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

2: 

3: 

4: \shorttitle{SNe in CS Bubbles}

5: \shortauthors{Dwarkadas.}

6: \begin{document} 

7: \input{macros}

8:  

9: \title{The Evolution of Supernovae in Circumstellar Wind Bubbles II:

10:   Case of a Wolf-Rayet star}

11:  

12:  

13: \author{ Vikram V. Dwarkadas}

14: 

15: \affil{Astronomy and Astrophysics, Univ of Chicago, 5640 S Ellis Ave

16: AAC 010c, Chicago IL 60637}

17:  

18: \email{vikram@oddjob.uchicago.edu }

19:  

20: 

21: \begin{abstract}

22: Mass-loss from massive stars leads to the formation of circumstellar

23: wind-blown bubbles surrounding the star, bordered by a dense

24: shell. When the star ends its life in a supernova (SN) explosion, the

25: resulting shock wave will interact with this modified medium. In a

26: previous paper \citep{d05} we discussed the basic parameters of this

27: interaction with idealized models. In this paper we go a step further

28: and study the evolution of SNe in the wind blown bubble formed by a 35

29: $\msun$ star that starts off as an O star, goes through a red

30: supergiant phase, and ends its life as a Wolf-Rayet star. We model the

31: evolution of the circumstellar medium throughout its lifetime, and

32: then the expansion of the SN shock wave within this medium. Our

33: simulations clearly reveal fluctuations in density and pressure within

34: the surrounding medium, due to the changing mass-loss parameters over

35: the star's evolution. The SN shock interacting with these

36: fluctuations, and then with the dense shell surrounding the wind-blown

37: cavity, gives rise to a variety of transmitted and reflected shocks in

38: the wind bubble. The interactions between these various shocks and

39: discontinuities is examined, and its effects on the emission from the

40: remnant, especially in the X-ray regime, is noted. In this particular

41: case the shock wave is trapped in the dense shell for a large number

42: of doubling times, and the remnant size is restricted by the size of

43: the surrounding circumstellar bubble. Our multi-dimensional

44: simulations reveal the presence of several hydrodynamic

45: instabilities. They show that the turbulent interior, coupled with the

46: large fluctuations in density and pressure, gives rise to an extremely

47: corrugated SN shock wave. The shock shows considerable wrinkles as it

48: impacts the dense shell, and the impact occurs in a piecemeal fashion,

49: with some parts of the shock wave interacting with the shell before

50: the others. As each interaction is accompanied by an increase in the

51: X-ray and optical emission, different parts of the shell will

52: `light-up' at different times. The reflected shock that is formed upon

53: shell impact will comprise of several smaller shocks with different

54: velocities, and which are not necessarily moving radially inwards. The

55: non-spherical nature of the interaction means that it will occur over

56: a prolonged period of time, and the spherical symmetry of the initial

57: shock wave is completely destroyed by the end of the simulation.

58: \end{abstract}

59: 

60: 

61: \keywords{hydrodynamics --- instabilities --- shock waves ---

62: supernovae: general --- stars: winds, outflows --- supernova remnants}

63: 

64: \section{INTRODUCTION}

65: 

66: Core-collapse supernovae (SNe), those generally classified as Type

67: Ib/c and Type II, arise from massive stars (M $>$ 8 $\msun$). These

68: stars lose a considerable amount of mass prior to their explosion as

69: supernovae.  Mass loss from the star will modify the medium

70: surrounding it, giving rise to wind-blown cavities surrounded by

71: expanding shells of gas.  When the supernova explodes, the resulting

72: blastwave will eventually interact with this modified circumstellar

73: medium (CSM) rather than the interstellar medium (ISM) into which the

74: star was born.

75: 

76: In the past decades several pieces of evidence have suggested that

77: many supernovae arise within circumstellar bubbles. The most famous

78: example is the exceptionally well-studied SN 1987A, which is thought

79: to have exploded within a bipolar circumstellar bubble \citep{sck05a,

80: sck05b}. This bubble is due to the interaction of the wind from the

81: blue-supergiant progenitor with the mass loss from a prior red

82: supergiant (RSG) stage \citep{lm91, bl93}. Other well know SNe that

83: have been interpreted as arising within a wind-blown bubble include

84: Cas A \citep{bsbs96}, G292+0.8 \citep{pr02}, RCW86 \citep{vkb97}, and

85: the Cygnus Loop \citep{l97}.  Circumstellar interaction models have

86: been proposed for N132D \citep{h87} and Kepler's SNR \citep{b87}.

87: 

88: The formation of wind-blown bubbles around stars has been studied in

89: detail, both analytically \citep{wmc77, km92a, km92b} and numerically

90: \citep{fm94, mf95,dcb96}. Such models usually assume an ad-hoc

91: prescription for the wind properties.  Some attempts have been made to

92: take the evolution of the wind into account in the case of planetary

93: nebulae \citep{m95, db98} and Wolf-Rayet bubbles \citep{gm95, bd97}.

94: 

95: The evolution of supernovae in different media has also been studied,

96: especially in a constant density medium, and in a medium whose density

97: decreases as a power-law with radius. Self-similar solutions for the

98: interaction of supernovae with a power-law density profile have been

99: derived by \citet{c82}. Ostriker \& McKee (1988) present a compilation

100: of solutions for the evolution of astrophysical blast waves in various

101: media.

102: 

103: The density structure within a CS bubble however is quite different

104: from that in the constant density ISM. Thus the evolution of the

105: supernova remnant within the pre-existing wind-blown bubble differs

106: considerably from its counterpart evolving in the pristine ISM. The

107: evolution of supernovae in wind-blown bubbles has not received quite

108: as much attention as it deserves. \citet{cd89} did some preliminary

109: work, and \citet{cl89} studied the evolution analytically. A series of

110: papers by Tenorio-Tagle and his colleagues \citep{tbf90, trfb91,

111: rtfb93} studied the numerical evolution in further detail. A CS

112: evolution model for Cas A was computed by \citet{bsbs96}.

113: 

114: In \citet[][hereafter Paper 1]{d05} we introduced the various aspects

115: of the problem, and carried out one-dimensional calculations that

116: illustrated the various parameters involved. The results are

117: summarized in \S \ref{sec:intro}. These calculations assumed idealized

118: winds with constant properties, appropriate for studying the various

119: factors that affect SN evolution. More realistic models however

120: require a prescription of the mass loss history of the star as its

121: ascends the HR diagram, coupling the evolution of the circumstellar

122: gas to that of the star itself.

123: 

124: An important step forward in this process was the computation of

125: detailed stellar evolution models of 35 and 60 $\msun$ stars by

126: Norbert Langer \citep{lhl94}. These models, which provided the

127: required mass loss history with time, were used by \citet{glm96a,

128: glm96b} to compute the dynamical evolution of the surrounding gas. As

129: they show in their work, taking the complete stellar mass-loss history

130: into account results in a much more complicated circumstellar

131: structure.  The presence of a variety of small scale structures,

132: various dynamical instabilities and multiple shock fronts presents a

133: complex morphology that compares well with observations.  These

134: calculations however did not compute the full evolution of the

135: circumstellar bubble in multi-dimensions, nor the final supernova

136: phase of the star and the resultant interaction with the surrounding

137: medium.

138: 

139: In this paper we take a more detailed look at a specific case, that of

140: the 35 $\msun$ star. Using the 35 $\msun$ star model, courtesy of Prof

141: Norbert Langer, we first study the formation of the medium around the

142: star as it evolves along the HR diagram. We then assume that the star

143: explodes as a SN, leaving behind a compact remnant, and study the

144: evolution of the SN shock wave within this medium. Initially we

145: compute spherically symmetric one-dimensional calculations that

146: illustrate the various shock structures and the dynamics

147: involved. This is followed by two-dimensional computations that take

148: multi-dimensional factors such as deviations from symmetry, the onset

149: of turbulence, and the presence of hydrodynamic instabilities into

150: account. 

151: 

152: Our aim in this paper is to show the impact of a single massive star

153: on the surrounding medium throughout its lifetime. Although we

154: concentrate on a specific case, our goal is to illustrate general

155: properties of the interaction which are more globally applicable, and

156: to identify the various features which distinguish this from the

157: interaction of a SN shock with the ISM. Preliminary results from this

158: work were outlined by \citet{d07a}. This work expands considerably on

159: the results outlined therein, describes in detail the spherically

160: symmetric calculations that are necessary to understand the shock

161: dynamics and structures, computes X-ray luminosity and surface

162: brightness, and elaborates on the intricacies of the multi-dimensional

163: calculations. In a companion paper \citep{d07b} we also provide a more

164: general analytic discussion for the properties of wind-blown bubbles

165: around massive stars.

166: 

167: The rest of this paper proceeds as follows: In \S 2 we give a brief

168: overview of supernova explosions within wind-blown cavities. In \S 3

169: we present one-dimensional simulations that illustrate the evolution

170: of the bubble density and pressure with time, and describe the various

171: evolutionary phases. This is followed by a 1D simulation of the

172: SN-bubble interaction, which captures the essence of the

173: hydrodynamics. \S 4 follows up with 2D simulations that show the

174: formation of various dynamical instabilities and other

175: higher-dimensional effects. \S 5 summarizes the paper, provides a

176: general discussion of the results and their applications, and outlines

177: followup work.

178: 

179: \section{Overview of SNR-wind bubble interaction}

180: \label{sec:intro}

181: 

182: \citet{tbf90} and \citet{d05} showed that the interaction of the

183: supernova ejecta with a wind-blown shell can be divided into various

184: regimes, depending on the ratio of the mass of the shell to the mass

185: of the ejected material, a quantity that we label as $\Lambda$. If

186: this ratio is small ($ \Lambda << 1$) then the presence of the shell

187: merely acts as a perturbation to the flow. Indications of the

188: interaction are visible in the density, velocity and temperature

189: profiles of the ejecta.  However once the shock has swept up an amount

190: of material exceeding a few times the shell mass, the ejecta `forget'

191: about the existence of the dense shell. The ejecta density profile

192: reverts back to the profile that would have existed in the absence of

193: the shell.  The expansion parameter `$\delta$' (where R$_{sn} \propto

194: {\rm t}^{\delta}$) which had dropped considerably at the point of

195: shell interaction, increases gradually till it reaches the value it

196: would have had in the absence of the shell. In about 10-20 doubling

197: times, the SNR will completely "forget" about the shell and continue

198: to evolve as if the shell had never existed. The density profile

199: changes to reflect this. Since the emission from the remnant after a

200: few months is mainly due to circumstellar interaction, the changing

201: density profile will be reflected in a change in the emission from the

202: remnant, such as the optical and X-ray emission. Dwarkadas (2005)

203: showed the change in the X-ray surface brightness profile due to

204: changes in the density profile.

205: 

206: As the ratio $\Lambda$ increases, the energy imparted by the remnant

207: to the shell is larger, and the evolution begins to change. The

208: interaction of the remnant with the shell drives a shock front into

209: the shell. The high pressure behind the shock-shell interface sends a

210: reflected shock back through the ejecta. Thermalization of the ejecta

211: is achieved in a much shorter time as compared to thermalization by

212: the SN reverse shock.  The ejecta after reaching the center bounce

213: back, sending a weaker shock wave that will collide again with the

214: shell. In time a series of shock waves and rarefaction waves are seen

215: to be traversing the ejecta. Each time a shock wave collides with the

216: dense shell a corresponding (but successively weaker) rise in the

217: X-ray emission from the remnant is seen.

218: 

219: The presence of the dense shell results in a deceleration of the shock

220: wave, transfer of ejecta energy to the shell as well as conversion of

221: kinetic to thermal energy, which in the case of a very dense shell may

222: be effectively radiated away. These effects tend to speed up the

223: evolution of the nebula. The `Sedov stage' may be reached later than

224: for evolution in a constant density medium (due to the lower interior

225: density) and may last for a shorter time (due to the dense shell). In

226: some cases when the value of $\Lambda >> 1$ the Sedov stage may be

227: completely by-passed. In such a situation the shock wave will merge

228: with the wind driven shell. The velocity of the blast wave is

229: sufficiently decreased that the flow time becomes comparable to the

230: age of the remnant. Radiative cooling begins to dominate and the

231: remnant may enter the radiative stage much earlier on in its lifetime.

232: 

233: Herein we have outlined the basics of SN shock wave interaction with

234: dense shells.  Further details can be found in \citet{d05} and

235: references therein.

236: 

237: \section{Evolution of the CSM around a 35 Solar Mass star}

238: 

239: \subsection{1-dimensional calculations}

240: 

241: In this section we outline the evolution of the Wolf-Rayet bubble

242: around a 35 solar mass star, and describe the eventual SN-shell

243: interaction. Although this has been previous carried out by

244: \citet[hereafter GLM96]{glm96b} we chose to redo the entire

245: calculation. It is necessary for us to run the calculation in order to

246: be able to remap it onto the grid, and set up the initial conditions

247: for the SNR to to interact with this medium. In the process of

248: carrying out this simulation we found several differences between our

249: work and that of \citet{glm96b}, especially in the two-dimensional

250: case. Our use of a lower ambient density (more characteristic of the

251: ISM) allowed the bubble to expand to a much larger size. The

252: differences in our simulations are mainly due to us being able to

253: compute the evolution over all the stages in two-dimensions as a

254: result of the considerable increase in computational power. This is

255: reflected in a change in the character of the medium into which the

256: SNR evolves. Given the size and morphology variations in our

257: computations as compared to those of \citet{glm96b}, and the necessity

258: in describing accurately the medium into which the SN shock wave is

259: evolving, we have chosen to present herein a rather detailed

260: description of the evolution of the bubble.

261: 

262: The simulations were carried out using the VH-1 code, a

263: two-dimensional finite-difference hydrodynamic code based on the

264: Piecewise Parabolic Method \citep{cw84}. The code employs an expanding

265: grid, that tracks the outer shock front and expands along with

266: it. This trick is very useful in cases where the dimensions of the

267: grid expand by many orders of magnitude over the course of a run. The

268: 1D simulations presented here were carried out on a grid of 2000

269: zones. 2-dimensional simulations were carried out on a grid of 600 by

270: 600 zones. Radiative cooling was implemented in the form of a cooling

271: function, adopted from the one given in \citep{sd93}, and modified to

272: extend to lower temperatures. The effect of the lower temperature is

273: mainly to make the shells thinner.

274: 

275: In the course of its evolution the star evolves through 3 phases. The

276: wind properties in each phase are computed generally from

277: parameterizations derived from fits to the observed data

278: \citep{kppa89, l89, nj90}. The first and longest lasting is the

279: main-sequence (MS) stage which lasts for about 4.56 million years. In

280: this stage the star, which starts its life as an O star, loses mass in

281: the form of a wind with a mass-loss rate on the order 10$^{-7}\; {\rm

282: to}\; 10^{-6} \msun$~yr$^{-1}$ and a wind velocity that starts close

283: to 4000 km~s$^{-1}$ and gradually decreases with time. The wind

284: velocity of the star over its evolution is shown in Figure

285: \ref{fig:bubvel}, and the mass-loss rate is depicted in Figure

286: \ref{fig:bubmdot}. At the end of the MS phase the star swells up

287: immensely in size to become a red supergiant (RSG). In this phase mass

288: is lost in the form of a very slow, dense, wind. The velocity in this

289: simulation decreases down to 75 km~s$^{-1}$, although in reality we

290: would expect RSG winds to be even slower, on the order of 10-20

291: km~s$^{-1}$. The mass-loss rate increases to close to 10$^{-4}

292: \msun$~yr$^{-1}$. Since the RSG phase lasts for about 250,000 years,

293: the total mass lost in this stage is very large, about 19.6

294: $\msun$. At the end of the RSG phase, the star sheds its H envelope

295: and becomes a Wolf-Rayet star. Stars in this stage lose mass in the

296: form of radiatively driven winds, with a mass loss rate that is a

297: factor of a few smaller than their RSG predecessors, but a velocity

298: that is 2 orders of magnitude higher. As can be surmised, the

299: recurrent changes in the wind properties can lead to continuous

300: changes in the structure of the surrounding medium into which the

301: stellar wind is expanding.

302: 

303: In order to accurately compute the surroundings we assume that the

304: star was born in a medium with constant density 2.34 $\times$

305: 10$^{-24}$ g cm$^{-3}$, a number density of about 1 particle/cm$^3$

306: for a medium with 90\% H and 10\% He. This density is lower than that

307: assumed by GLM96, who used an artificially high density to avoid large

308: computational domains. Our use of an expanding grid partially

309: circumvents the latter problem, since we do not need to start from a

310: grid extending out to about 100 parsecs. Our initial grid extends out

311: to about 10$^{15}$ cms. The lower external density results in a bubble

312: about twice the radius obtained by GLM96.

313: 

314: It is not clear what the density of the medium around the star is over

315: its lifetime. On the one hand many observations suggest that the mean

316: surface density of a massive star forming region is around 1 g

317: cm$^{-2}$ \citep[see review in][]{m04}, which would indicate a very high

318: volume density. But OB stars have high velocities that may cause them

319: to drift away from their birthplace \citep{mc05}. The presence of GRBs

320: occurring several hundred parsecs away from massive star forming

321: regions \citep{hfs06} shows that some massive stars do end their lives

322: in low density regions, irrespective of where they were

323: born. Observations of isolated HII regions which are far removed from

324: their host galaxies \citep{rw04} also indicate that sometimes massive

325: star formation can occur in very low density regions. Taking all these

326: factors into account, we have assumed a density of 1 particle/cm$^3$

327: as an average interstellar medium density over the stellar

328: lifetime. We note that if the density is higher, many of the results

329: will scale appropriately.

330: 

331: The interaction of the MS wind with the surrounding medium leads to

332: the formation of a double shocked structure, consisting of an outer

333: shock expanding into the ISM, and a reverse shock, often referred to

334: as a ``wind termination shock''. The termination shock separates the

335: free-streaming wind from the shocked wind region, and moves inwards in

336: a Lagrangian sense, i.e.~the entire structure expands outwards, but

337: the termination shock eventually moves towards the center with respect

338: to the outer shock. Most of the volume is occupied by the shocked wind

339: region, forming a hot, tenuous region that may emit in soft X-rays.

340: The formation of wind-blown bubbles is further described in Paper 1

341: and references therein. The density and pressure structure at various

342: timesteps during the evolution of the bubble in this stage are shown

343: in figure \ref{fig:ms1d}. The title of each plot gives the evolution

344: time in years. The two numbers in the top right hand corner denote the

345: wind velocity in km/s (upper) and the mass-loss rate in solar masses

346: per year (lower). The initial wind velocity is close to 4000 $\kms$,

347: and decreases slowly with time, accompanied by a corresponding

348: increase in its mass loss rate. This happens in such a manner that the

349: mechanical luminosity (0.5 $\times \dot{M} {v_w}^2$ ) is almost

350: constant. \citet{wmc77} calculated an analytic solution that describes

351: the evolution of the outer shock with time under such circumstances.

352: The radius R of the shell increases as R $\propto$ t$^{0.6}$, which

353: confirms reasonably well with the simulations throughout most of the

354: MS stage, and the structure of the bubble is reasonably consistent

355: with our expectations based on this paper. As mentioned the main

356: sequence stage in this model lasts for about 4.5 $\times$ 10$^6$

357: years. The total amount of mass lost in this stage is on the order of

358: 2.5 $\msun$.

359: 

360: Towards the end of the main-sequence stage there is an abrupt drop in

361: wind velocity as the star swells up considerably in size, and enters

362: the red supergiant stage.  The large drop in velocity, and

363: corresponding rise in mass loss rate leads to a much higher density

364: for the RSG wind, and a change in the wind ram pressure. A new system

365: of pressure equilibrium is established in the bubble interior, and the

366: position of the termination shock adjusts accordingly (figure

367: \ref{fig:rsg1d}, top two panels).  It is not clear that a pressure

368: equilibrium can always be established. If the wind velocity is very

369: low, then the pressure at the base of the RSG wind will never be equal

370: to that in the MS shell. In this case, where the wind velocity is set

371: (in an ad-hoc fashion) to a minimum of about 70 km/s, the ram pressure

372: is sufficient to establish a new equilibrium (figure \ref{fig:rsg1d},

373: at time T $\sim$ 4.75e6 years). Once the pressure equilibrium reaches

374: steady-state a region of freely expanding RSG wind is seen, separated

375: from the surrounding medium by a shock. The RSG wind is abruptly

376: decelerated at this shock and piles up against it, forming a thin RSG

377: shell. Note that the RSG wind velocity is much smaller than that of

378: the material into which it is expanding. Thus the expansion of the RSG

379: wind into the MS wind does not result in a wind-blown bubble. Even

380: though the duration of the RSG stage is small (only about 230,000

381: years), the mass loss rate is 2 orders of magnitude higher than in the

382: previous stage. Thus the total amount of mass lost is estimated by

383: GLM96 to be about 18.6 $\msun$.

384: 

385: At the end of the RSG phase the star enters the Wolf-Rayet phase. The

386: wind velocity increases by over two orders of magnitude, to a value of

387: around 2000 $\kms$, while the mass loss rate drops by a factor of a

388: few. The fast, low-density wind from the W-R star collides with the

389: free-streaming RSG wind, creating a thin shell of swept up material,

390: and forming an inner shock that separates the unshocked and shocked

391: winds (figure \ref{fig:wr1d}, top left panel). Given the high velocity

392: of the W-R wind and the fact that the mass-loss is lower only by a

393: factor of a few, the momentum of the W-R shell considerably exceeds

394: that of the RSG shell. The W-R shell collides with the RSG shell in

395: about 10,000 years (figure \ref{fig:wr1d}, time T $\sim$ 4.79 million

396: years).  The collision leads to a reflected shock that moves back into

397: the unshocked W-R wind, and a transmitted shock that enters the thin

398: shell. The emergent shock wave drags the RSG shell along with it as it

399: enters the low density MS bubble.  Eventually this structure will

400: collide with the main-sequence shell (figure \ref{fig:wr1d}, time T

401: $\sim$ 4.88 million years). We note that this whole phase, which has

402: extremely important implications for the bubble and SN evolution, was

403: not modeled by \citet{glm96a}.  The energy transferred to the

404: main-sequence shell is not significant and does not cause appreciable

405: motion of the latter. The collision results in an increase in pressure

406: and a rise in the X-ray emission from the shell. The pressure behind

407: the compressed shell is sufficient to send a reflected shock expanding

408: back into the wind material, that ends up compressing the already

409: shocked material.  The wrinkles seen in the density profile are due to

410: small scale changes in the wind properties. This is partially due to

411: the fact that the boundary conditions are calculated in a

412: discontinuous fashion. It may be possible to interpolate over the

413: terminal wind velocity and mass loss rate as a function of time and

414: store them as continuous functions; however we have not attempted to

415: do so. Note that at the very end of the star's lifetime the wind speed

416: rises to about 3000 $\kms$, whereas the mass loss rate decreases

417: gently, resulting in a drop in the wind density. The reverse shock,

418: which is moving back into a high-density medium, suddenly finds itself

419: plowing through a much lower-density environment. The situation

420: finally reaches an equilibrium when the dynamic pressure of the freely

421: expanding wind equals the pressure behind the reverse shock.

422: 

423: The X-ray luminosity of the bubble over the evolution is shown in

424: Figure \ref{fig:xlumbub}. Our aim is to obtain a reasonable

425: approximation to the X-ray luminosity without carrying out complex

426: calculations, which is not the intention of this paper. We have

427: therefore used the approximate fit to the \citet{rcs76} cooling curve,

428: as suggested by \citet{cf94}: The cooling function $\Lambda = 2.5

429: \times 10^{-27}\, T^{0.5}$ ergs cm$^3$ s$^{-1}$ for temperature T $> 4

430: \times 10^7$ K, and $\Lambda = 6.2 \times 10^{-19}\, T^{-0.6}$ ergs

431: cm$^3$ s$^{-1}$ for temperatures 10$^6 < $ T $< 4 \times 10^7$ K.

432: This cooling function accounts for thermal bremsstrahlung at the

433: higher temperatures, and line emission in the lower range. We see that

434: throughout the early evolution of the bubble, the luminosity of the

435: bubble is quite low, a few times 10$^{32}$ ergs/s. The variations are

436: not as important as the average luminosity indicated.  Using the

437: calculators on the Chandra website we find that a bubble with this

438: luminosity in our galaxy at a distance of a few kiloparsecs would not

439: be observable by the Chandra X-ray satellite, as the background count

440: rate would exceed that from the source. The luminosity increases after

441: the appearance of the W-R wind, and especially after the collision of

442: the W-R wind with the RSG wind. The reason is that the density of the

443: RSG wind, into which the W-R wind is expanding, is much higher than

444: the density of the surrounding bubble. Thus the emission measure is

445: much larger for the X-ray emission from the W-R wind, than for that

446: from the earlier phases. As expected the emission decreases once the

447: W-R wind passes the RSG wind and encounters the low-density MS

448: interior. However it rises again once the expanding shock wave

449: collides with the dense MS shell.

450: 

451: We note that the X-ray luminosity is in general quite low. The

452: \citet{wmc77} picture suggests a very hot interior temperature, and

453: observable diffuse X-ray emission. However several authors have

454: pointed out the paucity of Wolf-Rayet bubbles with diffuse X-ray

455: emission \citep{cgg06, wcmw05,cgggw03}. Furthermore even the bubbles

456: seen show a much lower temperature than expected from the models. Our

457: results suggest that the emission measure from these bubbles is low

458: enough that diffuse emission may not be seen, and this may account for

459: some of the discrepancy. However, the temperatures in our models,

460: although lower than those suggested in the analytic models of

461: \citet{wmc77} due to the mixing and turbulence in the interior, are

462: still around a few times 10$^6$ to 10$^7$ K, much higher than those

463: actually observed. Perhaps this is partly due to the fact that much of

464: the observed X-ray emission may be coming from high-density,

465: lower-temperature clumps in the unstable, turbulent interior. We will

466: investigate this suggestion more carefully in future, using our

467: multi-dimensional simulations.

468: 

469: According to GLM96, the star at this point is in the transition phase

470: from a WN to a WC star. The model calculations terminate at this

471: point, and the density and pressure profiles at the end of its

472: evolution are shown in figure \ref{fig:bub1dfinal}. Note that the

473: overall structure is quite similar to what one would expect from a

474: two-wind interaction, although there are considerable fluctuations in

475: the density profile in the bubble interior. In particular there is one

476: region in the bubble interior where there is an increase in density by

477: almost two orders of magnitude compared to the surroundings. This will

478: be somewhat smoothed out by instabilities in multi-dimensions.

479: 

480: It must be kept in mind that the evolution described above is only

481: 1-dimensional.  In two dimensions, as we shall show below (\S

482: \ref{sec:bub2d}), strong instabilities may develop in the shells, the

483: interior of any surrounding bubble will not be isobaric, and the

484: structure of the circumstellar medium is not as clearcut as in the ID

485: models. It is however clear from this picture (Figure

486: \ref{fig:bub1dfinal}) that in general the medium surrounding a massive

487: star is a low-density medium, surrounded by a high-density thin

488: shell. This is similar to what was assumed in the calculations carried

489: out in Paper 1. It is significant also that the shell size is

490: essentially set by the MS stage, while the composition of the bubble

491: consists mostly of material emitted in the RSG phase. This material

492: would not have gone too far however were it not for the high-momentum

493: W-R wind, which mixes all the material out to the radius of the dense

494: shell. Thus what we refer to as a ``Wolf-Rayet bubble'' is the

495: cumulative consequence of the previous stages of evolution.

496: 

497: \subsection{Interaction of the SN shock with the Surrounding Medium}

498: 

499: \label{sec:sncsm1d}

500: 

501: The 35 $\msun$ model presented here ends its evolution as a 9.15

502: $\msun$ star. In order to study the further evolution, we assume that

503: the star subsequently explodes as a supernova of energy 10$^{51}$

504: ergs.  The outer layers are expelled in the explosion, leaving behind

505: a remnant neutron star of mass 1.4 $\msun$. Thus the amount

506:  of mass ejected in the explosion is about 7.75 $\msun$. Given the

507:  mass and energy, one more parameter, the exact form of the ejecta

508:  density profile, is needed in order to model the SN explosion. As

509:  described in Paper 1, we have assumed that the ejecta are well

510:  described by a power law profile in the outer parts, with a power-law

511:  index of 7, and a constant density profile in the interior

512:  \citep{cf94}. The choice of power-law index is necessarily arbitrary

513:  and meant to be illustrative, although it is useful to remember that

514:  less steep power-laws are thought to be more appropriate to describe

515:  the ejecta of SNRs arising from compact stars \citep{bsbs96}.

516: 

517: The initial interaction of the supernova shock with the free wind was

518: carried out separately on an expanding grid. The wind parameters used

519: were those existing at the end of the star's W-R stage. The shock

520: structure was then interpolated and mapped back onto the first few

521: (typically 75-100) zones of the grid containing the final stages of

522: the W-R star. Inflow boundary conditions are used.

523: 

524: The innermost wind termination shock of the W-R bubble is at a radius

525: of about 11.2 pc.  The SN shock takes about 880 years to reach this

526: point. Note that this time is shorter than that predicted by the

527: self-similar solution \citep{c82, cf94}. The reason is that the

528: self-similar solution assumes the external medium has a negligible

529: velocity, whereas in this case the ambient velocity (due to the W-R

530: wind) is a finite (and non-negligible) fraction of the SN shock

531: velocity. The collision of the SN shock with the inner wind

532: termination shock results in a reflected shock moving back into the SN

533: ejecta and a transmitted shock advancing into the low-density

534: bubble. The reflected shock can be seen (Figure \ref{fig:snbuba1d},

535: 3743 years ) climbing the steep power-law part of the ejecta profile.

536: 

537: The SN shock wave at the start of the simulation was moving close to

538: 13,000 $\kms$ (see Figure \ref{fig:snvel}). The collision with the

539: termination shock reduces the velocity considerably, and the

540: transmitted shock emerges into the interior cavity of the MS bubble

541: with a velocity close to 5,000 $\kms$. The shock continues to sweep up

542: more of the surrounding material, but the low density implies that the

543: swept-up mass is quite small, and therefore its velocity does not

544: change appreciably. As shown in Figure \ref{fig:snbuba1d}, a

545: high-density perturbation exists about 41 pc from the center, with a

546: maximum density of about 0.01 particles cm$^{-3}$ about 47 pc from the

547: center. The collision of the transmitted shock with this region

548: results in a corresponding drop in velocity, and a weak reflected

549: shock whose presence is mainly discernible as a slight perturbation in

550: the pressure profile, but which tends to otherwise blend in the

551: density structure (Figure \ref{fig:snbuba1d}, 9299 years). When the

552: shock hits the highest density part of this region (at about 47 pc)

553: another reflected shock is seen which is slightly stronger (labeled

554: $r_1$ in \ref{fig:snbuba1d}). The collision compresses the dense

555: region, increasing its density (Figure \ref{fig:snbuba1d}, 12531

556: years), and results in a transient increase in the X-ray emission

557: ({figure \ref{fig:xlumsn}).

558: 

559: The transmitted shock that emerges from this collision has a velocity

560: that is closer to 2000 $\kms$. Note that this is lower than the

561: velocity the shock would have if it were expanding in a medium with a

562: constant, low density equal to the bubble interior density. It is the

563: frequent collisions with the higher density fluctuations that tend to

564: slow it down. The forward shock moves roughly at constant velocity

565: till it collides with the main sequence shell, at about 22000 years

566: (Figure \ref{fig:snbubb1d}). The mass of the shell far exceeds the

567: mass of the shocked material colliding with it, so this collision

568: falls in the regime where $\Lambda >> 1$ (see Paper 1).  The shock

569: front merges with the shell, and its velocity drops almost to

570: zero. Since the shock has essentially merged with the shell it is very

571: difficult to track the shock and note its actual radius, and hence

572: velocity, resulting in the velocity fluctuations seen after about

573: 23000 years in Figure \ref{fig:snvel}. Meanwhile, a strong reflected

574: shock resulting from the collision with the dense shell moves back

575: into the already shocked material.  This reflected shock collides with

576: the remains of the previously shocked high-density perturbation,

577: resulting in a weaker forward shock that expands outwards and

578: subsequently impacts the main-sequence shell (Figure

579: \ref{fig:snbubb1d}, around 30000 years). The result of all these

580: frequent collisions is a plethora of shock waves criss-crossing the

581: region of the remnant in the vicinity of the main-sequence shell. A

582: high-pressure region is formed there, and some weaker shocks are later

583: seen to expand outwards and subsequently collide with the

584: main-sequence shell (Figures \ref{fig:snbubc1d} and

585: \ref{fig:snbubd1d}).

586: 

587: In the meanwhile, the reflected shock $r_1$ from the earlier

588: collisions has reached the flat part of the density profile. This

589: reflected shock moves towards the center at a rapidly increasing

590: velocity, with maximum velocities approaching 6,000 $\kms$. On

591: reaching the center, and given our inflow boundary conditions, the

592: shock bounces back. This much weaker re-reflected shock will expand

593: outwards and eventually collide with the main-sequence shell in about

594: another 30,000 years, and the cycle tends to repeat itself (Figures

595: \ref{fig:snbubc1d} and \ref{fig:snbubd1d}). However, the SN density at

596: the origin (center) is decreasing as t$^{-3}$. The density in the

597: interior increases when the ejecta are shocked, although it will then

598: again decline with time. In general then the shocks are continuously

599: interacting with material of lower density, and therefore any

600: radiation signatures from the interaction, such as increasing X-ray

601: emission, are continually diminished. In fact the subsequent reverse

602: shock interactions with the shell hardly result in an increase in the

603: X-ray luminosity of the remnant, although the interaction of the

604: re-reflected shock with the dense main-sequence shell certainly leads

605: to a noticeable increase.

606: 

607: A few salient points of the interaction are noticeable in Figures

608: \ref{fig:snbuba1d} to \ref{fig:snbubd1d}:

609: 

610: \begin{enumerate}

611: \item At several times during the evolution, a variety of reflected,

612: transmitted and re-reflected shocks are visible in the SN density, and

613: especially pressure profile (e.g.~at time 89485 years and 100187

614: years). The pressure and density profiles are very different from

615: those assumed for a SN interacting with a constant density medium or a

616: wind, and therefore the emission computed from these will vary.

617: \item In the 1D calculations the high-density fluctuation that is

618: initially present is visible almost throughout the simulation. In

619: multi-dimensions this will probably be somewhat flattened due to the

620: presence of instabilities. But fluctuations in density will exist in

621: multi-dimensions, and the simulations reasonably depict the behavior

622: in such cases.

623: \item Once the SNR shock impacts the dense shell the evolution is more

624: or less restricted to the W-R bubble for 5-6 doubling times, and the

625: motion of the shell is negligible. Thus the size of the remnant in

626: this period is confined to that of the W-R bubble, and the remnant

627: will appear to have stalled.

628: \item Changes in the radiation signatures during this time are almost

629: entirely due to the effects of the reflected and other shocks within

630: the bubble, and the forward shock has very little role to play.

631: \item The complex surrounding structure results in a variety of shock

632: waves traversing the bubble at any given time. A large range of gas

633: velocities will be observed over the interior of the remnant. When the

634: SNR shock is heading towards the dense shell, gas velocities ranging

635: from -2000 km/s to +2000 km/s are seen in the interior. Once the shock

636: hits the shell the forward shock velocity is considerably reduced, but

637: the gas velocity behind the reverse shock can increase to 5000 km

638: s$^{-1}$ as the shock expands in a continually lower density medium.

639: Thus line emission or absorption spectra from different parts of the

640: remnant will show vastly different velocities, sometimes differing by

641: thousands of km s$^{-1}$. If a spectrum is taken that shows lines

642: arising from different parts of the remnant, it will reveal a very

643: confusing and complicated velocity structure.

644: \item By the end of the calculation the transmitted shock can be seen

645: expanding outwards, pushing the dense shell with it. It will take

646: considerably more time before this shock separates from the shell and

647: is visible as a separate entity (see Paper 1).

648: \end{enumerate}

649: 

650: The evolution of the SNR is thus very different from that seen if it

651: had expanded in the pure interstellar medium. The shock wave entering

652: the MS shell causes the shell to expand. As shown in paper 1, the

653: shock more or less merges with the shell in this case, and the

654: transmitted shock does not seem to appreciably separate from the

655: shell. Over time the density of the shell decreases, but its thickness

656: appears to increase as the forward shock expands without effectively

657: separating from the shell.  Even as late as 150,000 years, or over 6

658: doubling times, the outer shock is seen at a distance of about 84 pc,

659: which means that is has moved just 3 pc in about 125,000 years. A

660: large fraction of the kinetic energy has been converted to thermal

661: energy and emitted as radiation. The ejecta are completely thermalized

662: even before reaching the so-called Sedov stage. In fact the remnant at

663: this stage does not fit anywhere in the classical evolutionary pattern

664: of free expansion, Sedov, radiative stage and eventual dispersal into

665: the ISM. It comes closest to being somewhere in between the Sedov and

666: radiative stages, much closer to the latter.

667: 

668: An important point to note is that most of the gas within the bubble

669: is shocked, and is at temperatures 10$^6$ degrees or higher.

670: Occasionally a passing shock will raise the temperature at some point

671: to greater than 10$^7$ K. The entire remnant will appear luminous in

672: X-rays, although the emission from the outer parts dominates through

673: much of the evolution. The inner regions may brighten up when the

674: reflected shocks collide with them. As the X-ray surface brightness

675: depends on the square of the density, we have computed this value at

676: various different times (Figure \ref{fig:xsbsn}, for details see

677: Paper 1). The quantity plotted in the various frames is the square of

678: the particle density, integrated along the line of sight, at all

679: points where the temperature exceeds 10$^6$ K, and normalized to the

680: maximum value.  This quantity shows approximately the surface

681: brightness of the remnant in X-rays. As can be seen, in most cases the

682: outer regions appear to dominate the surface brightness. It is

683: noteworthy that in the last few plots in Figure \ref{fig:xsbsn}, the

684: emission appears to arise from the entire remnant, but the emission

685: measure at this point is so low that it would likely not be

686: observable.

687: 

688: The total X-ray luminosity from the remnant varies considerably with

689: time, depending strongly on the behavior of the various shock waves

690: criss-crossing the remnant. While an exact computation of the X-ray

691: emission, which would involve taking non-equilibrium processes into

692: account during the early evolution, is far beyond the scope of this

693: paper, Figure \ref{fig:xlumsn} illustrates the evolution of the X-ray

694: emission over the first 150,000 years of the remnant, in a similar

695: manner to Figure \ref{fig:xlumbub}.  The X-ray luminosity is on

696: average about 10$^{34}$ ergs s$^{-1}$, with occasional periods when it

697: increases by a few orders of magnitude.  A large extended source with

698: this average luminosity at a distance of 10 kpc would hardly be

699: visible with the Chandra satellite above the galactic background. Thus

700: the presence of the low density CSM considerably reduces the emission

701: from the remnant. However the periodic brightening would increase the

702: visibility considerably. The initial increase in luminosity is due to

703: the collision of the shock wave with the MS shell, while the secondary

704: maximum at about 89000 years is due to a combination of the

705: re-reflected shock hitting the main sequence shell and various

706: reflected shocks moving back into the ejecta. This shows that even

707: older remnants may experience some brightening in X-rays if the

708: supernova explodes within a pre-existing cavity, rendering them

709: visible even at late times.

710: 

711: The above description illustrates the differences between the

712: evolution of SNRs in a constant density medium as compared to more

713: realistic structured environments sculpted by the pre-supernova

714: progenitor star. In this section we have shown 1D calculations that

715: accurately track the expansion of the SN shock wave through the

716: surrounding medium. The 1D simulations can capture the dynamics of the

717: multitude of shocks that appear to criss-cross the remnant at various

718: times, and enable us to understand the complicated hydrodynamics and

719: kinematics. However, in order to obtain a complete picture one needs

720: to carry out multi-dimensional simulations that can capture effects

721: such the formation of hydrodynamical instabilities and deviations from

722: spherical symmetry which cannot be seen in one-dimensional

723: simulations. To this effect we have also carried out two-dimensional

724: simulations, as described below.

725: 

726: \section{Two-Dimensional Computations}

727: 

728: \subsection{Evolution of the surrounding Medium}

729: \label{sec:bub2d}

730: 

731: Unlike GLM96, we have chosen to carry out 2-dimensional simulations

732: from the start of the MS stage. While computationally this is

733: extremely time-consuming, since the simulation runs for over a million

734: timesteps, we feel that a thorough treatment of the problem requires

735: 2-dimensional modeling of the entire evolution from the beginning of

736: the star's life. As we will show below, the MS bubble in two

737: dimensions shows a much more complicated structure, with a

738: non-isobaric interior that shows considerable density and pressure

739: fluctuations. In our simulations, the RSG and WR shells also become

740: unstable. Only the instability of the WR shell was noticed by

741: GLM96. Our simulations show differences from theirs right from the

742: start. Besides, they did not model the interaction of the WR wind with

743: the MS shell and its consequences, which are important to us to define

744: the pre-SN stage.

745: 

746: The two-dimensional simulations were carried out on a spherical

747: ($r-\theta$) grid. The simulations described used 600 zones in both

748: the radial and azimuthal directions. In order to accurately compute

749: the inner boundary conditions, i.e. the velocity, the density from the

750: mass-loss rate and wind velocity, and the temperature, the simulation

751: runs for about 1.79 million time steps, one good reason why it was not

752: carried out in full two-dimensions by GLM96.

753: 

754: In figure {\ref{fig:bub2d_ms}} we show density images of the evolution

755: of the MS bubble at various timesteps. The radius scale is in

756: parsecs. In the top right corner of each figure we show the current

757: wind parameters - the velocity in km s$^{-1}$ and the mass-loss rate

758: in $\msun yr^{-1}$ - as well as the time in years. The color scale

759: shows the logarithm of the gas density, calibrated in g cm$^{-3}$.

760: 

761: The evolution of the MS bubble in two dimensions starts off as in the

762: 1D case. A thin shell of swept-up material is formed, bounded on the

763: outside by a highly radiative shock and on the inside by a contact

764: discontinuity. The outer shock sweeps up the surrounding ISM into a

765: thin shell. The expansion of the bubble closely resembles that shown

766: in the previous sections and in \citet{wmc77}, with the radius

767: increasing approximately as t$^{0.6}$. The interior is initially

768: isobaric. However after about 75,000 years minor perturbations in

769: density and pressure seem to appear within the interior. Perturbations

770: appear to start close in to the inner shock, and somewhat later just

771: inside of the contact discontinuity. We have not added any

772: perturbations to the initial conditions. The origin of the

773: perturbations can be traced to irregularities that arise at the inner

774: shock, and we suggest that it is the response of the inner shock to

775: the changing wind parameters that gives rise to the fluctuations.  As

776: the wind parameters vary, the shock position varies correspondingly at

777: every timestep. The variation in the position of the inner shock

778: causes variations in the pressure around the shock, and since the

779: shock position is continuously varying, and the interior is subsonic

780: with respect to the reverse shock, the pressure waves do not have

781: enough time to isotropize within the interior. This leads to further

782: variations and density inhomogeneities, which result in the

783: development of turbulence in the interior, visible in Figure

784: \ref{fig:bub2d_ms}.

785: 

786: 

787: The fluctuations in the position of the shock front, and the shear

788: associated with it, result in a considerable amount of vorticity being

789: deposited into the shocked wind. As the shocked wind is expanding

790: outwards, the deposited vorticity is carried out with the wind, and

791: does not dissipate to smaller scales. Pressure variations due to the

792: deposition of vorticity near the inner shock lead to density

793: variations within the region, as regions at different temperatures

794: cool differently. The net result is the formation of higher density

795: regions within the interior. The stellar wind is forced to flow around

796: these obstacles, leading to slower moving regions within the radial

797: flow. The buildup of vorticity in these regions is clearly

798: demonstrated in Fig {\ref{fig:bub2d_ms}}.  The cumulative effect is

799: the growth of turbulence within the interior.

800: 

801: At the same time, we find that the dense, thin shell also shows some

802: signs of shear instabilities. Although we do not include any initial

803: perturbations in our calculation, small perturbations can be initiated

804: within the shell due to the shear flow between the contact

805: discontinuity and the cavity just interior to it, due to the

806: difference in the flow velocities. Such perturbations can already be

807: seen in the top right panel in figure {\ref{fig:bub2d_ms}}. These

808: instabilities tend to persist throughout the growth of the bubble, but

809: their effect is minor, and does not lead to any significant distortion

810: of the spherical shell. This is clearly seen in the last panel (bottom

811: right), which shows the structure of the bubble towards the end of the

812: MS stage. This complicated turbulent structure is the medium into

813: which the RSG wind will expand.

814: 

815: We note here that in previous low-resolution simulations \citep{d01,

816: d02, d04}, we have reported the presence of an unstable thin shell in

817: the main sequence phase, which we attributed to a Vishniac-type thin

818: shell instability \citep{v83, vr89}. This was seen in simulations with

819: grid resolutions from 200 $\times$ 200 to 400 $\times$ 400. However,

820: this instability disappeared when we ran simulations with the even

821: higher resolution reported herein. We feel that the higher resolution

822: simulations are far more believable, although we have been unable to

823: understand the reasons for triggering of the instability at lower

824: resolutions. However, the presence of this instability (or lack of it)

825: does not affect the subsequent evolution of the bubble or the SN

826: shock, which is our main concern for this paper. The remaining

827: behavior is seen as expected at both lower and higher resolutions,

828: with the higher resolutions providing increasingly sharper clarity to

829: view the R-T instabilities and turbulence within the interior. It is

830: curious though that we have noted similar behavior when carrying out

831: simulations of a 40 $\msun$ star from a stellar model provided by Dr

832: Georges Meynet. The MS shell instability was seen at a resolution of

833: 400 $\times$ 400 zones, but not at a higher resolution of 600 $\times$

834: 600 zones. The current situation should be unstable to the Vishniac

835: ram pressure instability \citep{v83}, but the appearance of this would

836: differ from the finger-like perturbations that we had written about

837: earlier. Thus the presence (or absence) of an instability in the MS

838: shell is still a puzzle that remains to be solved.

839: 

840: The interior of the cavity thus differs considerably from the

841: isotropic spherical bubble assumed by GLM96.  At the end of about 4.5

842: million years, with the MS shell at a radius close to 75pc from the

843: star, the wind velocity begins to decrease rapidly as the star enters

844: the RSG stage. As mentioned earlier, the higher mass loss rate and

845: lower velocity of the RSG wind causes the pressure equilibrium within

846: the bubble to change, and a termination shock is formed where the ram

847: pressure of the RSG wind is equal to the thermal pressure in the

848: interior. The RSG shell is also found to be unstable, as shown in

849: figure \ref{fig:bub2d_pms}. GLM96 suggest that the shell is stable to

850: Rayleigh-Taylor instabilities, but this would be strictly true only if

851: the wind velocity was constant or monotonic.  In this case the RSG

852: wind velocity varies from about 95 $\kms$ to about 70 $\kms$ over the

853: evolution, and the pressure equilibrium also varies. The RSG shell is

854: found to be decelerating as it expands outwards. The high density

855: behind the shock decelerated by the high pressure of the wind-blown

856: cavity within the MS bubble provides the right conditions for the R-T

857: instability, and we see clearly the growth of Rayleigh-Taylor fingers

858: in our simulations. As the dense shell is being decelerated the

859: fingers tend to expand outwards. Figure \ref{fig:bub2dzoom} zooms in

860: on the RSG region towards the end of the RSG stage, showing vividly

861: the thin expanding R-T filaments, with slightly bulbous heads due to

862: the formation of Kelvin-Helmholtz instabilities at the tip of the

863: fingers, resulting from the shear flow between the fingers and the

864: surrounding material. The filaments grow in length over the evolution

865: of the shell.

866: 

867: At the end of the RSG stage the surface temperature of the star begins

868: to increase, it sheds its outer envelope and enters the W-R phase. The

869: WR wind expanding into the RSG wind will form a miniature version of

870: the MS bubble, complete with an inner shock, a contact discontinuity

871: and a thin shell. The W-R shell is also found to be unstable and

872: filamentary structure can be seen at the inner edge of the shell. We

873: attribute the growth of these structures also to the development of

874: the Rayleigh-Taylor instability.

875: 

876: In the ideal course of the collision of two winds with constant

877: wind-properties, the wind-blown shell would expand with a constant

878: velocity, and the conditions would not allow for the development of

879: the R-T instability. However in this particular case, although the W-R

880: wind is expanding within the RSG wind, there are some important

881: differences. Firstly the wind parameters are not constant, either in

882: the W-R or the RSG stage. Secondly, the W-R bubble is so highly

883: supersonic that it does not cool efficiently, and therefore the outer

884: shock is not highly radiative, and the shell that forms is not

885: thin. The formation of a thin, dense, cool shell may have led to the

886: development of thin-shell instabilities. Instead due to the variation

887: in parameters we find that the W-R shell is accelerating down the ramp

888: of the RSG wind. The pressure inside the dense shell (in the interior

889: cavity of the W-R bubble) is much larger than the pressure outside (in

890: the RSG wind). The high pressure pushing on the high density shell

891: provides the appropriate conditions for the R-T instability to

892: develop. In this case however, as opposed to the RSG case, the fingers

893: expand inwards, from the high density shell into the bubble, rather

894: than outwards, as is expected from the physical conditions. The

895: structure at this stage is shown in Figure \ref{fig:bub2d_pms}, and

896: magnified in the upper right panel in Figure \ref{fig:bub2dzoom}.

897: 

898: It is interesting to note that although the formation of the R-T

899: instability is seen in both the RSG and W-R cases, the structure looks

900: quite different. One reason is that in the RSG case the shell is

901: decelerating outwards, whereas in the W-R case the shell is

902: accelerating outwards. Another reason is that the density contrast, or

903: the Atwood number, is very different between the two cases. The time

904: that the instability can develop is very short in the W-R case, as the

905: W-R wind quickly advances in the RSG wind before colliding with the

906: shell. For these reasons the appearance of the R-T fingers is

907: different in the two cases, as is apparent in figure

908: \ref{fig:bub2dzoom}. In our simulation we do not have enough

909: resolution in the radial direction to study the instability in the W-R

910: wind in great detail.

911: 

912: The extremely large velocity of the W-R wind (on the order of 3300

913: $\kms$), coupled with a mass loss rate which is only a factor of a few

914: smaller than that of the RSG, means that the ram pressure of the W-R

915: wind far exceeds that of the RSG wind.  The W-R wind slams into the

916: RSG shell, causing it to fragment completely, and is seen to break-out

917: of the shell at many different points (Figure \ref{fig:bub2dzoom}),

918: primarily along the axis, (although this may be a numerical

919: effect). The formation of the funnel-like feature close to the axis is

920: certainly a numerical effect, due to the presence of very narrow zones

921: along the axis. However the simulation clearly reveals the break-up of

922: the RSG shell, with the fragments being dispersed throughout the

923: interior of the W-R bubble.  The momentum of the W-R wind carries some

924: of the RSG material along with it towards the MS shell (Figure

925: \ref{fig:bub2d_pms}).  The W-R gas collides with the MS shell and

926: bounces back, in the process distributing the RSG material throughout

927: the interior of the nebula. Thus although the RSG wind itself was not

928: able to penetrate even a fifth of the MS radius, much of the mass in

929: the interior of the nebula will be comprised of material dredged up

930: from the RSG stage. This material is then shocked by the W-R wind,

931: perhaps leading to an overabundance of N and/or C in the nebular

932: material. The collision of the W-R wind with the MS shell results in a

933: shock being driven into the shell and a reflected shock back into the

934: unshocked wind. This reflected shock penetrates as far back as it can

935: towards the center, before its advance is arrested by the ram pressure

936: of the outflowing W-R wind, and an inward facing wind-termination

937: shock forms.

938: 

939: Figure \ref{fig:bub2dzoom}, bottom right panel, shows a combined

940: contour and vector plot of the structure of the bubble at the end of

941: the simulation. The complex nature of the velocity field is due to the

942: various evolutionary phases, although the imprint of the W-R phase is

943: unmistakable in the inward flow near the axis and the complicated

944: behavior in the equatorial regions. The intersection of the ram

945: pressure of the radially outward flow and the thermal pressure behind

946: the reflected shock leads to the formation of the W-R wind termination

947: shock. It is of consequence to note that the wind termination shock is

948: {\em not spherical} but slightly elongated towards the equatorial

949: latitudes. This is due to a combination of factors. The unstable W-R

950: wind pushes out on the unstable RSG wind, fragmenting the RSG

951: shell. The material is not carried out in a spherically symmetrical

952: manner. This asphericity is enhanced when it travels through the

953: cavity due to the pressure and and density fluctuations, and again on

954: this material colliding with the shell. Therefore, when the reflected

955: shock's progress towards the center is finally halted, the equilibrium

956: between the isotropic pressure of the W-R wind and the varying thermal

957: pressure behind the reflected shock leads to an aspherical wind

958: termination shock. This has important consequences for the subsequent

959: evolution of the SN shock wave.

960: 

961: Qualitatively a radial cut through the nebula resembles the 1D

962: profile. However there is considerably more structure in the 2D

963: profile.  Also since the 2D resolution is about a third of the 1D case

964: the structure is more smeared out, and the sharp shock fronts of the

965: 1D model are spread out over a larger distance in 2D.

966: 

967: We note that our final picture of the W-R bubble agrees well with

968: observations, including the large size and the complicated internal

969: structure \citep{cec03}. The size of the bubble of course is a direct

970: consequence of the external density that we have assumed. A much

971: higher density would lead to a smaller size. This would however just

972: compress the entire picture into a small radius, and therefore

973: increase the density in the bubble interior, but it would not

974: appreciably change the dynamics and kinematics that we see.

975: 

976: \subsection{Evolution of the SN Shock Wave}  

977: 

978: 

979: In the next stage the star was assumed to undergo a supernova

980: explosion as outlined in \S \ref{sec:sncsm1d}. The SN profile was

981: interpolated onto the CSM grid. The SN profile occupies 63 zones, and

982: the interpolation results in a smearing out of the shock front and

983: other features. The simulation was then run to study the evolution of

984: the SN shock wave into the surrounding medium. In this discussion we

985: will concentrate on describing the main multi-dimensional effects and

986: departures from the spherically symmetric case discussed in \S

987: \ref{sec:sncsm1d}. 

988: 

989: The overall evolution of the SN shock wave proceeds as in the 1D case,

990: but with one major difference - the SN shock does not remain

991: spherical. We elaborate on this aspect, and its consequences, below.

992: 

993: The evolution of the SN shock wave in the freely expanding wind

994: proceeds as expected, and in a fashion similar to the 1D case, up

995: until the time that the shock reaches the wind termination shock. Note

996: that due to the aspherical nature of the wind-termination shock, the

997: interaction first takes place in the region of the symmetry axis. This

998: is shown in Figure \ref{fig:snbub2dpre} (at 2597 years), which shows

999: snapshots of the pressure at various times during the evolution. The

1000: pressure is chosen as a quantity to highlight the shocked interaction

1001: region between the forward and reverse SN shocks. As explained in \S

1002: \ref{sec:sncsm1d}, the interaction results in a transmitted and a

1003: reflected shock wave. The transmitted shock wave has been decelerated

1004: by the interaction with the wind shock, and is therefore slower than

1005: the rest of the shock, which is still undecelerated as it has not yet

1006: encountered the wind shock. Thus this portion lags slightly behind the

1007: rest of the shock wave. As the next part of the shock wave hits the

1008: wind shock, it also gets decelerated. Since the velocity of the SN

1009: shock is decreasing as it moves outwards, each subsequent impact of

1010: some part of the SN shock with the wind shock happens at a lower

1011: velocity. The net result of this impact is that different parts of the

1012: transmitted (and reflected) shocks travel outwards at slightly

1013: different velocities. Both shock waves assume the shape of the

1014: aspherical wind shock to a certain degree.

1015: 

1016: The effect is further accentuated by the fact that the interior cavity

1017: is not isotropic, but shows considerable variation in pressure and

1018: density, with several regions that are higher density than the

1019: surroundings. Furthermore, as shown in Figure \ref{fig:bub2dzoom},

1020: lower right panel, the interior also contains several vortices with

1021: velocities as high as a thousand km/s. The interaction of the SN shock

1022: wave with this highly turbulent material, and the fact that the

1023: average pressure behind the shock wave is not significantly higher

1024: than the pressure of the interior, causes further corrugations in the

1025: shock wave. The result is a highly wrinkled shock wave, with various

1026: bumps and wiggles, by the time it reaches the outer dense shell.

1027: 

1028: Figure \ref{fig:snbub2dden} illustrates how the spherical shock slowly

1029: becomes a wrinkled and corrugated structure. In order to display this

1030: effect vividly, we plot the density profile of the shockwave at

1031: various timesteps as it expands towards the wind-blown shell. Although

1032: the outer and inner shock are not very well resolved in the density

1033: color scale, it is easy to spot their location, especially using

1034: figure \ref{fig:snbub2dpre}. In Figure \ref{fig:snbub2dden} at time

1035: 4932 years, the shock wave is encountering a density fluctuation

1036: within the bubble. The shock appears slightly depressed in that

1037: region, and a reverse shock (purple in color) can be seen reflecting

1038: off the perturbation. At later times this event is repeated, adding to

1039: the asymmetry of the shock wave each time, until finally the shock

1040: wave becomes extremely distorted just before it is about to collide

1041: with the shell, at about 20000 years.

1042: 

1043: 

1044: 

1045: Since the shock is not spherical, the expansion is not completely

1046: radial. The bumps in the shock wave, and the crinkled nature of the

1047: shock, results in just one or two extended sections of the shock

1048: hitting the dense shell at about 22,000 years, as opposed to the

1049: entire shock wave in the one-dimensional case. Each collision will

1050: result in a rise in the X-ray and optical emission, as we have seen

1051: before (see Paper 1). However, in this case, since the shock collides

1052: with the shell in a piecemeal fashion, different parts of the shell

1053: will brighten up in the X-ray and optical at different

1054: times. Therefore, instead of seeing a glowing shell, what will be seen

1055: are different sections of the shell ``lighting up'' at different

1056: times. Eventually, in this case over a timescale of about 15000 years,

1057: the entire shell will brighten up as the entire extent of the shock

1058: has collided with the shell.

1059: 

1060: Another effect of this piecemeal collision is that instead of having

1061: one reflected shock bouncing off the shell, we will have several small

1062: ``shocklets'', with velocity vectors pointing in different

1063: directions. This results in an even more aspherical reverse shock, as

1064: is shown in figure \ref{fig:snbub2dpre} after about 35000

1065: years. Different parts of the reverse shock will then move at

1066: different velocities towards the center. Since the velocities are not

1067: all radial, some portions will advance preferentially towards the

1068: symmetry axis or the equatorial axis, and collide with it. In the last

1069: panel in figure \ref{fig:bub2dzoom}, just after 45000 years, one part

1070: of the reflected shock can be clearly seen to have collided with the

1071: axis of symmetry, and a re-reflected is just starting to move

1072: back. Note that this shock is directed almost perpendicular to the

1073: axis. In part this is due to an axis effect, as has often been

1074: discussed for two-dimensional axisymmetric simulations. Therefore we

1075: caution into reading too much into the {\em specific} behavior of the

1076: shock outlined in this case, emphasizing more the existence of an

1077: overall global asymmetry, and reflected shocks that are not moving

1078: radially inwards.

1079: 

1080: Due to this piecewise interaction the reflected shock takes a longer

1081: time to reach the center in multi-dimensions. In the one-dimensional

1082: case the reflected shock takes about 31000 years or so to reach the

1083: center and bounce back. In the two-dimensional case, only a very small

1084: fraction of the reverse shock has reached the center in almost 45,000

1085: years. This is not surprising, considering that the entire forward

1086: shock itself has barely just finished colliding with the dense shell,

1087: and that the reverse ``shocklets'' that are formed have a significant

1088: azimuthal component rather than just a radial, centrally directed

1089: component.

1090: 

1091: 

1092: The SN shock wave gets essentially trapped in the dense shell in this

1093: case, as outlined in \S \ref{sec:sncsm1d}. A transmitted shock

1094: eventually emerges, as shown in the one-dimensional case. However in

1095: this case the transmitted shock takes an even longer time to emerge as

1096: compared to the one-dimensional case, for the same reasons that we

1097: have outlined above, mainly the slower motion of the SN shock wave and

1098: the larger time taken for the entire shock wave to collide with the

1099: shell. This will thus introduce an even larger degree of asymmetry in

1100: the transmitted shock wave as different parts of the shock emerge from

1101: the shell at different times. The appearance will be of a very

1102: aspherical remnant. Unfortunately due to computational time

1103: constraints we have not carried out the simulations further. In future

1104: we plan to use a parallel adaptive mesh code to carry out this

1105: simulation in three-dimensions, thus removing any axis of symmetry.

1106: 

1107: The final panel in Figure \ref{fig:snbub2dden}, at about 42000 years,

1108: displays the density profile long after shock-shell interaction. The

1109: filamentary nature of the resultant remnant is very clearly evident in

1110: this picture, resulting from dense shell material expanding into the

1111: low density cavity after the shock-shell interaction, suggestive of

1112: Richtmeyer-Meshkov and Rayleigh-Taylor instabilities. The final

1113: morphology is a combination of the various asymmetries and the

1114: piecemeal collision, together with the various pieces of reverse shock

1115: that also interact with each other. The result is a spaghetti-like

1116: mesh of filaments emanating from the inner walls of the nebula. The

1117: color scale gives the logarithm of the density. It can be seen that

1118: the filaments are about two orders of magnitude higher in density.

1119: 

1120: 

1121: It is clear from the description that both the transmitted and

1122: reflected shocks may be considerably aspherical. Furthermore, the

1123: radial symmetry of the remnant is gradually lost over increasing

1124: interactions. In our case we started with the SN shock evolving in a

1125: highly turbulent, but still spherical bubble, yet ended up with a

1126: structure with a very aspherical shock wave and reflected shock. In a

1127: case where the circumstellar bubble around the remnant is itself not

1128: spherical but bipolar, such as the very well observed SN 1987A, this

1129: effect will be even more pronounced. Thus many of our simplistic

1130: notions of radially expanding outgoing and incoming shock waves in

1131: SNRs need to be re-evaluated.

1132: 

1133: \section{Summary and Discussion}

1134: 

1135: Continuing our series of papers on the evolution of SNe in structured

1136: wind environments, we have herein explored the case of a 35 $\msun$

1137: Wolf-Rayet star via numerical simulations. The star starts its life on

1138: the main-sequence as an O star, evolved through the red supergiant

1139: phase, and then becomes a Wolf-Rayet star. As it evolves, it loses

1140: mass via winds, whose properties change dramatically over the entire

1141: evolution. The mass-loss leads to the formation of a structured

1142: wind-blown bubble around the star. A main-sequence bubble with a dense

1143: shell is formed initially. The slow and dense RSG wind does not expand

1144: very far inside this shell, and its low velocity does not result in

1145: the formation of a wind-blown bubble, but a termination shock is

1146: formed and the wind piles up against it. The high-momentum of the

1147: Wolf-Rayet wind pushes the RSG material outwards, until it collides

1148: with the MS shell and rebounds back. Finally a Wolf-Rayet wind

1149: termination shock is formed at the radius where the ram pressure of

1150: the freely expanding wind and the thermal pressure behind the shock are

1151: equal.

1152: 

1153: This is the basic description of the evolution. Multi-dimensional

1154: calculations add further details to the overall picture. The

1155: constantly fluctuating position of the reverse shock in the MS phase

1156: results in the buildup of pressure fluctuations and the deposition of

1157: vorticity into the shocked wind. The vorticity is carried out with the

1158: shocked expanding flow. These effects result in the formation of

1159: eddies and the onset of turbulence within the shocked medium.

1160: 

1161: The RSG wind and the W-R wind shell are also found to be unstable to

1162: R-T instability. The high-momentum W-R wind pushes out on the RSG

1163: material, causing it to fragment, and carrying the material far beyond

1164: it would have otherwise traveled. This is important for the formation

1165: of W-R bubbles - they may be composed in some cases mainly of RSG

1166: material, perhaps material that has been dredged up. The W-R wind is

1167: instrumental in dispersing this material over an area of tens of

1168: parsecs, which the RSG wind by itself, with its low velocity, could

1169: not accomplish.

1170: 

1171: The global structure of the bubble in multi-dimensions is not very

1172: different from that predicted in one-dimension, but with considerable

1173: fluctuations in the density and pressure of the interior. The various

1174: instabilities and turbulence result in a Wolf-Rayet wind-termination

1175: shock that is not spherical, but slightly elongated towards equatorial

1176: latitudes. This has implications for the subsequent evolution of the

1177: SN shock wave within this medium.

1178: 

1179: We note here that in some cases such as the one presented here, the

1180: Wolf-Rayet wind-termination shock is not formed by the direct

1181: interaction of the Wolf-Rayet wind with the surrounding medium, or

1182: even with the wind from a pre-existing stage. Rather it results from

1183: the Wolf-Rayet wind interaction with the MS shell, and a reflected

1184: shock bouncing back until pressure equilibration with the ram pressure

1185: of the freely expanding wind is achieved. This means that it is not

1186: possible to predict the radius of this shock {\it a priori} from

1187: wind-wind interaction, as is sometimes done nowadays for calculations

1188: of the structure around gamma-ray bursts. This could lead to an

1189: erroneous answer. {\em In calculating the radius of the Wolf-Rayet

1190: wind-termination shock, one must take into account the previous

1191: evolution of the circumstellar medium}.

1192: 

1193: Our results have important implications for the surroundings of

1194: massive stars, and the environment in which supernovae, and possibly

1195: gamma-ray bursts, evolve. We do point out that in our calculations we

1196: have not considered the effects of the ionizing radiation from the

1197: star. These were briefly detailed in Paper 1. It is possible that the

1198: ionization front sweeping through the star may have a dynamical effect

1199: that needs to be taken into account. We are now working on a code that

1200: includes the effects of the ionization from the star. These

1201: simulations will be detailed in forthcoming papers.

1202: 

1203: During the writing of this paper we have realized that the ionizing

1204: effects in a similar case have recently been considered in simulations

1205: by \citet{fhy06}. Unfortunately they do not provide details of the

1206: shock structures in a one-dimensional model, which would have been

1207: very useful to compare the direct effects of the ionization front. And

1208: even though their highest resolution was larger than what we use here,

1209: the fuzziness of their published figures precludes a detailed

1210: comparison with this work. They see the formation of an ionization

1211: front instability in the dense shell in the MS stage, but they do not

1212: see in their simulations the Rayleigh-Taylor instabilities that are

1213: mentioned herein in the RSG and W-R stages. Conversely they note the

1214: formation of an R-T instability when the W-R wind passes over the

1215: boundary between the RSG and MS wind. We do see the formation of

1216: filamentary structure in this case (figure \ref{fig:bub2dzoom}, lower

1217: left panel) but it is not clear that this is just due to the existing

1218: instabilities that we have noted, the formation of a new R-T

1219: instability, or even a combination of the two. Since they started with

1220: a surrounding medium density different from that used here, the size

1221: of the bubble and its properties would differ correspondingly, so it

1222: is difficult to get a good read on how much the parameters such as

1223: density, pressure and velocity are affected by the ionizing

1224: radiation. The ionization front will raise the temperature, and

1225: therefore the pressure, of the ionized region, which will affect the

1226: evolution of the wind bubble in that region. One point of direct

1227: comparison is the X-ray luminosity of the bubble over time, which can

1228: be directly compared to Figure 19 of their paper. The variations in

1229: the luminosity are much larger in our plot, possibly due to our more

1230: approximate method. The luminosity calculated by us is slightly

1231: larger, by a factor of a few, compared to their plot. This is partly

1232: because we include all emission larger than 10$^6$K, whereas they

1233: include only emission from 0.1-2.4KeV. However the overall similarity

1234: in the plots, especially the large rise in luminosity during the W-R

1235: phase, is striking. This also attests to the validity of our simpler

1236: method of calculation.  In the final analysis, and without access to

1237: their data, we conclude from their Figure 15 that while the formation

1238: of an HII region plays a role, the pre-SN state of the gas at the end

1239: of their calculation is not significantly different from that found

1240: herein, and this is the main quantity that we are interested in.

1241: 

1242: Our most interesting results deal with the expansion of the SN shock

1243: wave within this medium, which is the main goal of this paper. We find

1244: several interesting effects. The evolution of the SN shock wave is

1245: confined within this bubble for a substantially large period of time

1246: of several doubling periods after the SN shock wave interacts with the

1247: dense shell. This will often be the case for SNe that arise from W-R

1248: stars (Type Ic SNe). The overall level of the emission from the

1249: remnant is significantly reduced due to the lower density within the

1250: cavity. Due to the fluctuations in density and pressure within the

1251: bubble, there are several shocks and rarefaction waves seen

1252: criss-crossing the remnant. A large range of velocities will be seen

1253: at any given time throughout the remnant. The interior is almost

1254: completely thermalized and heated to high, X-ray emitting temperatures

1255: throughout. Thus, as suggested in Paper 1, SN shock waves in shells

1256: could be one explanation for remnants which show centrally peaked

1257: emission. However it must be pointed out that at least in this

1258: particular case the density within the remnant is so low that the

1259: emission measure in the interior is very small, and most X-ray

1260: emission will be seen to arise only from the edge of the remnant.

1261: 

1262: The spherical SN shock wave interacting with an aspherical

1263: wind-termination shock results in an aspherical, and considerably

1264: wrinkled, transmitted shock. The corrugated nature of the shock wave

1265: results in the interaction of the shock with the dense shell taking

1266: place in a bit-by-bit fashion, with different parts of the shockwave

1267: interacting with the dense shell at different times. As pointed out in

1268: \S \ref{sec:sncsm1d}, the interaction of the shock wave with the shell

1269: leads to a brightening up of the shell due to an increase in X-ray and

1270: optical emission. In this case, as different parts of the shock wave

1271: collide with the shell at different times, the shell will brighten up

1272: in different places at different times, almost like blinking Christmas

1273: lights.

1274: 

1275: This effect is reminiscent of the situation in SN 1987A. Optical

1276: observations have shown the presence of very bright "hot spots" on the

1277: equatorial ring surrounding the SN. The first spot was seen around

1278: 1997, and gradually over the next few years many more have been

1279: visible. The latest HST pictures show spots almost all the way around

1280: the ring. The ring itself is known to be the equatorial waist of a

1281: circumstellar bipolar nebula surrounding the progenitor star, and the

1282: spots can be interpreted as the interaction of a wrinkled shock wave

1283: with the ring. This could be one example of the kind of situation

1284: outlined in the previous paragraph. We caution that our simulations

1285: are not mean to represent the situation in SN 1987A, which is

1286: considerably more complex, and whose progenitor was probably a much

1287: lower mass B3Ia star. Furthermore, in the case of SN 1987A we know

1288: that the ionization from the star is important in creating an HII

1289: region interior to the dense shell \citep{cd95}. And that perhaps

1290: there are fingers (instabilities) pushing inwards from the ring with

1291: which the shock is interacting. But nevertheless, the similarities are

1292: intriguing. Even though this particular simulation may not be

1293: representative of SN 1987A, it shows that it is possible for the SN

1294: shell interaction within a bubble to occur in a discontinuous fashion

1295: due to a wrinkled shock wave. SN 1987A may be the rare case where it

1296: is possible to investigate these effects.

1297: 

1298: Our results provide useful pointers for investigating SNRs in

1299: wind-blown cavities. A case in point is the Oxygen-rich remnant RCW

1300: 86. For several reason, including the low $n_e t$ values, the faint

1301: emission and the X-ray profiles, it has been suggested \citep{vkb97}

1302: that the remnant was formed in a wind-blown cavity. Our simulations

1303: could be useful in testing some of these theories and predicting the

1304: future evolution. Another remnant which may have been formed in a

1305: wind-blown cavity is G292.0+1.8. This remnant shows the interesting

1306: presence of instabilities, identified by some authors as

1307: Rayleigh-Taylor instabilities that arise from the initial explosion

1308: \citep{ghw05}. We note that in our simulations, such filamentary

1309: structures are also formed after the shock-shell interaction, while

1310: the reverse shock is headed back into the remnant and the transmitted

1311: shock is trapped in the dense shell. This may provide an alternative

1312: explanation for the presence of such structures. In future papers we

1313: hope to study individual remnants in much further detail.

1314: 

1315: Our simulations illustrate that it is possible to start with a

1316: spherical shock wave from a SN and still end up with a highly

1317: aspherical remnant due to the complexity of the surrounding

1318: medium. And as noted we have not even taken global asymmetries in the

1319: surrounding medium into account, as in the case of SN 1987A. Although

1320: we caution against reading too much into the specific details of a

1321: single calculation, it serves to illustrate the basic point that SN

1322: evolution in wind blown cavities differs considerably from that in a

1323: wind or a constant density medium, and that the structure and

1324: evolution of the remnant, its dynamics and kinematics, may all change

1325: as a result. Furthermore, the X-ray emission from the remnant evolving

1326: in the low-density cavity may be considerably reduced, with occasional

1327: periods of increased luminosity. It is therefore necessary to

1328: investigate SN shock waves while taking an accurate picture of the

1329: ambient medium into which it expands. This however is only possible if

1330: stellar evolution calculations including the parameters of the

1331: mass-loss from massive stars for every timestep during its evolution

1332: are readily available. Fortunately, the situation is consistently

1333: improving, and more and more such calculations are now being

1334: published.

1335: 

1336: In future papers we will investigate further the evolution of SN shock

1337: waves in environments created by the pre-SN star. We will look at the

1338: effects of more massive stars, and especially rotating stars, where

1339: strong rotation leads to the formation of a wind that is faster and

1340: denser at the poles of the star as compared to the equator

1341: \citep{md01, do02}. These environments can have considerable effect on

1342: the evolution of the shock wave.

1343: 

1344: \acknowledgements

1345: 

1346: VVDs research is supported by award \# AST-0319261 from the National

1347: Science Foundation, and by NASA through grant \# HST-AR-10649 awarded

1348: by the Space Science Telescope Institute. We would like to acknowledge

1349: several very constructive discussions with Roger Chevalier, especially

1350: in identifying the various instabilities that have been seen. Comments

1351: and suggestions from John Blondin, Thierry Foglizzo, and Robin

1352: Williams have proved extremely useful. We would like to thank the 

1353: anonymous referee for some very useful comments and suggestions.

1354: 

1355: 

1356: 

1357: \clearpage

1358: 

1359: \begin{thebibliography}{}

1360: \bibitem[Bandiera(1987)]{b87} Bandiera, R. 1987, \apj, 319, 885

1361: Lundqvist, P. 1993, \apj, 405, 337

1362: \bibitem[Blondin \& Lundqvist(1993)]{bl93} Blondin, J. M., \&

1363: Lundqvist, P.~1993, \apj, 405, 337

1364: \bibitem[Borkowski et al.(1996)]{bsbs96} Borkowski, K., Szymkowiak,

1365: A. E., Blondin, J. M. \& Sarazin, C. L. 1996, \apj, 466, 866

1366: \bibitem[Brighenti \& D'Ercole(1997)]{bd97} Brighenti, F., \&

1367: D'Ercole, A.~1997, \mnras, 285, 387

1368: \bibitem[Cappa et al.(2003)]{cec03} Cappa, C. E., Arnal, E. M.,

1369: Cichowolski, S., Goss, W. M., Pineault, S, in ``A Massive Star

1370: Odyssey: From Main Sequence to Supernova'', IAUS 212, Eds. K. A. van

1371: der Hucht, A. Herrero, \& C. Esteban, SF: ASP, 596

1372: \bibitem[Chevalier \& Dwarkadas(1995)]{cd95} Chevalier, R. A. \&

1373: Dwarkadas, V. V. 1995, \apj, 452, L45

1374: \bibitem[Chevalier \& Fransson(1994)]{cf94} Chevalier, R. A. \&

1375: Fransson, C. 1994, \apj, 420, 268

1376: \bibitem[Chevalier \& Liang(1989)]{cl89} Chevalier, R. A. \&

1377: Liang, E. P. 1989, \apj, 344, 332

1378: \bibitem[Chevalier(1982)]{c82} Chevalier, R. A. 1982, \apj, 258, 790

1379: \bibitem[Chu, Gruendl, \& Guerrero(2006)]{cgg06} Chu, Y.-H., Gruendl,

1380: R.~A., \& Guerrero, M.~A.~2006, in: Proceedings of the "The X-ray

1381: Universe 2005", Ed. A. Wilson., ESA SP-604, Volume 1, (Noordwijk: ESA

1382: Publications Division), 363

1383: \bibitem[Chu et al.(2003)]{cgggw03} Chu, Y.-H., Guerrero, M.~A.,

1384: Gruendl, R.~A., García-Segura, G., \& Wendker, H. L.~2003, \apj, 599,

1385: 1189

1386: \bibitem[Ciotti \& D'Ercole(1989)]{cd89} Ciotti, L., \& D'Ercole,

1387: A. 1989, \aap, 215, 347

1388: \bibitem[Colella \& Woodward(1984)]{cw84} Colella, P., \& Woodward,

1389: P. R. 1984, J. Comput. Phys., 59, 264

1390: \bibitem[Dwarkadas(2007b)]{d07b} Dwarkadas, V. V, 2007, in preparation

1391: \bibitem[Dwarkadas(2007a)]{d07a} Dwarkadas, V. V, 2007, ApSS, 307, 153

1392: \bibitem[Dwarkadas(2005)]{d05} Dwarkadas, V. V, 2005, \apj, 630, 892

1393: \bibitem[Dwarkadas(2004)]{d04} Dwarkadas, V. V, 2004, in ``Cosmic

1394: explosions in three dimensions : Asymmetries in supernovae and

1395: gamma-ray bursts'', Eds P. Hoflich, P. Kumar and J. C. Wheeler,

1396: (Cambridge:CUP), 274

1397: \bibitem[Dwarkadas \& Owocki(2002)]{do02} Dwarkadas, V. V., \& Owocki,

1398: S. P. 2002, \apj, 581, 1337

1399: \bibitem[Dwarkadas(2002)]{d02} Dwarkadas, V. V, 2002, in

1400: ``Interacting Winds from Massive Stars'', ASP Conference Proceedings

1401: 160, eds A.~F. J. Moffat and N.~St-Louis, (San Francisco: ASP), 141

1402: \bibitem[Dwarkadas(2001)]{d01} Dwarkadas, V. V, 2001, JKAS, 34, 243

1403: \bibitem[Dwarkadas \& Balick(1998)]{db98} Dwarkadas, V. V., \& Balick,

1404: B. 1998, \apj, 497, 267

1405: \bibitem[Dwarkadas et al.(1996)]{dcb96} Dwarkadas, V. V., Chevalier,

1406: R. A., \& Blondin, J. 1996, ApJ, 457, 773

1407: \bibitem[Frank \& Mellema(1994)]{fm94} Frank, A., \& Mellema, G. 1994, \aap, 

1408: 289, 937

1409: \bibitem[Freyer, Hensler \& Yorke(2006)]{fhy06} Freyer, T., Hensler, G., \& Yorke, H.~W.~2006, \apj, 638, 262

1410: \bibitem[Garcia-Segura \& Maclow(1995)]{gm95} Garcia-Segura, G.,

1411: \& MacLow, M.-M. 1995, \apj, 455, 160

1412: \bibitem[Garcia-Segura et al.(1996b)]{glm96b} Garcia-Segura, G.,

1413: Langer, N., \& MacLow, M.-M. 1996, \aap, 316, 133

1414: \bibitem[Garcia-Segura et al.(1996a)]{glm96a} Garcia-Segura, G.,

1415: MacLow, M.-M., \& Langer, N.~ 1996, \aap, 305, 229

1416: \bibitem[Ghavamian, Hughes, \& Williams(2005)]{ghw05} Ghavamian, P.,

1417: Hughes, J.~P., \& Williams, T.~B.~2005, \apj, 635, 365

1418: \bibitem[Hammer et al.(2006)]{hfs06} Hammer, F., Flores, H., Schaerer,

1419: D., Dessauges-Zavadsky, M., Le Floc'h, E., Puech, M.~2006, A\&A, 454,

1420: 103

1421: \bibitem[Hughes(1987)]{h87} Hughes, J. P. 1987, \apj, 314, 103

1422: \bibitem[Koo \& Heiles(1995)]{kh95} Koo, B.-C., \& Heiles, C. 1995,

1423: \apj, 442, 679

1424: \bibitem[Koo \& McKee(1992a)]{km92a} Koo, B.-C., \& McKee, C. F. 1992a, \apj,

1425: 388, 93

1426: \bibitem[Koo \& McKee(1992b)]{km92b} Koo, B.-C., \& McKee, C. F. 1992b,

1427: \apj, 388, 103

1428: \bibitem[Kudritzki et al.(1989)]{kppa89} Kudritzki, R. P., Pauldrach,

1429: A., Puls, J., \& Abbott, D. C.~1989, \aap, 219, 205

1430: \bibitem[Langer et al.(1994)]{lhl94} Langer, N., Hamann, W.-R.,

1431: Lennon, M., Najarro, F., Pauldrach, A. W. A., Puls, J. 1994, \aap,

1432: 372, 9 290, 819

1433: \bibitem[Langer(1989)]{l89} Langer, N. 1989, \aap, 220, 135

1434: \bibitem[Langer(1994)]{l94} Langer, N. 1994, in ``Circumstellar Media

1435: in the Late Stages of Stellar Evolution'', Proceedings of the 34th

1436: Herstmonceaux conference, eds. R.E.S. Clegg, I. R. Stevens \&

1437: W. P. S. Meikle, Cambridge: CUP, 1

1438: \bibitem[Levenson et al.(1997)]{l97} Levenson, N. A. et al., 1997,

1439: \apj, 484, 304

1440: \bibitem[Luo \& McCray(1991)]{lm91} Luo, D., \& McCray, R.~1991, \apj, 379, 659

1441: \bibitem[Maeder \& Desjacques(2001)]{md01} Maeder, A., \& Desjacques,

1442: V.~2001, \aap, 372, 9

1443: \bibitem[McKee(2004)]{m04}McKee, C. F. 2004, in ``Star Formation in

1444: the Interstellar Medium: In Honor of David Hollenbach, Chris McKee and

1445: Frank Shu'', ASP Conference Proceedings, Vol. 323, eds. D.~Johnstone,

1446: F.C. Adams, D.N.C. Lin, D.A. Neufeld, and E.C. Ostriker, (San

1447: Francisco: Astronomical Society of the Pacific), 21

1448: \bibitem[Mdzinarishvili \& Chargeishvili(2005)]{mc05} Mdzinarishvili

1449: T. G., \& Chargeishvili K. B. 2005, A\&A, 431, L1

1450: \bibitem[Mellema(1995)]{m95} Mellema, G.~1995, \mnras, 277, 173

1451: \bibitem[Mellema \& Frank(1995)]{mf95} Mellema, G., \& Frank, A. 1995, \mnras, 

1452: 273, 401

1453: \bibitem[Nieuwenhuijzen \& de Jager(1990)]{nj90} Nieuwenhuijzen, H., \& de

1454: Jager, C.,~1990, \aap, 231, 134

1455: \bibitem[Ostriker \& McKee(1988)]{om88} Ostriker, J.~P., \& McKee,

1456: C.~F.~1988, RvMP, 60, 10

1457: \bibitem[Park et al.(2002)]{pr02} Park, S., Roming, P.~W.~A., et al.~2002, \apj, 564, L39

1458: \bibitem[Raymond, Cox \& Smith(1976)]{rcs76} Raymond, J.~C., Cox,

1459: D.~P., \& Smith, B.~W.~1976, \apj, 204, 290

1460: \bibitem[Rozyczka et al.(1993)]{rtfb93} Rozyczka, M., Tenorio-Tagle,

1461: G., Franco, J., \& Bodenheimer, P. 1993, \mnras, 261, 674

1462: \bibitem[Ryan-Webber et al.(2004)]{rw04} Ryan-Weber, E. V., Meurer, G. R., Freeman, K. C., 

1463: et al.~2004, \aj, 127, 1431

1464: \bibitem[Sugerman et al.(2005b)]{sck05b}Sugerman, B.~E.~K., Crotts,

1465: A.~P.~S., Kunkel, W.~E., Heathcote, S.~R., \& Lawrence, S.~S.~2005b,

1466: \apjs, 159, 60

1467: \bibitem[Sugerman et al.(2005a)]{sck05a}Sugerman, B.~E.~K., Crotts,

1468: A.~P.~S., Kunkel, W.~E., Heathcote, S.~R., \& Lawrence, S.~S.~2005a,

1469: \apj, 627, 888

1470: \bibitem[Sutherland \& Dopita(1993)]{sd93} Sutherland, R., S., \&

1471: Dopita, M. A. 1993, ApJS, 88, 253

1472: \bibitem[Tenorio-Tagle et al.(1991)]{trfb91} Tenorio-Tagle, G.,

1473: Rozyczka, M., Franco, J., \& Bodenheimer, P. 1991, \mnras, 251, 318

1474: \bibitem[Tenorio-Tagle et al.(1990)]{tbf90} Tenorio-Tagle, G.,

1475: Bodenheimer, P., Franco, J., \& Rozyczka, M. 1990, \mnras, 244, 563

1476: \bibitem[Vink et al(1997)]{vkb97} Vink, J., Kaastra, J.~S., \&

1477: Bleeker, J.~A.~M.~1997, \aap, 328, 628

1478: \bibitem[Vishniac \& Ryu(1989)]{vr89} Vishniac, E. T., \& Ryu,

1479: D.~1989, \apj, 337, 917

1480: \bibitem[Vishniac(1983)]{v83} Vishniac, E. T.~1983, \apj, 274, 152

1481: \bibitem[Weaver et al.(1977)]{wmc77} Weaver, R., McCray, R., Castor,

1482: J., Shapiro, P., \& Moore, R. 1977, \apj, 218, 377

1483: \bibitem[Wrigge et al.(2005)]{wcmw05} Wrigge, M., Chu, Y-H., Magnier,

1484: E.~A., \& Wendker, H.~J.~2005,ApJ, 633, 248

1485: \end{thebibliography}

1486: 

1487: \clearpage

1488: 

1489: \begin{figure}

1490: \includegraphics[angle=90,scale=0.75]{f1.eps}

1491: \caption{Evolution of the velocity of the wind from the star with time}

1492: \label{fig:bubvel}

1493: \end{figure}

1494: 

1495: 

1496: \begin{figure}

1497: \includegraphics[angle=90,scale=0.75]{f2.eps}

1498: \caption{Evolution of the mass-loss rate of the wind from the star with time}

1499: \label{fig:bubmdot}

1500: \end{figure}

1501: 

1502: \begin{figure}

1503: \includegraphics{f3.eps}

1504: \caption{Density and Pressure Profiles at various timesteps in the

1505: evolution of the wind-blown nebula, during the Main Sequence

1506: Stage. The solid line shows the density, the dashed lines the

1507: pressure. Density scale is on LHS in g cm$^{-3}$, pressure on RHS in

1508: cgs units. Time is given in years at the top of each panel. The two

1509: numbers in the top right denote the velocity in km s$^{-1}$, and the

1510: mass-loss rate in $\msun {\rm yr}^{-1}$. The X-axis scale is in

1511: parsecs. Note that the grid is expanding with time. }

1512: \label{fig:ms1d}

1513: \end{figure}

1514: 

1515: \begin{figure}

1516: \includegraphics{f4.eps}

1517: \caption{Density and Pressure Profiles at various timesteps in the

1518: evolution of the wind-blown nebula, during the Red Supergiant

1519: Stage. Other details are as in Figure \ref{fig:ms1d}}

1520: \label{fig:rsg1d}

1521: \end{figure}

1522: 

1523: \begin{figure}

1524: \includegraphics[scale=0.85]{f5.eps}

1525: \caption{Density and Pressure Profiles at various timesteps in the

1526: evolution of the wind-blown nebula, during the Wolf-Rayet Stage. Other

1527: details are as in Figure \ref{fig:ms1d}}

1528: \label{fig:wr1d}

1529: \end{figure}

1530: 

1531: \begin{figure}

1532: \includegraphics[angle=90,scale=0.75]{f6.eps}

1533: \caption{Density and Pressure Profiles within the Circumstellar Bubble

1534: at the end of the star's life, just prior to its death in a SN

1535: explosion. Other details are as in Figure \ref{fig:ms1d}}

1536: \label{fig:bub1dfinal}

1537: \end{figure}

1538: 

1539: 

1540: \begin{figure}

1541: \includegraphics[angle=90,scale=0.75]{f7.eps}

1542: \caption{The X-ray (thermal bremstrahhlung) luminosity during the evolution of the circumstellar wind-blown bubble.}

1543: \label{fig:xlumbub}

1544: \end{figure}

1545: 

1546: \begin{figure}

1547: \includegraphics{f8.eps}

1548: \caption{Density and Pressure Profiles at various timesteps in the

1549: evolution of the supernova remnant during the bubble. }

1550: \label{fig:snbuba1d}

1551: \end{figure}

1552: 

1553: 

1554: \begin{figure}

1555: \includegraphics{f9.eps}

1556: \caption{Density and Pressure Profiles at various timesteps in the

1557: evolution of the supernova remnant during the bubble.}

1558: \label{fig:snbubb1d}

1559: \end{figure}

1560: 

1561: \begin{figure}

1562: \includegraphics{f10.eps}

1563: \caption{Density and Pressure Profiles at various timesteps in the

1564: evolution of the supernova remnant during the bubble.}

1565: \label{fig:snbubc1d}

1566: \end{figure}

1567: 

1568: \begin{figure}

1569: \includegraphics{f11.eps}

1570: \caption{Density and Pressure Profiles at various timesteps in the

1571: evolution of the supernova remnant during the bubble.}

1572: \label{fig:snbubd1d}

1573: \end{figure}

1574: 

1575: \begin{figure}

1576: \includegraphics[angle=90,scale=0.8]{f12.eps}

1577: \caption{The velocity of the forward shock with time over the

1578: expansion of the remnant. }

1579: \label{fig:snvel}

1580: \end{figure}

1581: 

1582: 

1583: \begin{figure}

1584: \includegraphics[angle=90,scale=0.75]{f13.eps}

1585: \caption{The X-ray (thermal bremsstrahlung) luminosity during the SN evolution}

1586: \label{fig:xlumsn}

1587: \end{figure}

1588: 

1589: 

1590: \begin{figure}

1591: \includegraphics[scale=0.9]{f14.eps}

1592: \caption{The X-ray surface brightness profiles during the SN evolution}

1593: \label{fig:xsbsn}

1594: \end{figure}

1595: 

1596: 

1597: \begin{figure}

1598: \includegraphics[scale=0.9, clip=true]{f15.eps}

1599: \caption{Density plots of the two-dimensional evolution of the bubble

1600: during the main-sequence stage. The numbers in the top right corner

1601: give the wind parameters at each stage - velocity (in km s$^{-1}$),

1602: mass loss rate (in $\msun {\rm}^{-1}$), and time (in years). }

1603: \label{fig:bub2d_ms}

1604: \end{figure}

1605: 

1606: \begin{figure}[t]

1607: \includegraphics[angle=90,scale=0.85]{f16.eps}

1608: \caption{The evolution of the wind-blown bubble in the red-supergiant

1609: (first two panels) and Wolf-Rayet (panels three to six) phases. The

1610: legend is as for Figure \ref{fig:bub2d_ms} }

1611: \label{fig:bub2d_pms}

1612: \end{figure}

1613: 

1614: \begin{figure}

1615: \includegraphics[scale=0.95]{f17.eps}

1616: \caption{4 frames emphasizing the bubble evolution at various phases,

1617: especially to show the development of instabilities in those

1618: phases. (a) Top left: The growth of Rayleigh-Taylor fingers in the Red

1619: Supergiant wind (b) Top right: Growth of R-T instabilities in the

1620: Wolf-Rayet wind (c) The interaction of the W-R and RSG winds, leading

1621: to the fragmentation of the RSG wind (d) Velocity vectors plotted over

1622: density contours at the end of the star's life. }

1623: \label{fig:bub2dzoom}

1624: \end{figure}

1625: 

1626: \clearpage

1627: 

1628: \begin{figure}[t]

1629: \includegraphics*[angle=90, scale=0.85, origin=cr]{f18.eps}

1630: \caption{These panels show the time-evolution of the supernova shock

1631: wave within the wind-blown bubble. The panels show the pressure rather

1632: than density as displayed in the previous figures. The time in years

1633: is given at the top right of each plot. }

1634: \label{fig:snbub2dpre}

1635: \end{figure}

1636: 

1637: \clearpage

1638: 

1639: \begin{figure}[t]

1640: \includegraphics*[angle=90, scale=0.85, origin=cr]{f19.eps}

1641: \caption{Here we show the density evolution at the various timesteps

1642: during the expansion of the SN shock wave. The panels clearly

1643: demonstrate how the spherical shock wave becomes a wrinkled and

1644: corrugated structure as it impacts the dense shell. }

1645: \label{fig:snbub2dden}

1646: \end{figure}

1647: 

1648: \end{document}

1649: \end

1650: 

1651: 

1652: