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

2: %% preprint produces a one-column, single-spaced document:

3: 

4: %\documentclass[preprint2]{aastex}

5: %% preprint2 produces a double-column, single-spaced document:

6: 

7: \documentclass{aastex}

8: \usepackage{emulateapj5}

9: \usepackage{apjfonts}

10: %% Produces an emulated ApJ-style looking paper

11: 

12: %-------------------------------------------------------------------------

13: % Personal Definitions

14: %-------------------------------------------------------------------------

15: \newcommand{\ea}{{\it et al.}}

16: \newcommand{\eg}{{\it e.g.}}

17: \newcommand{\ie}{{\it i.e.}}

18: \newcommand{\etc}{{\it etc.}}

19: \newcommand{\rmd}{{\mathrm{d}}}

20: 

21: %aastex defines \sun for \msol to give $M_{\sun}$ Slightly larger sun sign.

22: %\newcommand{\msol}{\rm{M}$_\odot$}

23: \newcommand{\mdot}{\dot{\mathrm{M}}}

24: \newcommand{\msol}{\mathrm{M}_\odot}

25: \newcommand{\msolyr}{\msol~\textrm{yr}^{-1}}

26: \newcommand{\kms}{km~$\rm{s}^{-1}$}

27: \newcommand{\kmsbig}{$^{\textrm{km}}/_{\textrm{s}}$}

28: \newcommand{\kmssmall}{$^{\mathrm{km}}/_{\mathrm{s}}$}

29: \newcommand{\cc}{$\rm{cm}^{-3}$}

30: \newcommand{\arcs}{\char'175}

31: 

32: \newcommand{\sfrac}[2]{\,{}^{#1}\!/_{#2}}

33: \newcommand{\beq}{\begin{equation}}

34: \newcommand{\eeq}{\end{equation}}

35: \newcommand{\bdm}{\begin{displaymath}}

36: \newcommand{\edm}{\end{displaymath}}

37: 

38: \shorttitle{Formation of Turbulent Cones in Accretion Disk Outflows}

39: \shortauthors{Poludnenko, Blackman, \& Frank}

40: 

41: \slugcomment{Submitted to the Astrophysical Journal Letters}

42: 

43: \begin{document}

44: 

45: \title{Formation of Turbulent Cones in Accretion Disk Outflows and 

46: Application to Broad Line Regions of Active Galactic Nuclei}

47: 

48: \author{A.Y. Poludnenko\altaffilmark{1}, E.G. Blackman\altaffilmark{2},

49: A. Frank\altaffilmark{3}}

50: \affil{Department of Physics and Astronomy,\\

51:        University of Rochester, Rochester, NY 14627-0171}

52: \altaffiltext{1}{wma@pas.rochester.edu}

53: \altaffiltext{2}{blackman@pas.rochester.edu}

54: \altaffiltext{3}{afrank@pas.rochester.edu}

55: 

56: \begin{abstract}

57: We consider the stability of an accretion disk wind to cloud formation

58: when subject to a central radiation force.  For a vertical launch

59: velocity profile that is Keplerian or flatter and the presence of a

60: significant radiation pressure, the wind flow streamlines cross in a

61: conical layer. We argue that such regions are highly unstable, and are

62: natural sites for supersonic turbulence and, consequently, density

63: compressions. We suggest that combined with thermal instability these

64: will all conspire to produce clouds. Such clouds can exist in

65: dynamical equilibrium, constantly dissipating and reforming.  As long

66: as there is an inner truncation radius to the wind, our model emerges

67: with a biconical structure similar to that inferred by Elvis (2000)

68: for the broad line region (BLR) of active galactic nuclei (AGN). Our

69: results may also apply to other disk-wind systems.

70: \end{abstract}

71: 

72: \keywords{hydrodynamics --- instabilities --- turbulence --- 

73: galaxies:active --- (galaxies:) quasars: emission lines --- 

74: galaxies: Seyfert --- (galaxies:) cooling flows}

75: 

76: \section{INTRODUCTION}

77: %-------------------------------------------------------------------------

78: 

79: Non-spherical outflows are a ubiquitous feature of AGN. In this Letter

80: we discuss the stability of such an outflow launched normally from the

81: disk with a decaying power law velocity profile. The motivation for

82: this work is to understand the conditions under which the flow may be

83: unstable to cloud formation, and to infer the structure of the region

84: in which such clouds may reside. \citet{Elvis} has argued that data

85: require AGN BLR clouds to reside in a narrow biconical structure. Our

86: work herein supports this possibilty.

87: 

88: A variety of launching mechanisms for AGN accretion disk outflows have

89: been studied: radiatively accelerated outflows \citep{Arav94},

90: hydrodynamic line-driven winds \citep{Proga99, Proga1},

91: hydromagnetic disk winds \citep{Konigl94}, \citep{Pudritz92}, thermal

92: wind-type outflows in low luminosity AGNs \citep{Pietrini}, and

93: others. In this work we parameterize the disk wind independent of the

94: launching mechanism.

95: 

96: A disk wind outflow can become unstable from linear and nonlinear

97: perturbations in velocity, density, temperature, ionization state,

98: etc. These can lead to cloud formation. The cloud model is a leading,

99: though not unanimously accepted, paradigm for the structure of the AGN

100: BLR regions: models exist with \citep{Elvis, Urry95} and without

101: \citep{Murray1, Murray2} clouds. The main reason for the lack of

102: agreement is the uncertainty about cloud survival in AGN enviroments

103: \citep{Mathews}. They may be unstable to evaporation (e.g., see 

104: \citep{Pier95}), or be insufficiently supported by pressure and be

105: dynamically unstable to shredding \citep{Mathews, Poludnenko}. Among

106: the suggested solutions has been magnetic confinement of clouds

107: \citep{Rees87, Bottorff1}.

108: 

109: However a key point is that a typical cloud needs only to exist long

110: enough to reprocess radiation \citep{Rees87, Celotti, Kuncic1,

111: Kuncic2}. If clouds are destroyed thereafter and new clouds are

112: formed, then the system can exist in dynamical equilibrium even though

113: each cloud loses its identity swiftly.  This may apply to the broad

114: line region. Short cloud survival and regeneration times can be

115: associated with turbulence \citep{Bottorff2, Bottorff3}. The balance

116: between formation and destruction maintains a constant BLR cloud

117: number density.

118: 

119: Our analysis suggests that a nonlinear wind instability resulting from

120: the combination of launching profile + radiation pressure can lead to

121: the formation of a turbulent biconical zone that might manifest itself

122: observationally as the BLR. This biconical structure is the region

123: where flow streamlines cross. Such a bicone is remarkably similar to

124: that inferred empirically by \citet{Elvis} and from numerical

125: simulations by \citet{Proga1}. Growing observational data

126: corroborates the existance of such a structure (e.g. NGC 1068

127: \citep{Arribas96}, Mrk 3 \citep{Ruiz01}, etc.)

128: 

129: We formulate the problem in section 2.1 and describe the flow field in

130: section 2.2. We analyze the linear and nonlinear stability of the flow

131: and a possible scenario for thermal instability resulting in formation

132: of the BLR clouds in section 2.3. In section 3 we discuss the results

133: and their implications for the AGN structure and dynamics.

134: 

135: 

136: \section{DISK WIND MODEL}

137: %-------------------------------------------------------------------------

138: 

139: \subsection{Formulation of the problem}

140: 

141: All of our calculations are performed in cylindrically symmetric

142: geometry, however we do not include axial rotation. At the origin we

143: assume a black hole of mass $M_{BH}$. Fig.~\ref{streamlines} shows the

144: setup.

145: 

146: We take a Keplerian, geometrically thin, optically thick accretion

147: disk to be located along the abscissae with the distance $\lambda$

148: measured along the disk. The central engine is taken to be a

149: point-like source at the origin coincident with the black hole. We

150: exclude disk surface illumination by the central source including all

151: disk properties into the disk-wind launch velocity profile.

152: 

153: Our wind originates at the disk surface with launch velocity profile

154: \beq

155: v_{z0}(\lambda) = v^*\bigg(\frac{\lambda}{\lambda^*}\bigg)^n,

156: \label{vz0}

157: \eeq

158: where $v^*$ is the wind velocity at the innermost launch point of the

159: disk with coordinate $\lambda^*$ and $n$ is an arbitrary non-positive

160: number\footnote{We discard positive values of $n$ as unphysical.}. We

161: assume that the wind propagates into a low density medium.

162: 

163: After the wind material leaves the disk, it is subjected to gravity

164: from the black hole and radiation from the central engine. 

165: \citet{Proga99, Proga1} discuss one possible approach to line driven winds 

166: in the context of numerical modeling. To capture the most important

167: qualitative features of the flow and to simplify the analysis we

168: assume a continuum driven wind and use only Thomson opacities. This

169: assumption somewhat underestimates the driving force since it excludes

170: the contribution of line-driving. However, on the other hand, it is an

171: overestimate since typically a significant fraction of the central

172: source spectrum lies in the Compton scattering regime. To compensate

173: for that we introduce a factor $f_T < 1$, intended to decrease the

174: total luminosity $L$. The choice of a specific value of $f_T$ depends

175: on a particular situation being considered.

176: 

177: In our model, a fluid parcel of density $\rho$ leaves the disk with

178: the velocity of equation (\ref{vz0}), determined by the coordinate of

179: its launch point, and finds itself in a centrally symmetric,

180: conservative potential field. For Thomson scattering the radiative

181: force per unit volume is

182: \beq

183: \vec{F}^{rad}(r)=\frac{1}{c}\frac{\vec{r}}{|\vec{r}|}\displaystyle\int

184: \alpha_{\nu}F_{\nu}d\nu = \frac{n\sigma_{T,e}}{c}F\frac{\vec{r}}{|\vec{r}|},

185: \eeq

186: and the effective potential field is

187: \beq

188: U(r) = U_{rad} + U_{grav} = \frac{\rho GM_{BH}}{r}\Bigg(

189: \frac{f_{T}L}{L_{Edd}}-1\Bigg)=\rho\frac{\alpha}{r},

190: \label{u}

191: \eeq

192: where $L_{Edd}= 1.25\cdot 10^{38} \ erg/s (M_{BH}/\msol)$ is the

193: Eddington luminosity. Note, that here we give an expression for the

194: potential energy per unit volume.

195: 

196: \subsection{Description of the flow field}

197: 

198: Finding a trajectory and velocity of a fluid parcel in the outflow

199: reduces to the problem of the motion of a particle in a centrally

200: symmetric repulsive potential, described by (\ref{u}).  The trajectory

201: is a hyperbola \citep{Landafshitz1}, and in polar coordinates is

202: \beq

203: r(\phi)=\frac{p}{e\cos \phi -1}.

204: \label{traj}

205: \eeq

206: Here $\phi \in [0,\pi/2]$. Eccentricity of the orbit $e$ and parameter

207: of the orbit $p$ are defined as

208: \beq

209: e=1+\frac{p}{r_0}, \ \ p=2K\bigg(\frac{r_0}{r^*}\bigg)^{2n+1}r_0,

210: \label{ep}

211: \eeq

212: where $r_0=\lambda_0$ is the coordinate of the fluid element launch

213: point. The dimensionless quantity

214: \beq

215: K=\lambda^*(v^*)^2/2\alpha,

216: \label{K}

217: \eeq

218: is the ratio of the initial kinetic energy of a fluid particle

219: launched at $\lambda^*$ to its potential energy and measures the

220: relative importance of the disk wind vs. central radiation source.

221: ($K$ and $n$ are the key parameters in our analysis). Equation

222: (\ref{traj}) can be considered as an implicit solution for $\lambda_0$

223: as a function of $z$ and $\lambda$. For any point $(\lambda,z)$ we can

224: uniquely determine the launch point of the streamline,

225: $\lambda_0$. Fig.~\ref{streamlines} shows a sample streamline pattern.

226: 

227: The velocity distribution along a streamline is

228: \beq

229: \left\{

230: \begin{array}{lll}

231: v_{\lambda}(\lambda_0,\lambda) & = & \displaystyle

232: \sqrt{\frac{\alpha}{\lambda_0}(e+1)}\frac{\sinh\xi}{e\cosh\xi+1}, \\

233: v_{z}(\lambda_0,\lambda) & = & \displaystyle

234: \sqrt{\frac{\alpha}{\lambda_0}(e^2-1)(e+1)}\frac{\cosh\xi}{e\cosh\xi+1}.

235: \end{array} \right.

236: \label{vstr}

237: \eeq

238: Parameter $\xi \in [0,+\infty)$ is defined by  

239: $\cosh \xi = (\lambda/\lambda_0) (e+1)-e$.

240: 

241: \subsection{Stabilty analysis}

242: 

243: We base our linear stability analysis for the inviscid case on the

244: \emph{Fj\o rtoft theorem}, which is derived from the inviscid 

245: limit of the Orr-Sommerfeld equation, i.e. Rayleigh equation

246: \citep{Yih, Maslowe}. We have studied the outflows with the values of

247: $-2.5 \le n \le 0.0$ and $0.05 \le K \le 2.5$. In all the regimes

248: studied, the disk wind is linearly stable to infinitesimal velocity

249: perturbations.

250: 

251: However a study of equation (\ref{traj}) shows that a non-linear

252: instability can be present in a subregion of the flow. Depending on

253: $n$, a streamline inclination angle at any given point $(r,\lambda)$

254: is either a monotonically increasing or monotonically decreasing

255: function of $\lambda_0$. The condition for the change in the character

256: of monotonicity can be obtained: using equation (\ref{traj}) we can

257: determine the inclination angle of a given streamline $\tan

258: \Theta(\lambda_0,\lambda)=dz/d\lambda.$ Using 

259: \beq

260: \frac{d}{d\lambda_0}\Big(\lim_{\lambda \rightarrow \infty}

261: \tan \Theta(\lambda_0,\lambda)\Big)=0,

262: \eeq

263: we find that the monotonicity changes for $n=-1/2$, i.e. the Keplerian

264: disk is the borderline case. This means that for $n>-1/2$, each

265: successive streamline is less inclined towards the abscissae axis than

266: the preceding one and eventually intersects all of the preceding

267: streamlines. For $n \leq -1/2$ the inclination angle is a

268: monotonically non-increasing function of $\lambda_0$ and then the

269: streamlines never intersect.

270: 

271: The intersection of two streamlines can be approximated as an

272: intersection of two jets of vanishing thickness whose interaction

273: leads to a resultant turbulent jet with an inclination angle lying

274: between that of the two original jets (e.g., see

275: \citep{Landafshitz2}). Therefore, streamline intersection causes the

276: biconical zone of compressible turbulence in the overall disk-wind

277: outflow. The extent of such a zone can be estimated: since the

278: streamline inclination angle is a monotonically increasing function of

279: $\lambda_0$, the streamline originating at the innermost radius of the

280: disk is the most inclined one. Therefore, it delineates the lower

281: boundary of the turbulent biconical zone. Its inclination angle is

282: \beq

283: \tan\Theta_{min} = \lim_{\lambda \rightarrow \infty}\frac{dz}{d\lambda}

284: (\lambda^*,\lambda)=2\sqrt{K+K^2}.

285: \label{thetamin}

286: \eeq

287: An exact shape of the upper boundary of the bicone can be found by

288: solving $dz/d\lambda_0 = 0$ for $\lambda_0$ as a function of $\lambda$

289: and then substituting it into the equation (\ref{traj}) to find the

290: exact dependence $z(\lambda)$ for the upper boundary. Instead, we give

291: an approximate expression for the inclination angle of the outer

292: boundary, found by analogy to equation (\ref{thetamin}):

293: \beq

294: \begin{array}{lll}

295: \displaystyle \tan\Theta_{max} & = & \lim_{\lambda \rightarrow \infty}

296: \frac{dz}{d\lambda}(\lambda_0,\lambda) \\

297: & = & 2\sqrt{K\Big(\frac{\lambda_0}{\lambda^*}\Big)^{2n+1}

298: +K^2\Big(\frac{\lambda_0}{\lambda^*}\Big)^{2(2n+1)}}.

299: \label{thetamax}

300: \end{array}

301: \eeq

302: This defines the basic geometry of the bicone: the \emph{inclination

303: angle} $\Theta = (\Theta_{max}+\Theta_{min})/2$ and the

304: \emph{divergence angle} $\Delta \Theta = \Theta_{max}-\Theta_{min}$.

305: See Figure~\ref{streamlines}.

306: 

307: Typically, the bicone has a divergence angle $\Delta\Theta$ that

308: slowly increases with radius. Therefore equation (\ref{thetamax}),

309: which gives the limiting inclination angle of the streamline

310: originating from the point $\lambda_0$, depends on the choice of such

311: a particular streamline. However, the increase in the divergence angle

312: is typically small. Moreover, a particular streamline used to

313: determine $\Theta_{max}$ can be determined independently from the

314: properties of the accretion disk or the estimated extent of the

315: outflow.

316: 

317: Three other effects will enter in a more detailed hydrodynamic

318: treatment that influence the divergence angle of the bicone. An effect

319: which tends to make the divergence angle smaller than that determined

320: using equations (\ref{thetamin}) and (\ref{thetamax}) arises because

321: the resultant of two intersecting jets lies at an intermediate angle

322: between the original two. Each successive interaction of resultant

323: jets would then tend to narrow the overall conical structure. In

324: addition, the change in the optical depth of the outflow after the

325: formation of the turbulent bicone would shade the outer parts of the

326: outflow. This will make the bicone narrower by increasing the

327: inclination angle of the outer streamlines due to the reduced

328: radiation pressure. In competition with both of these above effects

329: is the turbulence itself, which could broaden the bicone.

330: 

331: The initial density fluctuations in the turbulent biconical zone

332: cannot themselves represent the BELR clouds; the density and

333: temperature contrasts are too small compared to the ambient warm

334: highly ionized gas (WHIM in the nomenclature of \citet{Elvis}).

335: However, thermal instability in such a medium can transform initial

336: turbulent overdensities into the BELR clouds (see \citep{Burkert,

337: Hennebelle99} for the discussion of two-phase medium formation via

338: linear thermal instability in optically thin regions). The formation

339: of the compressible turbulent bicone from intersecting flows can

340: produce inhomogeneities with density contrast up to 4 in the adiabatic

341: case and higher if one allows for the possibility of radiative

342: cooling. Such high density contrasts (compared to the linear density

343: fluctuations discussed by \citet{Burkert}) can serve as a seed for

344: thermal instability. The instabilty proceeds in the nonlinear regime

345: from the onset. Rapid cooling would then produce clumps. The maximum

346: clump density reached at a given scale is determined by how much

347: cooling can occur during an eddy turnover time on that scale, i.e.

348: during an average clump survival time.

349: 

350: This mechanism can lead to a dynamical equilibrium described by the

351: balance between in situ formation, destruction, and re-formation of

352: clumps comoving with the ambient flow. As long as any given clump

353: survives long enough to reprocess radiation, problems of clump

354: confinement against thermal conduction and destruction via high

355: velocity differences between the clumps and the ambient medium are

356: eliminated. The dynamical equilibrium of the turbulent flow, ensures a

357: constant number density of BELR clouds in the turbulent biconical zone

358: over observation durations.

359: 

360: Finally, there is a site where a linear instability of the

361: Kelvin-Helmholtz type might operate, namely the shear layer between

362: the disk wind outflow and the infalling (or stationary) circumpolar

363: material, that fills the cone of the outflow. Simulations of

364: \citet{Proga1} show the formation of clumpy structure in 

365: that boundary shear layer. They attribute it to the Kelvin-Helmholtz

366: instability. However, in regimes for which the outflow develops the

367: turbulent bicone, such linear instability will be completely

368: overwhelmed by the nonlinear instability induced by the intersection

369: of streamlines. In cases when the bicone is not formed, the outflow

370: density decreases from the maximum values near the plane of the disk

371: to nearly the values of the circumpolar infalling material in the

372: upper parts of the outflow with the density transition occurring over

373: the large range of angles. BLR clouds are not expected in this case.

374: \footnote{Note, that even in the cases when the turbulent bicone is 

375: absent the upper boundary of the outflow is still roughly defined by

376: the streamline launched at $\lambda^*$.}.

377: 

378: 

379: \section{DISCUSSION}

380: %-------------------------------------------------------------------------

381: 

382: We considered the linear and nonlinear stability of a disk outflow and

383: found that it is linearly stable to infinitesimal velocity

384: perturbations. However, when $n>-1/2$ and a supereddington radiation

385: source is present, nonlinear instability leads to the formation of a

386: biconical zone of compressible turbulence. High density contrasts in

387: this zone may trigger thermal instability and lead to the further

388: condensation of clumps into BELR clouds. Turbulence plays a key role

389: in establishing the dynamical equilibrium which maintains a steady

390: cloud number density, even though each cloud is short lived.

391: 

392: In the regimes when the outflow can develop the turbulent bicone, the

393: geometry is remarkably similar to that described by \citet{Elvis}.

394: For a disk outflow with $K \approx 0.05$ and an initial launch

395: velocity profile slightly flatter than Keplerian, namely $n=-0.45$,

396: the turbulent bicone inclination angle is $\Theta \approx 28^o$ with

397: the divergence angle $\Delta\Theta \approx 7^o$. This is in

398: quantitative and qualitative agreement with \citet{Elvis}: (1) no

399: absorbers are along the line of sight passing above the outflow; (2)

400: along the line of sight passing through the turbulent bicone broad

401: absorption and emission line features will be observed; (3) narrow

402: absorption lines (NAL) will be seen along sight lines that fall inside

403: of the bicone. Figure~\ref{streamlines} shows the geomtry of the

404: outflow and the emerging bicone with the angle values found here.

405: 

406: The outflow geometry in our picture depends only on dimensionless

407: quantities $K$ and $n$. However, in addition to $\Theta$ and

408: $\Delta\Theta$ determined above, we can also infer the inner scale by

409: matching observed cloud velocities. We assume the same value of $n$

410: and $K$ as in the previous paragraph. We take $M_{BH} =10^9\ \msol$

411: and broad absorption line velocity $v_{BAL}=10^4$ \kms (the velocity

412: observed when the line of sight passes directly along the bicone

413: towards the central source). In our model $v_{BAL}$ is the terminal

414: velocity at infinity of material launched with the largest initial

415: velocity and potential, i.e. at the point $\lambda_0=\lambda^*$.

416: Setting $\lambda_0=\lambda^*$ in (\ref{vstr}) and using the

417: expressions for $v_{\lambda}(\lambda^*,\lambda)$ and

418: $v_{z}(\lambda^*,\lambda)$ in the limit $\lambda \rightarrow \infty$,

419: we find 

420: \beq

421: \lambda^* =\frac{2\alpha(1+K)}{v_{BAL}^2}.

422: \label{lstar}

423: \eeq

424: Assuming $L=1.5L_{Edd}$ and $f_{T}=0.7$ the corresponding value of

425: $\lambda^* =\lambda_0 \approx 1000 \ a.u.$, which matches

426: \citet{Elvis}. The corresponding maximum launch velocity is $\approx

427: 2000$ \kms. Therefore, the observed velocity in the NAL region is

428: $\sim 1000$ \kms.

429: 

430: In principle, the paradigm presented here applies not only to luminous

431: AGN but to any source with a centrally symmetric potential, a

432: sufficiently luminous central radiation source, and a disk wind,

433: e.g. disk winds in young stellar objects.

434: 

435: We have not discussed the role of magnetic fields, disk rotation,

436: metallicity, wind density fall off, or the physics of line driving.

437: These should be considered in future work.

438: 

439: \acknowledgements

440: 

441: The authors thank V. Pariev for extremely important comments and

442: suggestions. AYP thanks N. Murray and M. Elvis for valuable

443: discussions. AYP and AF acknowledge support from NSF grant AST-9702484

444: and NASA grant NAG5-8428. EGB acknowledges support from DOE grant

445: DE-FG02-00ER5460. The authors acknowledge DOE support from the

446: Laboratory for Laser Energetics.

447: 

448: \begin{thebibliography}{}

449: %

450: \bibitem[Arav \ea (1994)]{Arav94}

451: Arav, N., Li, Z.-Y., Begelman, M. C. 1994, \apj, 432, 62

452: %

453: \bibitem[Arribas \ea (1996)]{Arribas96}

454: Arribas, S., Mediavilla, E., Garcia-Lorenzo, B. 1996, \apj, 463, 509

455: %

456: \bibitem[Bottorff \& Ferland (2000)]{Bottorff1}

457: Bottorff, M. C., Ferland, G. J. 2000, \mnras, 316, 103

458: %

459: \bibitem[Bottorff \& Ferland (2001)]{Bottorff2}

460: Bottorff, M. C., Ferland, G. 2001, \apj, 549, 118

461: %

462: \bibitem[Bottorff \ea (2000)]{Bottorff3}

463: Bottorff, M., Ferland, G., Baldwin, J., Korista, K. 2000, \apj, 542, 644

464: %

465: \bibitem[Burkert \& Lin (2000)]{Burkert}

466: Burkert, A., Lin., D. N. C. 2000, \apj, 537, 270

467: %

468: \bibitem[Celotti \ea (1992)]{Celotti} 

469: Celotti, A., Fabian, A. C., Rees, M. J. 1992, \mnras, 255, 419

470: %

471: \bibitem[Elvis (2000)]{Elvis}

472: Elvis, M. 2000, \apj, 545, 63

473: %

474: \bibitem[Hennebelle \& P\'erault (1999)]{Hennebelle99}

475: Hennebelle, P., P\'erault, M. 1999, \aap, 351, 309

476: %

477: \bibitem[K\"onigl \& Kartje (1994)]{Konigl94}

478: K\"onigl, A., Kartje, J. F. 1994, \apj, 434, 446

479: %

480: \bibitem[Kuncic \ea (1996)]{Kuncic1}

481: Kuncic, Z., Blackman, E. G., Rees, M. J. 1996, \mnras, 283, 1322

482: %

483: \bibitem[Kuncic \ea (1997)]{Kuncic2}

484: Kuncic, Z., Celotti, A., Rees, M. J. 1997, \mnras, 284, 717

485: %

486: \bibitem[Landau \& Lifshitz (1959)]{Landafshitz2}

487: Landau, L. D., Lifshitz, E. M. 1959, Fluid Mechanics

488: (Reading : Addison-Wesley)

489: %

490: \bibitem[Landau \& Lifshitz (1976)]{Landafshitz1}

491: Landau, L. D., Lifshitz, E. M. 1976, Mechanics (Oxford, New York : Pergamon Press)

492: %

493: \bibitem[Maslowe (1985)]{Maslowe}

494: Maslowe, S. A. 1985, in Hydrodynamic Instabilities and Transition to Turbulence,

495: ed. H.L. Swinney, J.P. Gollub (2nd ed.;Springer-Verlag), 181

496: %

497: \bibitem[Mathews \& Ferland (1987)]{Mathews} 

498: Mathews, W. G., Ferland, G. J. 1987, \apj, 323, 456

499: %

500: \bibitem[Murray \& Chiang (1995)]{Murray1}

501: Murray, N., Chiang, J. 1995, \apjl, 454, L105

502: %

503: \bibitem[Murray \ea (1995)]{Murray2}

504: Murray, N., Chiang, J., Grossman, S. A., Voit, G. M. 1995, \apj, 451, 498

505: %

506: \bibitem[Pelletier \& Pudritz (1992)]{Pudritz92}

507: Pelletier, G., Pudritz, R. E. 1992, \apj, 394, 117

508: %

509: \bibitem[Pier \& Voit (1995)]{Pier95}

510: Pier, E. A., Voit, G. M. 1995, \apj, 450, 628

511: %

512: \bibitem[Pietrini \& Torricelli-Ciamponi (2000)]{Pietrini}

513: Pietrini, P., Torricelli-Ciamponi, G. 2000, \apj, 363, 455

514: %

515: \bibitem[Poludnenko \ea (2002)]{Poludnenko}

516: Poludnenko, A. Y., Frank, A., Blackman, E. G. 2002, \apj, in press

517: %

518: \bibitem[Proga \ea (1999)]{Proga99}

519: Proga, D., Stone, J. M., Drew, J. E. 1999, \mnras, 310, 476

520: %

521: \bibitem[Proga \ea (2000)]{Proga1}

522: Proga, D., Stone, J. M., Kallman, T. R. 2000, \apj, 543, 686

523: %

524: \bibitem[Ruiz \ea (2001)]{Ruiz01}

525: Ruiz, J. R., Crenshaw, D. M., Kraemer, S. B., Bower, G. A., Gull, T. R., 

526: Hutchings, J. B., Kaiser, M. E., Weistrop, D. preprint (astro-ph/0108521)

527: %

528: \bibitem[Rees (1987)]{Rees87}

529: Rees, M. J. 1987, \mnras, 228, 47p

530: %

531: \bibitem[Urry \& Padovani (1995)]{Urry95}

532: Urry, C. M., Padovani, P. 1995, \pasp, 107, 803

533: %

534: \bibitem[Yih (1969)]{Yih}

535: Yih, C. S. 1969, Fluid Mechanics (2d ed.; McGraw-Hill, New York 1969)

536: 

537: \end{thebibliography}

538: 

539: %--------- Figure 1  -----------------------------------------------------------

540: \clearpage

541: 

542: \begin{figure}

543: \plotone{f1.eps}

544: \caption{Sample streamline pattern for a disk wind outflow. Shown is the case 

545: $n=-0.45$ and $K=0.06$. Shaded region is the turbulent bicone formed

546: by the intersection of streamlines. Values of the angles shown are

547: discussed in the text.

548: \label{streamlines}}

549: \end{figure}

550: 

551: \end{document}