded in the denser cluster core.

541: 

542: We suggest that the observed features represent the first evidence for the

543: development of a large scale hydrodynamic instability in the cluster

544: atmosphere.  This interpretation appears to be consistent with the

545: previous cluster dynamic interpretation proposed by Vikhlinin et al.

546: (2001a,b).   

547: 

548: The discovery of this instability represents an important step toward

549: a full understanding of the physics of mergers.

550: In particular it shows that, although

551: the cold front prevents the gases of the merging objects to mix along the

552: direction of motion, strong turbulent mixing processes, on scales comparable

553: to the size of the merging subclump, may occur at large angles to the

554: direction of motion. 

555: This may favor the deposition of a non-negligible quantity of

556: thermal energy right in the 

557: cluster center  with important consequences for the development and/or 

558: evolution  of a central  cooling flow.

559: 

560: %

561: %

562: \acknowledgments

563: %We thank ............

564: We thank M. Markevitch and W. Forman for useful comments and suggestions.

565: P.M. acknowledges an ESA fellowship and thanks the Center for

566: Astrophysics for its hospitality. 

567: Support for this study was provided

568: by NASA contract NAS8-39073, grant NAG 5-9217, and by the Smithsonian

569: Institution. 

570: 

571: 

572: \begin{references}

573: 

574: %\reference{} Gerwin, R. A. 1968, Rev. Mod. Phys., 40, 652

575: 

576: \reference{} Esch, R.E. 1957, J. Fluid Mech. 3, 289

577: 

578: \reference{} Ettori, S.~\& Fabian, A.~C.\ 2000, \mnras, 317, L57

579: 

580: \reference{} Inogamov, N.~A.\ 1999, Astrophysics and Space Physics Reviews, 10, 1

581: 

582: 

583: \reference{} Mazzotta, P., Markevitch, M., Vikhlinin, A., Forman, W.~R., David, L.~P., \& VanSpeybroeck, L.\ 2001, \apj, 555, 205

584: 

585: 

586: \reference{} Markevitch, M., Sarazin, C. L., \& Vikhlinin, A. 1999, ApJ, 521, 526 

587: 

588: \reference{} Markevitch, M.~et al.\ 2000, \apj, 541, 542

589: 

590: \reference{} Markevitch, M., \&  Vikhlinin, A. 2001, ApJ, 563, 95

591: 

592: \reference{} Miles, J. W. 1958, J. Fluid Mech., 4, 538 

593: 

594: \reference{} Vikhlinin, A., Markevitch, M., Murray, S.S. 2001a, ApJ, 551, 160

595: 

596: \reference{} Vikhlinin, A., Markevitch, M., Murray, S.S. 2001b, \apjl, 549, L47

597: 

598: 

599: 

600: 

601: \end{references}

602: 

603: 

604: 

605: 

606: 

607: \end{document}

608: 

609: