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

2: \documentclass[numberedappendix]{emulateapj}

3: 

4: \begin{document}

5: 

6: \title{Energetic Impact of Jet Inflated Cocoons in Relaxed Galaxy Clusters}

7: \author{John C. Vernaleo and Christopher S. Reynolds}

8: \affil{Department of Astronomy, University of Maryland, College Park,

9:   MD 20742-2421}

10: \email{vernaleo@astro.umd.edu, chris@astro.umd.edu}

11: 

12: \begin{abstract}

13: Jets from active galactic nuclei (AGN) in the cores of galaxy clusters

14: have the potential to be a major contributor to the energy budget of

15: the intracluster medium (ICM).  To study the dependence of the

16: interaction between the AGN jets and the ICM on the parameters of the

17: jets themselves, we present a parameter survey of two-dimensional

18: (axisymmetric) ideal hydrodynamic models of back-to-back jets injected

19: into a cluster atmosphere (with varying Mach numbers and kinetic

20: luminosities).  We follow the passive evolution of the resulting

21: structures for several times longer than the active lifetime of the

22: jet.  The simulations fall into roughly two classes, cocoon-bounded

23: and non-cocoon bounded sources.  We suggest a correspondence between

24: these two classes and the Faranoff-Riley types.  We find that the

25: cocoon-bounded sources inject significantly more entropy into the core

26: regions of the ICM atmosphere, even though the efficiency with which

27: energy is thermalized is independent of the morphological class.  In

28: all cases, a large fraction (50--80\%) of the energy injected by the

29: jet ends up as gravitational potential energy due to the expansion of

30: the atmosphere.

31: \end{abstract}

32: 

33: \keywords{cooling flows -- galaxies: active -- galaxies: jets

34:   --- hydrodynamics -- X-rays: galaxies: clusters} 

35: 

36: \section{Introduction}

37: \label{sec:intro}

38: 

39: Jets from powerful Active Galactic Nuclei (AGN) play a major role in

40: shaping their environment on both the large and small scales.  In many

41: clusters of galaxies, the intracluster medium (ICM) bears the marks of

42: repeated episodes of AGN jet activity: e.g., bubbles

43: \citep{2000MNRAS.318L..65F, 2000ApJ...534L.135M, 2001ApJ...558L..15B,

44: 2002ApJ...579..560Y}, ghost bubbles \citep{2001ApJ...562L.149M,

45: 2002ApJ...569L..79H, 2004ApJ...606..185C, 2000MNRAS.318L..65F},

46: ripples \citep{2003MNRAS.344L..43F,2005MNRAS.360L..20F}, shells

47: \citep{2000MNRAS.318L..65F}, and filaments.  Besides affecting the

48: gross morphology of the ICM, AGN have the potential to dramatically

49: alter the energy and entropy budgets of the ICM.  This is especially

50: of interest in relation to the cooling flow problem of galaxy

51: clusters~\citep{1977ApJ...215..723C, 1977MNRAS.180..479F,

52: 1994ARA&A..32..277F}.  In this paper, we attempt to determine the

53: relationship between the jet parameters (Mach number and kinetic

54: luminosity) and the resulting energetics and structures in the ICM.

55: 

56: X-ray observations show that the ICM radiates strongly in the X-ray

57: wavelengths due to thermal bremsstrahlung radiation.  In the central

58: regions of rich, relaxed clusters, the ICM has a cooling time shorter

59: than the Hubble time (sometimes as low as a few $\times 10^8$ years).

60: With such short cooling times, we would expect to see large quantities

61: of cool gas and/or star formation in the center of rich clusters.

62: Equivalently, the cD galaxy at the center of the cluster would be

63: accreting cooled matter from the cooling ICM.  The density-squared

64: dependence of the emission ensures that, if this simple picture were

65: correct, the cooling would increase and eventually reach a cooling

66: catastrophe in a finite time.   

67: 

68: Observations strongly suggest that large quantities of cooled gas are

69: not being deposited in the central galaxy.  The observed star

70: formation rates in the central cD galaxies are not sufficient to match

71: the inferred cooling

72: rates~\citep{2004ApJ...612..131O,2005ApJ...635L...9H}.  Furthermore,

73: the mass function of galaxies shows a high-mass truncation that argues

74: against the continued ICM accretion growth of cD galaxies and requires

75: a feedback process significantly more efficient than star

76: formation (see~\citet{2003ApJ...599...38B}).  More directly,

77: high-resolution X-ray spectroscopy with the Reflection Grating

78: Spectrometer (RGS) on {\it XMM-Newton} shows that the ICM cools from

79: the virial temperature to approximately one third of the virial

80: temperature but reveals an absence of plasma below this temperature.

81: The apparent contradiction between the absence of cool gas and the

82: short cooling times of the ICM core is the cooling flow

83: problem~\citep{1994ARA&A..32..277F}.

84: 

85: These observational results suggest that some form of ICM heating is

86: required to offset the radiative cooling.  Intermittent activity by a

87: central AGN remains an attractive solution.  A large amount of

88: theoretical work has been performed on the possible effects of AGN on

89: cooling cluster cores.  With ever increasing computer power, most of

90: the recent work has focused on hydrodynamic models of the AGN/ICM

91: interaction

92: \citep{1997MNRAS.284..981C,2001ApJ...554..261C,2001MNRAS.325..676B,2002Natur.418..301B,2002MNRAS.332..271R,2003MNRAS.339..353B,2004MNRAS.348.1105O,2004MNRAS.350L..13O,2004MNRAS.355..995D,2004ApJ...601..621R,2005A&A...429..399Z}.

93: Some groups have also begun initial explorations of the effects of

94: magnetic fields \citep{2004ApJ...601..621R,2005ApJ...624..586J},

95: plasma processes \citep{2004ApJ...611..158R,2005MNRAS.357..242R},

96: feedback prescriptions~\citep{2006ApJ...645...83V}, and realistic

97: (cosmological) background motions~\citep{2006MNRAS.373L..65H}.

98: 

99: Collectively, this body of simulation work has allowed us to explain

100: many of the observed features of AGN/ICM interactions, investigate how

101: AGN induced flows mix metals within the ICM core, and study entropy

102: injection and heating of the ICM.  However, most of the current sets

103: of simulation are performed for a limited set of jet parameters and

104: hence it is unclear how to generalize the results to the population of

105: as a whole.  In this paper, we present a moderately large set of

106: (axisymmetric) high-resolution hydrodynamic simulations of jet/ICM

107: interactions that survey a wide range of jet powers and jet

108: velocities.  We study the morphology of the radio-galaxy as well as

109: the injection of energy and entropy into the ICM as a function of the

110: jet properties.  In Section~\ref{sec:setup}, we will discuss the

111: details of our setup and the code used.  In

112: Section~\ref{sec:analysis}, we will present the analysis of our

113: simulations and their classifications.  In Section~\ref{sec:disc} we

114: will discuss the results, followed by conclusions in

115: Section~\ref{sec:conc}.

116: 

117: \section{Simulation Setup}

118: \label{sec:setup}

119: 

120: Our goal is to model a relaxed cluster and its interaction with a

121: central radio galaxy that produces back-to-back jets.  To do this, we

122: start out with an initial cluster that is both spherically symmetric

123: and isothermal, with the (adiabatic) sound speed $c_s=1$ everywhere.

124: The initial cluster gas is assumed to follow a $\beta$-model density

125: profile,

126: \begin{equation}

127:   \label{eq:beta}

128:   \rho(r)=\frac{1}{[1+(\frac{r}{r_0})^2]^{3/4}},

129: \end{equation}

130: where the core radius is $r_0=2.0$ in code units.

131: 

132: A static gravitational potential,

133: \begin{equation}

134:   \label{eq:grav}

135:   \Phi=\frac{c^2_s}{\gamma}\ln(\rho),

136: \end{equation}

137: is set to make the gas initially in hydrostatic equilibrium and is

138: assumed to remain fixed in time.  This is equivalent to a potential

139: set entirely by stationary dark matter that dominates the system, so

140: the self gravity of the gas is ignored.  The intracluster gas, while

141: containing a large fraction of the baryons in the cluster, is indeed

142: not a significant contributor to the overall cluster mass.

143: 

144: We use spherical polar coordinates ($r$, $\theta$, $\phi$) and impose

145: symmetry in the $\phi$ direction ($\partial/\partial\phi=0$).  The

146: radial coordinate varied between $0.1$ and $30.0$ in scaled code

147: units.  The comparison between code units and physical units is given

148: in Section~\ref{sec:real}.  The values of $\theta$ were allowed to

149: vary between $0$ and $\pi$ (i.e., we are modeling the full sphere).

150: Since we have the full range of angles, this allows us to have

151: back-to-back jets.  Thus we can model interaction between the

152: backflows from the jets rather than the commonly employed technique of

153: imposing reflecting boundary conditions on the $\theta=\pi/2$ plane

154: (e.g., see ~\citet{2007ApJ...656L...5S,2006ApJ...645...83V}).  We

155: included second order differences in grid spacing in the $\theta$

156: direction for each hemisphere,

157: giving us two realizations of the jet/cocoon structure per

158: simulations.  In no cases were there major differences between the two

159: sides, confirming that the details of the griding do not affect our

160: results.  A small circle around $r=0$ was excluded from the

161: computational domain to avoid the coordinate singularity.  

162: 

163: All simulations were run with a $n_r\times n_\theta=1200\times600$

164: grid.  In the radial direction, a ratioed grid was used, with

165: successive cells increasing by a factor of $1.003$.  The $\theta$

166: direction also used a ratioed grid increasing by a factor of $0.998$

167: from $0$ to $0.5\pi$ and then decreasing by a factor of $1.002$ from

168: $0.5\pi$ to $\pi$.  This provides the greatest number of zones near

169: the center and near the two jet axes while providing the second-order

170: differences in the $\theta$ direction mentioned above.

171: 

172: Clearly jets are three dimensional structures.  However, there is a

173: long history of two-dimensional modeling of jets which supports the

174: usefulness of two dimensional jets~\citep{2002ApJS..141..371C,

175: 2001ApJ...554..261C, 2001JKAS...34..329M, 2002MNRAS.332..271R}.  There

176: are however several aspects of the dynamics that cannot be captured in

177: two dimensions.  There is no way to have realistic random motions in

178: the background.  This lack of random motions produces lobes which are

179: far more regular and symmetric than those found in any real

180: source. The two dimensional assumption also enhances mixing between

181: the high entropy material and ambient material. This is because

182: non-axisymmetric Kelvin-Helmholtz (KH) and Rayleigh-Taylor (RT) modes 

183: are clearly not possible in an axisymmetric system. Without

184: non-axisymmetric modes, all of the mixing must be done by the

185: axisymmetric modes, which are more efficient at mixing

186: \citep{2002MNRAS.332..271R}. Fortunately, this will be most

187: problematic only at late times when there has been enough time

188: for significant mixing to occur.

189: 

190: The hydrodynamic evolution of the jet/ICM system was followed using

191: the ZEUS-3D code~\citep{1992ApJS...80..753S, 1992ApJS...80..791S}.

192: ZEUS is a fixed grid Eulerian code, explicit in time and second order

193: accurate (when using van Leer advection as we do).  Artificial

194: viscosity is used to handle shocks.  All simulations were run as ideal

195: hydrodynamic cases, neglecting magnetic fields and self gravity.  ZEUS

196: solves the standard equations of hydrodynamics,

197: \begin{equation}

198:   \label{eq:hd1}

199:   \frac{D \rho}{D t}+\rho\nabla\cdot v = 0,

200: \end{equation}

201: 

202: \begin{equation}

203:   \label{eq:hd2}

204:   \rho\frac{D v}{D t}=-\nabla P -\rho\nabla\Phi,

205: \end{equation}

206: 

207: \begin{equation}

208:   \label{eq:hd3}

209:   \rho\frac{D }{D t} \left( \frac{e}{\rho} \right) =-P\nabla\cdot v - \Lambda,

210: \end{equation}

211: where

212: \begin{equation}

213:   \label{eq:hd4}

214:   \frac{D }{D t}\equiv\frac{\partial}{\partial t} + v\cdot\nabla.

215: \end{equation}

216: Radiative cooling, represented by the $\Lambda$-term in

217: Equation~\ref{eq:hd3}, is

218: neglected in order to concentrate on the hydrodynamic evolution and

219: the thermodynamic response of the ICM to the jets.

220: 

221: \subsection{Modeling the jets}

222: \label{sec:jet}

223: Two, back-to-back jets were injected into the center of the model

224: cluster.  This was achieved with an inflow boundary condition on the

225: inner radial boundary.  All the jets were highly supersonic with

226: respect to the ambient sound speed and moderately to highly supersonic

227: with respect to their internal sound speeds.  Direct injection allows

228: us to keep the details of jet acceleration and central engines outside

229: of the grid, so we are able to solve the equations of hydrodynamics

230: consistently {\it within our computational grid}.  

231: 

232: The injected jets are initially conical with half-opening angles of

233: $15^\circ$, and are injected in pressure balance with the ICM core.

234: The drop of internal pressure associated with subsequent jet expansion

235: leads to external, pressure driven recollimation and internal shocks,

236: ultimately resulting in some new, self-consistently determined opening

237: angle.  Given these (fixed) assumptions, the properties of a given jet

238: are parameterized by the kinetic luminosity $L_{\rm kin}$ and internal

239: Mach number of the injected jet matter.  These are related to the

240: injection density $\rho$, pressure $p$ and velocity $v$ by

241: \begin{eqnarray}

242:   \mathcal{M}_{jet}=\frac{v_j}{c_s}&=v_j\sqrt{\frac{\rho}{\gamma(\gamma-1)e}},\\

243:   L_{kin}&=\frac{1}{2}\rho_j v_j^3 \pi r_{in}^2.

244:   \label{eq:lkin}

245: \end{eqnarray}

246: where $r_{\rm in}$ is the radius of the ``injection nozzle'' for the

247: jet.  The jet parameters for the base run ($L_{\rm kin}=26.47$ and

248: $\mathcal{M}_{\rm jet}=10.55$; $\rho_j=0.01$ and $v_j=105.5$) are the same

249: as the run presented in \citet{2002MNRAS.332..271R}.

250: 

251: At total of 26 models were run, where only the kinetic luminosity

252: and/or the Mach number of the jets were varied.  In two of these

253: simulations, the jet lifetimes were also varied.  The full list of

254: simulations is given in Table~\ref{t:sims}.  Internal Mach numbers in

255: the range 3.7--21.1, and kinetic luminosity in the range 6.6---68.8

256: were explored.  The choice of parameters used was based on raising or

257: lowering the Mach number and/or kinetic luminosity by a factor of two

258: from one of the previous runs.

259: 

260: %earpage

261: 

262: \begin{table}

263: 

264: \caption{Parameters defining the 26 simulations presented in this

265: paper.  Also given are the (visual) morphological classification of

266: the resulting structure, as judged at time $t=2.0$.  See text for more

267: details.}

268: \begin{center}

269: \begin{tabular}{c c c c c } \hline \hline

270: Run & Jet Active Time & $L_{kin}$ & $\mathcal{M}_{jet}$ & Morphology \\ \hline

271: 1 & 1.0 & 26.47 & 10.55 & Cocoon\\

272: 2 & 1.0 & 52.94 & 21.1 & Cocoon\\

273: 3 & 1.0 & 26.43 & 15.89 & Cocoon\\

274: 4 & 1.0 & 26.31 & 7.39 & Cocoon\\

275: 5 & 1.0 & 68.97 & 10.46 & Cocoon\\

276: 6 & 1.0 & 23.32 & 10.54 & Cocoon\\

277: 7 & 1.0 & 52.94 & 10.55 & Cocoon\\

278: 8 & 0.5 & 26.47 & 10.55 & Cocoon\\

279: 9 & 1.5 & 26.47 & 10.55 & Cocoon\\

280: 10 & 1.0 & 8.91 & 6.47 & Cocoon\\

281: 11 & 1.0 & 17.91 & 9.48 & Cocoon\\

282: 12 & 1.0 & 19.98 & 15.0 & Cocoon\\

283: 13 & 1.0 & 31.09 & 20.02 & Cocoon\\

284: 14 & 1.0 & 13.24 & 21.1 & Non-Cocoon\\

285: 15 & 1.0 & 13.23 & 10.55 & Non-Cocoon\\

286: 16 & 1.0 & 26.47 & 21.1 & Non-Cocoon\\

287: 17 & 1.0 & 6.62 & 10.55 & Non-Cocoon\\

288: 18 & 1.0 & 9.96 & 8.99 & Non-Cocoon\\

289: 19 & 1.0 & 21.01 & 13.0 & Non-Cocoon\\

290: 20 & 1.0 & 30.1 & 18.02 & Non-Cocoon\\

291: 21 & 1.0 & 24.98 & 19.0 & Non-Cocoon\\

292: 22 & 1.0 & 52.99 & 5.28 & Unresolved\\

293: 23 & 1.0 & 26.44 & 3.71 & Unresolved\\

294: 24 & 1.0 & 26.43 & 5.27 & Unresolved\\

295: 25 & 1.0 & 10.05 & 5.01 & Unresolved\\

296: 26 & 1.0 & 12.15 & 6.54 & Unresolved\\

297: \hline

298:   \end{tabular}

299: \end{center}

300: \label{t:sims}

301: \end{table}

302: %earpage

303: 

304: 

305: \subsection{Comparison to Real Units}

306: \label{sec:real}

307: Throughout this paper, quantities are quoted in dimensionless code

308: units unless explicitly stated otherwise.  Since we are solving the

309: equations of ideal hydrodynamics (Equations~\ref{eq:hd1}-\ref{eq:hd4})

310: with no additional physics added, there is no unique set of physical

311: scales to our problem.  Each simulation may therefore be compared to

312: several different sets of physical scales.  Following Reynolds et

313: al. (2002), we shall quote two scalings for our simulations (although

314: there is actually an entire three-parameter family of scalings).  For

315: our rich cluster scaling, we set the core radius $r_0=100\,{\rm kpc}$,

316: meaning one code unit in r is equal to $50\,{\rm kpc}$.  Such clusters

317: are hot, with sound speed, $c_s=1000\,{\rm km}\,{\rm s}^{-1}$. This,

318: along with the typical central number density $n_0=0.01\,{\rm

319: cm}^{-1}$, gives a time unit of $50$\,Myrs and a kinetic luminosity

320: unit of $L_{kin}=3.5\times10^{44}{\rm erg}\,{\rm s}^{-1}$.  We can

321: also consider a poor cluster scaling, $r_0=50\,{\rm kpc}$,

322: $c_s=500\,{\rm km}\,{\rm s}^{-1}$, and $n_0=0.1\,{\rm cm}^{-1}$.

323: These give a length unit of $25$\,kpc, a time unit of $10$ Myrs and a

324: kinetic luminosity unit of $L_{kin}=4.4\times10^{42}\,{\rm erg}\,{\rm

325: s}^{-1}$.

326: 

327: \section{Analysis and Results}

328: \label{sec:analysis}

329: 

330: The primary diagnostic for distinguishing injected jet material from

331: ambient and shocked-ambient material is the specific entropy index,

332: \begin{equation}

333:   \label{eq:ent}

334:   \sigma=\frac{P}{\rho^{\gamma}}.

335: \end{equation}

336: The ambient material (both undisturbed and shock heated) has a low

337: specific entropy index (which ranges from $0.6$ to $9$) and the

338: injected jet material has a specific entropy index on the order of

339: $1000$ for our base simulation (run 1).  The specific entropy of

340: injected material does vary with jet power, going from near $1$ for

341: the weakest jets to $10^7$ for the most powerful.  With only two ways

342: to change entropy (increase with shocks and increase or decrease

343: through numerical mixing), specific entropy generally provides a

344: powerful way to distinguish injected (and shocked) material from the

345: background gas.  A cutoff of $\sigma=10$ was set to separate jet

346: material from ambient material.  This is close to the highest ambient

347: value so we are not likely to artificially miss any material that

348: should count as part of the cocoon/high entropy material.  We use the

349: same entropy-index cutoff for all simulations to allow for direct

350: comparisons to be made.  This value was chosen to avoid counting

351: ambient material which has the same range of values in all

352: simulations.  In the case of the weakest jets, this entropy index

353: cutoff cannot pick out the pre-shocked jet material.  Once the jet

354: material passes through the shock terminal shock, however, it is

355: always clearly distinguished from ambient material by a $\sigma=10$

356: cutoff.  For a further discussion of the entropy cutoff diagnostic,

357: see~\citet{2002MNRAS.332..271R}.

358: 

359: \subsection{Morphological Classification}

360: \label{sec:morph}

361: 

362: Based on the entropy maps, the simulations were visually placed into

363: three classes.  Simulations with a distinct, well-defined monolithic

364: cocoon of jet-originating plasma were labeled as ``Cocoon bounded

365: sources'' (or simply ``cocoon sources'').  We consider a cocoon to be

366: a jet inflated region of high entropy material with a coherent

367: structure that is well separated from the low entropy background and

368: relatively safe from mixing (until late times).  Simulations which

369: showed faint, wispy plumes of jet-originating material instead of well

370: defined cocoons were labeled ``Non-Cocoon bounded sources''

371: (``non-cocoon sources'').  The final group were those simulations that

372: do not appear sufficiently resolved for an accurate classification.

373: In the discussion below, we shall abbreviate the term

374: ``jet-originating material'' to ``jet material''.

375: 

376: An example of a typical cocoon source can be seen in

377: Figure~\ref{fig:coc}.  Figure~\ref{fig:noncoc} shows a typical

378: non-cocoon source.  A poorly resolved simulation can be seen in

379: Figure~\ref{fig:unres}.  We also observe that it is possible to

380: separate the cocoon-bounded from the non-cocoon bounded sources by use

381: of density contours.  Figures~\ref{fig:coc-cont}

382: and~\ref{fig:ncoc-cont} show density contours for a cocoon bounded and

383: non-cocoon bounded source respectively.  Simulations in the cocoon

384: category appear to have the high entropy jet material in a distinct

385: region separated from the background by a solid density contour and

386: remain that way until fairly late times.  Non-cocoon sources do not

387: have such distinctions and the jet material seems mixed in with the

388: background even early on.

389: %earpage

390: 

391: \begin{figure*}

392:   \centering

393:   \epsscale{0.4}

394:   %This is run28

395:   \plotone{f1.eps}

396:   \caption{Entropy map of central regions of typical cocoon bounded

397:     source (run 1) at t=2.0.}

398:   \label{fig:coc}

399: \end{figure*}

400: 

401: \begin{figure*}

402:   \centering

403:   \epsscale{0.4}

404:   %This is run49

405:   \plotone{f2.eps}

406:   \caption{Entropy map of central regions of typical non-cocoon

407:     bounded source (run 21) t=2.0.} 

408:   \label{fig:noncoc}

409: \end{figure*}

410: 

411: \begin{figure*}

412:   \centering

413:   \epsscale{0.4}

414:   %This is run19

415:   \plotone{f3.eps}

416:   \caption{Entropy map of central regions of typical unresolved source

417:     (run 23) at t=2.0.}

418:   \label{fig:unres}

419: \end{figure*}

420: 

421: \begin{figure*}

422:   \centering

423:   \epsscale{0.6}

424:   %This is run28

425:   \plotone{f4.eps}

426:   \caption{Density contours for inner half of cocoon bounded source

427:   (run 1) at t=2.0}

428:   \label{fig:coc-cont}

429: \end{figure*}

430: 

431: \begin{figure*}

432:   \centering

433:   \epsscale{0.6}

434:   %This is run49

435:   \plotone{f5.eps}

436:   \caption{Density contours for inner half of non-cocoon bounded

437:   source (run 21) at t=2.0}

438:   \label{fig:ncoc-cont}

439: \end{figure*}

440: %earpage

441: 

442: 

443: These classifications were performed at the same time for all

444: simulations, $t=2.0$.  At this time, the structures have evolved

445: passively for the same amount of time that they were driven by the

446: jet.  However, our classification does not appear to depend

447: sensitively on the choice of time, although there are some

448: limitations.  It the very earliest stages of the active jet phase

449: ($t<0.2-0.4$), the jets are still pushing out from the inner boundary

450: of the simulation (i.e., the morphology is strongly determined by the

451: artificial aspects of the simulation) and classification is not

452: meaningful.  At late times ($t>10-15$), mixing has disrupted all of our

453: simulated sources and, again, eliminated any morphological

454: distinction.

455: 

456: Using this classification, we can study how morphology depends on

457: $L_{\rm jet}$ and $\mathcal{M}_{\rm jet}$.  Our results are shown in

458: Fig.~\ref{fig:pspace}.  We see three separate regions on this

459: parameter space.  At the high kinetic luminosities and lower Mach

460: numbers, the runs are unresolved.  From this point on, the unresolved

461: sources were excluded from all analysis.  Low luminosity sources seem

462: to fall in the non-cocoon category, almost regardless of Mach number

463: (except that at the high Mach number they can have a higher kinetic

464: luminosity).  The middle region contains the cocoon bounded sources.

465: In Fig.~\ref{fig:pspace2}, we show the parameter space again, on a

466: density-velocity plot.  Although the overlap between the morphologies

467: is more apparent in this parameter space, we can see that there

468: appears to be a continuum of morphologies with the fastest and

469: lightest jet producing unresolved sources at one end and the slow

470: heavy jets producing non-cocoon sources.  Again, the cocoon sources

471: occupy the middle ground between the other two.  A discussion of the

472: physical origin of the cocoon will be reserved for the Discussion

473: (Section~\ref{sec:disc}).

474: 

475: As revealed by Fig.~\ref{fig:pspace}, the line separating cocoon

476: bounded and non-cocoon bounded sources on the $(L_j,{\cal M}_{\rm

477: jet})$ is approximately linear (i.e., $L_j\propto {\cal M}_{\rm

478: jet}$).  This can be understood from the analytic models of cocoon

479: expansion from \citet{1989ApJ...345L..21B}.  Equation (4) of

480: \citet{1989ApJ...345L..21B} states the the pressure of a cocoon is

481: given by

482: \begin{equation}

483: p_c\sim \frac{(L_{\rm kin}v_j\rho_aA_h)^{1/2}}{A_c}

484: \end{equation}

485: where $A_h$ is the area of the working surface over which the jet

486: deposits its momentum, $A_c$ is the cross-sectional area of the

487: cocoon, and $\rho_a$ is the density of the ambient ISM/ICM.  Well

488: defined cocoons can only be sustained if $p_c$ exceeds the pressure of

489: the ambient medium, $p_a$.  Noting that the axial ratio of the cocoons

490: are approximately constant (i.e., $A_h/A_c\sim {\rm constant}$), and

491: that the density and pressure of our model ambient ICM atmosphere

492: outside of the ICM core in the same in all simulations and drops off

493: as $r^{-3/2}$, the condition that $p_c>p_a$ can be written as the

494: condition 

495: \begin{equation}

496: (L_{\rm kin}v_j)^{1/2}>\Upsilon_{\rm crit},

497: \label{eqn:cocoon_pressure}

498: \end{equation}

499: where $\Upsilon_{\rm crit}$ is a weak function of $r$ ($\Upsilon

500: \propto r^{1/4}$) and hence time.  Since the jets are injected with

501: fixed initial pressure, it can be shown that initial jet velocity

502: depends on the kinetic luminosity and initial internal Mach number

503: according to $ v_j\propto L_{\rm kin}/{\cal M}_{\rm jet}^2$.  Thus, we

504: can use eqn.~\ref{eqn:cocoon_pressure} to see that the line

505: separating cocoon from non-cocoon sources (i.e., where the cocoon just

506: comes into pressure balance with the ambient ICM) can be written as

507: $L_{\rm jet}\propto \Upsilon {\cal M}_{\rm jet}$.  

508: 

509: %earpage

510: 

511: \begin{figure*}

512:   \centering

513:   \epsscale{0.6}

514:   \plotone{f6.eps}

515:   \caption{Morphology vs. jet parameters.  Lines separating the

516:   different regions were done by eye.  Although the

517:   lines between the different regions are not perfect, there appear to

518:   be regions in which different classes are clearly excluded.}

519:   \label{fig:pspace}

520: \end{figure*}

521: %earpage

522: 

523: \begin{figure*}

524:   \centering

525:   \epsscale{0.6}

526:   \plotone{f7.eps}

527:   \caption{Morphology vs. jet parameters.  On the density-velocity

528:     plane, we see the previously distinct regions as more of a

529:     continuum moving from one type of source to the next.}

530:   \label{fig:pspace2}

531: \end{figure*}

532: %earpage

533: 

534: \subsection{Mass Distribution and Energetics}

535: \label{sec:energetics}

536: 

537: To examine the mass distribution and energetics during the

538: simulations, several quantities were calculated at each output time.

539: The total mass of jet material at each time is given by:

540: \begin{equation}

541:   M_{cocoon}=\int_C \rho d V

542:   \label{eq:mcoc}

543: \end{equation}

544: where $C$ is the region with $\sigma>10$.  The internal, kinetic, and

545: potential energies were calculated for both the jet and the ambient

546: material,

547: \begin{eqnarray}

548:   E_{int}^{C,A}&=\frac{1}{\gamma-1}\int_{C,A} P d V,\\

549:   E_{kin}^{C,A}&=\frac{1}{2}\int_{C,A} \rho v^2 d V,\\

550:   E_{pot}^{C,A}&=-\int_{C,A}\rho\Phi d V.

551: \end{eqnarray}

552: where $A$ is the region with $\sigma\le 10$.  In all cases, the

553: integral is taken over only the region of interest (i.e., cocoon or

554: ambient only, as determined by the entropy cutoff).  We then subtract

555: the initial energies in order to derive the change of energy $\Delta

556: E_{int}^{C,A}$, $\Delta E_{kin}^{C,A}$, and $\Delta E_{pot}^{C,A}$.

557: 

558: The results of these calculations for a representative cocoon bounded

559: simulation are shown in Figure~\ref{fig:cmasses}.

560: Figure~\ref{fig:ncmasses} shows similar plots for a non-cocoon bounded

561: simulation.  These figures are similar to Figure 4

562: of~\cite{2002MNRAS.332..271R} and primarily differ due to our

563: increased outer radius (and the plots are continued until later

564: times).  

565: The upper-left panel in each figure shows the total mass

566: of the cocoon material compared to the amount of injected material.

567: In both cases, this increases until the jet turns off and then begins

568: to decrease.  For the cocoon-bounded source, this decrease is mostly

569: steady for the entire lifetime of the simulation.  For the non-cocoon

570: source, the drop is much greater early on, so that about halfway through

571: the simulation most of the high entropy material has mixed with the

572: background and thus the decay rate has lessened.  This enhanced mixing

573: is a direct consequence of the lack of a well defined contact

574: discontinuity in the non-cocoon sources.  Note that the cocoon mass

575: does not correspond to the cocoon mass plot

576: in~\cite{2002MNRAS.332..271R} as we are interested in the total cocoon

577: mass, not the change in cocoon mass.

578: 

579: Also shown in Figs.~\ref{fig:cmasses} and \ref{fig:ncmasses}

580: (upper-right panel) is the length of the region containing jet

581: material.  In the cocoon bounded sources, the length of the cocoon

582: evolves in a smooth manner, gradually decelerating as the cocoon comes

583: into pressure balance with the ambient ICM.  The discontinuity at late

584: times occurs well into the phase in which the cocoon has transformed

585: into two buoyantly rising plumes and corresponds to the complete

586: mixing/dispersal of the leading part of the plume.  The evolution of

587: the length of the non-cocoon bounded sources, on the other hand,

588: suffered a sharp break at the time that the jet is shut off.

589: 

590: The two lower panels of Figs.~\ref{fig:cmasses} and \ref{fig:ncmasses}

591: show the change in energies for the jet material and the ICM material

592: respectively.  For the jet material, there is little difference in the

593: time-dependence of the energies for the two classes with one

594: exception.  In the non-cocoon case, the changes in internal and

595: kinetic energy peak at a comparable value although the kinetic drops

596: to almost zero very soon after.  For the cocoon case, the change in

597: kinetic energy never reaches a comparable level to the change in

598: internal energy.  Physically, this is close to the heart of the

599: difference between the cocoon and non-cocoon sources.  In each case, a

600: set amount of energy is injected by the jet (along with a set amount

601: of mass as seen in the potential change which is not very dramatic).

602: The evolution of the system then determines how this energy is split

603: between the internal energy and the kinetic energy (and eventually how

604: much is transfered to the ICM).  In the case of the cocoon, the hot

605: jet and shocked material is kept separate from the background, and

606: goes almost entirely into the internal energy of the shocked jet gas.  For

607: the non-cocoon case, the energy is split nearly evenly

608: between the internal and the kinetic energy.  This means there is no

609: longer enough internal energy available to inflate a cocoon of shocked

610: gas; more of the energy goes to the kinetic energy of the mostly bare

611: jet (with wispy areas of hot, shocked gas around it).  This is

612: consistent with the analytic estimate of~\citet{1992ApJ...392..458C}.

613: 

614: The time-dependence of the ICM energies show a greater difference

615: between the two cases.  The potential energy suffers a greater change

616: (within a few times the jet lifetime) in the cocoon case due to the

617: well-defined shell of ICM that is lifted up by the expanding cocoon,

618: while the non-cocoon case has a much more gradual change in the

619: potential energy.  The internal energy in the cocoon case drops back

620: down to the initial value around the same time the potential peaks

621: while the non-cocoon case has what looks like a slow trade-off between

622: the two.  Finally, the kinetic energy changes are comparable in the

623: two cases, as there appears to be a similar amount of disturbance in

624: the ICM regardless of the nature of the inflated structure.

625: 

626: In its standard configuration, ZEUS is not a strictly energy

627: conserving code.  We do not however believe that this is a major

628: impact on our results.  Fig.~\ref{fig:conserve} shows that during the

629: majority of time during the simulation, total energy is conserved to a

630: sufficient degree.  At very early times, the jet injects hot material

631: onto the grid which is clearly not conservative.  At very late times,

632: material (and sound waves carrying energy) leave the outer boundary of

633: the grid.  During  the time in between, the energy change remains flat

634: with only minor variations (at the $10^{-5}$ level), showing that

635: energy is mostly conserved.

636: %earpage

637: 

638: \begin{figure*}

639:   \centering

640:   %This is run28

641:   \epsscale{0.35}

642:   \plotone{f8a.eps}

643:   \plotone{f8b.eps}

644: 

645:   \plotone{f8c.eps}

646:   \plotone{f8d.eps}

647:   \caption{Mass and Energy vs. time for cocoon bounded simulation

648:     (run 1).

649:     Starting from the top left and moving clockwise, the panels are:

650:     Mass of the cocoon material divided by the injected

651:     mass, the maximum radial extent of the high entropy material, the

652:     change in internal, kinetic, and potential energies for the

653:     ambient, low entropy 

654:     material compared to the initial state, and the same energy

655:     changes for the cocoon material.}  

656:   \label{fig:cmasses}

657: \end{figure*}

658: 

659: \begin{figure*}

660:   \centering

661:   %This is run49

662:   \epsscale{0.35}

663:   \plotone{f9a.eps}

664:   \plotone{f9b.eps}

665: 

666:   \plotone{f9c.eps}

667:   \plotone{f9d.eps}

668:   \caption{Mass and Energy vs. time for non-cocoon bounded

669:     simulation (run 21).

670:     Starting from the top left and moving clockwise, the panels are:

671:     Mass of the cocoon material divided by the injected

672:     mass, the maximum radial extent of the high entropy material, the

673:     change in internal, kinetic, and potential energies for the

674:     ambient, low entropy 

675:     material compared to the initial state, and the same energy

676:     changes for the cocoon material.}

677:   \label{fig:ncmasses}

678: \end{figure*}

679: %earpage

680: 

681: \begin{figure*}

682:   \centering

683:   \epsscale{0.6}

684:   \plotone{f10.eps}

685:   \caption{Fractional change in total energy within the simulated

686:   domain as a function of time.  All forms of energy (kinetic,

687:   internal and gravitational) for all of the gas in the computational

688:   domain are considered.  Shown here is the change in total energy

689:   compared with the initial time $\Delta E$ divided by the initial

690:   total energy $E_0$.}

691:   \label{fig:conserve}

692: \end{figure*}

693: %earpage

694: 

695: \subsection{Entropy evolution and thermalization efficiencies}

696: \label{sec:en}

697: 

698: We now examine the effect of the jets on the entropy of the background

699: gas.  First, we define the specific entropy,

700: \begin{equation}

701:   \Delta S=S_1-S_0=\log{\sigma_1}-\log{\sigma_0},

702: \end{equation}

703: where $\sigma$ is the specific entropy index from

704: Equation~\ref{eq:ent}.

705: 

706: Figure~\ref{fig:scomp} shows the entropy difference (compared to the

707: initial condition) at the final output of the simulations.  At this

708: point, the systems were allowed to evolve passively for nineteen times

709: the active lifetime of the jet.  For each simulation, the total

710: entropy at each radius was calculated and then averaged over the

711: angular coordinate.  The split between the different morphologies is

712: evident in this figure.  Within the core, the cocoon sources have a

713: higher $\Delta S$ overall.  Further discussion of this issue will be

714: reserved for Section~\ref{sec:disc}.

715: %earpage

716: 

717: \begin{figure*}

718:   \centering

719:   \epsscale{0.6}

720:   \plotone{f11.eps}

721:   \caption{Angle averaged change in entropy.  See discussion in text

722:     for full description.}

723:   \label{fig:scomp}

724: \end{figure*}

725: %earpage

726: 

727: As a final look at the effect of the jets on the cluster energetics,

728: we show the efficiencies for conversion of the injected energy into

729: various forms within the ICM as measured at the final time for

730: both the cocoon and non-cocoon sources.  This is defined as the change

731: in each type of energy for the ICM material divided by the total

732: amount of energy injected by the jet at the final output time of the

733: simulation.  Figures~\ref{fig:change-int}, \ref{fig:change-kin} and

734: \ref{fig:change-pot} show the efficiency for conversion of injected

735: energy into ICM internal energy, kinetic energy and potential energy,

736: respectively.  There is no clear segregation between the two types

737: although the cocoon sources do appear to cluster together somewhat.

738: In all cases, a large fraction of the injected energy (50--80\%) ends

739: as gravitational potential energy of the ICM as it expands in response

740: to the AGN heating.

741: %earpage

742: 

743: \begin{figure*}

744:   \centering

745:   \epsscale{0.6}

746:   \plotone{f12.eps}

747:   \caption{Efficiency for change in Internal Energy vs.\,\,Kinetic

748:   Luminosity for ambient medium.}

749:   \label{fig:change-int}

750: \end{figure*}

751: 

752: \begin{figure*}

753:   \centering

754:   \epsscale{0.6}

755:   \plotone{f13.eps}

756:   \caption{Efficiency for change in Kinetic Energy vs.\,\,Kinetic

757:   Luminosity for ambient medium.}

758:   \label{fig:change-kin}

759: \end{figure*}

760: 

761: \begin{figure*}

762:   \centering

763:   \epsscale{0.6}

764:   \plotone{f14.eps}

765:   \caption{Efficiency for change in Potential Energy vs.\,\,Kinetic

766:   Luminosity for ambient medium.}

767:   \label{fig:change-pot}

768: \end{figure*}

769: %earpage

770: 

771: \section{Discussion}

772: \label{sec:disc}

773: 

774: In the (resolved) simulations, we see a split in the morphology of the

775: jet-created structures.  As we saw in the parameter space plot

776: (Figure~\ref{fig:pspace}), the presence or absence of a cocoon is not

777: related to the jet speed or power in a simple matter.  So the forms

778: that we see do not tell us directly about the parameters of the jet

779: that formed them.  What they do tell us is something about the way the

780: jet interacted with the ICM.

781: 

782: The cocoon itself is formed by shocks.  The supersonic jet initially

783: flows freely through the ICM.  After traveling some distance (usually

784: short on the scale of these simulations), the jet will build up enough

785: material in front of it that it can no longer flow freely.  The

786: material at the head of the jet is shock heated and expands.  Up to

787: this point, the evolution is the same for both cocoon and non-cocoon

788: sources.  As this shocked material expands, the evolution diverges.

789: 

790: The hot, shocked material expands into the surrounding medium.  At the

791: boundaries of the shocked material, fluid instabilities (i.e., the

792: Rayleigh-Taylor (RT) and Kelvin-Helmholtz (KH) instabilities) work to

793: shred the contact discontinuity and destroy the forming bubble.  In

794: the cocoon bounded sources, the bubble is inflated before it is

795: shredded and continues to inflate until it reaches pressure

796: equilibrium with the background.  While active, the jet deposits

797: energy only at point where it comes into contact with the shocked ICM

798: and freshly shocked material flows back from there, continuing to

799: inflate the bubble.  Once the jet ceases, the bubble floats buoyantly

800: from the cluster center.

801: 

802: In the case of non-cocoon sources, the jet is unable to inflate a

803: bubble.  If the jet cannot deposit enough energy at the contact point

804: to inflate the bubble, the fluid instabilities will mix the shocked

805: material with the background material, preventing the bubble from

806: ever growing.  This leaves areas of hot, shocked gas that can rise,

807: but never produces a single coherent structure.

808: 

809: The clearest distinction in the thermodynamics of cocoon and

810: non-cocoon sources can be seen in the spatial distribution of the

811: entropy enhancement.  We find that in the core regions, cocoon sources

812: cause a greater enhancement of the ICM entropy than non-cocoon sources

813: although there is some overlap.  Outside the core (around $r=5$), both

814: appear the same and cause a slight negative change in the entropy.

815: This negative change corresponds to lower entropy material that has

816: been pushed upwards by the expansion of the core.  For the non-cocoon

817: sources, this change persists to the edge of the simulation.  Some,

818: but not all, of the cocoon sources have a spiky, positive change when

819: approaching the edge of the grid --- this corresponds to the remnants

820: of the buoyant cocoon that, in some cases, remains intact until the

821: end of the simulation.  It is important to note that the comparison of

822: overall entropy change was done at the final output ($t=20$).  This is

823: long after our classifications were done.  At this point, mixing has

824: disturbed the structure enough that it would no longer be possible to

825: separate the simulations into cocoon or non-cocoon categories.  The

826: fact that we still see a difference at this late a time shows that the

827: jets have long lasting effects on the energetics of the ICM and that

828: these effects vary based on the jet properties in a way that is more

829: complicated than simply depending on jet power.

830: 

831: The energy efficiencies (Figures~\ref{fig:change-int},

832: \ref{fig:change-kin}, and \ref{fig:change-pot}) are initially

833: puzzling.  The non-cocoon simulations generally seem to coincide with

834: the highest efficiency cocoon simulations for internal and potential

835: energy efficiency.  For the kinetic energy

836: (Figure~\ref{fig:change-kin}), the non-cocoon runs seem to have both

837: close to the highest values and the lowest values.  This is in

838: contrast with the situation for entropy where the cocoon simulations

839: have a greater impact.  It is fairly clear why the cocoon sources have

840: a great entropy enhancement overall: shocks.  The cocoon is formed by

841: shocked gas, so jets that are capable of forming cocoons are also

842: generally the ones that shock more ICM (although this is not the

843: entire story as the rate of shocked gas vs. mixed gas matters as

844: well).  It is not as immediately obvious why jets that are good at

845: increasing the entropy are not always as good at increasing the energy

846: and why there can be a spread in efficiencies for jets with similar

847: parameters.  To see why, we must go back to Equation~\ref{eq:ent}, the

848: definition of the specific entropy index (keeping in mind that the

849: pressure, $P$, is directly proportional to the internal energy.  For a

850: given change in internal energy, the change in entropy can be higher

851: (or lower) by having it take place at a lower (or higher) density.  So

852: if the shocks happen higher in the potential well (and hence at a

853: lower density), it is possible to have a greater change in entropy for

854: the same amount of energy input.  The spread in kinetic energies shows

855: that it is possible to stir up the gas significantly with both the

856: cocoon and the non-cocoon producing jets.  Also, as stated in the

857: discussion, the cocoon producing jets leave more energy available as

858: internal energy which can go towards shocks while the non-cocoon

859: producing jets have more of their energy in kinetic energy which does

860: not go to shocks and therefore is not available to increase the

861: entropy.  It is also the case that different jets will cause a

862: different amount of material and energy to be lost from the system at

863: the outer edge of the grid.  We consider it to be reasonable to count

864: this as lost energy as anything that has not managed to heat the gas

865: by the time it reaches the outer edge is clearly not going to be able

866: to change the temperature of the system.

867: 

868: This disconnect between entropy increase and energy increase also fits

869: in with the results from~\cite{2006ApJ...645...83V}, which showed that

870: even with large amounts of energy available, a jet can produce large,

871: well formed cocoons yet fail to significantly heat the ICM.  The large

872: cocoons indicate a large change in entropy, but the the central

873: regions were still not heated significantly.  This helps to reaffirm

874: the point that in pure hydrodynamics, it is hard to energetically

875: couple powerful jets to the {\it core} regions of the ICM (where it is

876: needed to solve the cooling flow problem!).

877: 

878: Along with the energetic issues, we may also compare these simulations

879: to observations in a broad sense.  Classically, radio galaxies are

880: split into two categories: Faranoff-Riley Type I (FR I) and

881: Faranoff-Riley Type II (FR II)~\citep{1974MNRAS.167P..31F}.  Lacking

882: any information on luminosities and radio emission, we cannot directly

883: compare our simulations to these classes of real radio galaxies, but

884: we can note some interesting correlations.  FR II galaxies encompass

885: the large, back-to-back classical double sided sources.  Our

886: cocoon-bounded sources are reminiscent of the classical doubles.  The

887: less lobe dominated and often bent FR I sources share a similar

888: structure with our non-cocoon sources.  Although our classifications

889: were done based on the passive phase of the system, we can also

890: compare the jets from the active phase to the FR I/FR II divide.  

891: As seen in Fig.~\ref{fig:pspace} and~\ref{fig:pspace2}, slow, heavy,

892: and lower luminosity jets (the non-cocoon cases) should correspond to

893: the FR I sources while fast, light, and more luminous jets should

894: correspond to the FR II type.  As FR II galaxies have higher

895: luminosity jets~\citep{1996AJ....112....9L}, this is at least

896: consistent with our simulations.

897: Even without a definite

898: correspondence between the two, we do show that it is possible to

899: produce qualitatively different structures with differences in the jet

900: parameters without any environmental changes.

901: 

902: Two additional simulations (run 8 and run 9) were performed where all

903: the parameters were kept the same as our standard case (run 1), but

904: the active time of the jet was varied.  In run 8, the jet had an

905: active time of only $t_j=0.5$ while run 9 had an active time of

906: $t_j=1.5$.  In both cases, these simulations fall in the same

907: cocoon-bounded category as run 1.  This give us some confidence that

908: our results are not simply based on the amount of

909: total energy inject by the jet, but instead depend on the detailed

910: hydrodynamic interactions between the jets and the ambient material.

911: 

912: \section{Conclusion}

913: \label{sec:conc}

914: 

915: We have performed a large number of axisymmetric simulations of AGN

916: jets in cluster atmosphere.  By limiting ourselves to pure

917: hydrodynamics with no radiative cooling, we are able to carefully

918: study the interaction of the atmosphere with the jets and the

919: formation of cocoons and bubbles in the cluster.  With the continued

920: interest in the probable role of jets and bubbles in the cooling flow

921: problem, it is important that we understand their dynamics as clearly

922: as possible.

923: 

924: Our main results are summarized in Figures~\ref{fig:pspace} and

925: \ref{fig:scomp}.  The morphology vs.\,\,jet parameters plot shows that we

926: are able to produce two distinct classes of structures, cocoon bound

927: and non-cocoon bound, by varying the initial jet parameters.  This

928: distinction does not change with a change in total injected energy (as

929: seen when we vary the jet lifetime) and is a function of both jet

930: parameters (Mach number and kinetic luminosity).  We can also visually

931: draw comparisons between our two classes and the FR I and FR II

932: distinction amongst real radio galaxies.  The angle averaged entropy

933: change vs. radius shows that along with the split in morphology, the

934: effect on the energetics of the ICM also depends on the jet

935: parameters.  We also see that even short lived single burst jets are

936: capable of a long-lived enhancement of the entropy of a cluster core.

937: 

938: \acknowledgements

939: We would like to thank the developers of ZEUS-3D and NCSA for

940: providing the code we used.  We would like to thank the anonymous

941: referee for useful comments that allowed us to enhance the discussion

942: in several ways.  All simulation in this paper were

943: performed on the Beowulf cluster (``the Borg'') in the Department of

944: Astronomy, University of Maryland.  We thank support from Chandra

945: Theory and Modeling Program under grants TM4-5007X and TM7-8009X.

946: 

947: \bibliographystyle{apj}

948: \bibliography{paper2d}

949: 

950: \end{document}

951: