1: \documentclass[preprint2]{aastex}

2: \usepackage{psfig}

3: %\usepackage{amsmath}

4: 

5: \singlespace

6: 

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

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

9: \newcommand{\eeqn}[1]{\label{#1}\end{equation}}

10: \newcommand{\dt}[1]{\frac{\partial  #1}{\partial t}}

11: \newcommand{\eq}[1]{(\ref{#1})}

12: \newcommand{\vO}{{\bf \Omega}}

13: \newcommand{\na}{ {\bf \nabla} }

14: \newcommand{\vv}{{\bf v}}

15: \newcommand{\vu}{{\bf u}}

16: \newcommand{\vn}{{\bf n}}

17: \newcommand{\ez}{{\bf e}_z}

18: \newcommand{\ddz}[1]{\frac{\partial^2  #1}{\partial z^2}}

19: \renewcommand{\char}{characteristic}

20: \newcommand{\vzero}{{\bf 0}}

21: \newcommand{\bu}{\overline{u}}

22: \newcommand{\lp}{ \left(}

23: \newcommand{\rp}{ \right)}

24: \newcommand{\tu}{\tilde{u}}

25: \newcommand{\cth}{ \cos\theta }

26: \newcommand{\beqa}{\begin{eqnarray*}}

27: \newcommand{\beqan}{\begin{eqnarray}}

28: \newcommand{\eeqa}{\end{eqnarray*}}

29: \newcommand{\eeqan}[1]{\label{#1}\end{eqnarray}}

30: \newcommand{\sth}{ \sin\theta }

31: \newcommand{\lac}{ \left\{}

32: \newcommand{\rac}{ \right\}}

33: \newcommand{\lc}{ \left[}

34: \newcommand{\rc}{ \right]}

35: \newcommand{\dzeta}[1]{\frac{\partial  #1}{\partial \zeta}}

36: \newcommand{\intvol}{ \int_{(V)}\! }

37: 

38: \bibliographystyle{astron}

39: 

40: \begin{document}

41: \title{Ekman layers and the damping of inertial r-modes in a spherical

42: shell: application to neutron stars}

43: 

44: \author{Michel Rieutord}

45: \affil{Observatoire Midi-Pyr\'en\'ees and Institut Universitaire de France}

46: \affil{14 avenue Edouard Belin, 31400 Toulouse}

47: \email{rieutord@obs-mip.fr}

48: \date{\today}

49: 

50: 

51: %\maketitle

52: 

53: \begin{abstract}

54: Recently, eigenmodes of rotating fluids, namely inertial modes, have

55: received much attention in relation to their destabilization when

56: coupled to gravitational radiation within neutron stars. However, these

57: modes have been known for a long time by fluid dynamicists. We give

58: a short account of their history and review our present understanding

59: of their properties. 

60: Considering the case of a spherical container, we then give the exact

61: solution of the boundary (Ekman) layer flow associated with inertial

62: r-modes and show that previous estimations all underestimated the

63: dissipation by these layers. We also show that the presence of an inner

64: core has little influence on this dissipation. As a conclusion, we

65: compute the window of instability in the temperature/rotation plane for

66: a crusted neutron star when it is modeled by an incompressible fluid.

67: \end{abstract}

68: 

69: \keywords{Hydrodynamics; rotating stars; neutron stars}

70: 

71: \section{Introduction}

72: 

73: Recently much work has been devoted to the study  of the rotational

74: instability of neutron stars resulting from a coupling between

75: gravitational radiation and the so-called ``r-modes" of a rotating star

76: \cite{anderss98,FM98,LOM98,KS99}.

77: Such an instability may indeed play a key role in the distribution of

78: rotation periods of neutron stars as well as it may be an important source

79: of gravitational radiation.

80: 

81: In this paper, we shall first clarify a point of history concerning

82: ``r-modes" which are in fact a special class of inertial modes; we shall

83: then review their singular properties which have been clarified only

84: very recently in \cite{RV97}, \cite{RGV99} and \cite{RGV00b}. The last

85: section will present the analytical derivation of the damping rate of

86: inertial r-modes in a neutron star with a crust and/or a core through

87: the boundary layer analysis within the framework of newtonian theory.

88: We conclude on the stability of crusted neutron stars when modeled by

89: an incompressible viscous fluid in a rotating sphere.

90: 

91: \section{A short point of history}

92: 

93: The very first work on rotating fluids oscillations which are presently

94: known as {\em inertial modes} dates back to Thomson\footnote{later Lord

95: Kelvin} (1880)\nocite{T1880} who analysed the case of a fluid contained in

96: a cylinder. However, another impetus to the study of these oscillations

97: was given soon after by Poincar\'e's (1885) \nocite{Poinc1885} work on

98: the stability of rotating self-gravitating masses, a work applied to

99: MacLaurin spheroids by \cite{Bryan1888}\footnote{ but

100: see the recent rederivation by \cite{LI99}.} and later continued

101: by \cite{Cartan22} who christened the equation of inertial modes

102: as ``Poincar\'e equation". In these studies, however, the effect of

103: rotation is combined to the one of gravity through (for an incompressible

104: fluid) surface gravity waves. In fact, except for the work of Thomson,

105: investigations on the oscillations specific of rotating fluids seem

106: to have started with the work of \cite{Bj33} where they are

107: called ``elastoid-inertial oscillations" since conservation of angular

108: momentum makes axis-centered rings of fluid behave elastically; but see

109: \cite{Fultz59} or \cite{Ald67} for an account on this part of history. In

110: the sixties, much work has been devoted to these oscillations, mainly by

111: Greenspan who introduced the terminology of ``inertial oscillations". The

112: presently used denomination ``inertial modes" has been ``officially"

113: given by Greenspan's book \citep{Green69}.

114: 

115: However, inertial modes are somewhat too general for applications in some

116: specific domains like atmospheric sciences. In this field indeed motions

117: are essentially two dimensional and inertial modes may be simplified

118: into the well-known Rossby (or planetary) waves.

119: 

120: The introduction of r-modes by \cite{PaPr78} was quite unfortunate

121: since they associated eigenmodes of rotating fluids with a very special

122: class of inertial modes, namely purely toroidal inertial modes.  This lead

123: following authors to introduce weird names such as ``hybrid modes" or

124: ``generalized r-modes" \citep{LF99} for describing the general class of

125: inertial modes. We therefore encourage authors to use,

126: as fluid dynamicists, inertial modes unless they discuss the very specific

127: r-modes.

128: 

129: \section{The present theory of inertial modes}

130: 

131: Inertial modes are a class of modes of oscillation of rotating fluids

132: which owe their existence to the Coriolis force. This force of inertia

133: has indeed a restoring action on perturbations of rotating fluids since

134: it insures the global  conservation of angular momentum. These modes

135: have many properties similar to those of gravity modes of stably

136: stratified fluids \citep{RN99}.

137: 

138: The dynamics of inertial modes may be appreciated when all other effects

139: are suppressed: no compressibility, no magnetic fields, no gravity,

140: etc... only an incompressible inviscid rotating (like a solid body)

141: fluid. In this case, perturbations of velocity $\delta \vv$ and

142: pressure $\delta P$ obey

143: 

144: \beq \dt{\delta\vv} + 2\vO\times\delta\vv = -\na\delta P,

145: \qquad \na\cdot\delta\vv =0 \eeqn{dimeq}

146: where $\vO$ is the angular velocity of the fluid. Concentrating on

147: time-periodic oscillations and choosing $(2\Omega)^{-1}$ as the time

148: scale, \eq{dimeq} can be written

149: 

150: \beq i\omega\vu+\ez\times\vu = -\na p, \qquad \na\cdot\vu=0\eeqn{eqmo}

151: with non-dimensional variables; $\omega$ is the non-dimensional (real)

152: frequency. When the velocity $\vu$ is eliminated in favor of the pressure

153: perturbation $p$, one is left with

154: 

155: \beq \Delta p -\frac{1}{\omega^2}\ddz{p} = 0 \eeqn{poincare}

156: which is known as Poincar\'e equation since \cite{Cartan22}. This

157: equation is remarkable in the fact that it is {\em hyperbolic spatially} since

158: $|\omega| \leq 1$ \citep{Green69}. As the solution of \eq{poincare} must

159: meet boundary conditions, namely $\vu\cdot\vn=0$, we see that inertial

160: modes are solutions of an ill-posed boundary value problem\footnote{Such

161: boundary conditions eliminate any distorsion of the surface due to the

162: fluid motion; for a free surface, these distorsion are surface gravity

163: waves (see Rieutord 1997\nocite{rieutord97}) but their inclusion (in

164: order to be more realistic) would not modify the ill-posed nature of

165: the problem.}. This property means that, in general, inertial modes are

166: singular; in other words they cannot exist physically if the fluid is

167: strictly inviscid.  These properties are detailed in \cite{RV97} and

168: \cite{RGV99}; to make a long story short, one may summarize the

169: situation as follows.

170: 

171: \begin{figure}[t]

172: \centerline{\psfig{figure=./dissec243.ps,width=8cm}}

173: \caption[]{The kinetic energy distribution in a meridional plane of

174: an inertial mode in a spherical shell associated with an

175: equatorial attractor. A co-existing polar attractor is also slightly

176: excited. The mode is axisymmetric with equatorial symmetry. Stress-free

177: boundary conditions have been used on both shells; this solution

178: was computed with an Ekman number of 2 10$^{-9}$ and required 1300

179: spherical harmonics and 450 radial grid points (Gauss-Lobatto). The

180: ratio of the inner radius to the outer radius is $\eta=0.35$.

181: $\omega=0.2429$ and the damping rate $\tau=-6.26\times10^{-4}$ are given

182: in dimensionless units as equation~\eq{eqmo}.}

183: 

184: \label{in_attr} \end{figure}

185: 

186: Let us first recall that in hyperbolic systems, energy propagates along

187: the \char s of the equation. For the Poincar\'e problem, these are

188: straight lines in a meridional plane. A way to approach the solutions of

189: this difficult problem is to examine the propagation of \char s as they

190: reflect on the boundaries. They define trajectories which depend

191: strongly on the container. Let us therefore concentrate on the case of

192: a spherical shell as a container; this configuration is relevant for

193: neutron stars with a central core due to some phase transition of the

194: nuclear matter \cite[see][]{Haen96}. In this case, it may be shown that \char\

195: trajectories generically converge towards attractors which are periodic

196: orbits. It may be shown \citep{RGV99} that in this case, the associated

197: solutions are singular, namely the velocity field is not

198: square-integrable. However, inertial r-modes are still solutions of

199: the problem since they meet the boundary conditions ($u_r=0$); in fact

200: they are the only regular (square-integrable) solutions of the

201: Poincar\'e problem in a spherical shell. In a more mathematical way, we

202: may say that the spectrum of eigenvalues of the Poincar\'e problem in a

203: spherical shell is empty except for the inertial r-modes. In this

204: sense, these modes are quite exceptional. This situation occurs because

205: there exists no system of coordinate in which the dependent variables of

206: the Poincar\'e equation can be separated. This is a consequence of the

207: conflict between the symmetry of the Coriolis force (cylindrical) and

208: the geometry of the boundaries. Thus, when this constraint is relaxed,

209: like in the case of a cylindrical container, regular solutions exist

210: and a dense spectrum of eigenvalues appears in the allowed frequency

211: range, namely $[0,2\Omega]$. In the case the container is a full sphere,

212: attractors also disappear and eigenmodes exist; they are also related to a

213: dense spectrum of eigenfrequencies. In this case, Poincar\'e equation

214: is exactly solvable \citep{Green69}.

215: 

216: However, real fluids have viscosity ($\nu$) and equation \eq{eqmo}

217: should be transformed into

218: 

219: \beq \lambda\vu +\ez\times\vu = -\na p+E\Delta\vu, \qquad

220: \na\cdot\vu=0\eeqn{eqvis}

221: where $\lambda$ is the complex eigenvalue and $E=\nu/2\Omega R^2$ is the

222: Ekman number ($R$ is the outer radius of the shell).

223: 

224: Using no-slip ($\vu=\vzero$) or stress-free boundary conditions,

225: \eq{eqvis} yields a well-posed problem. Yet, the singularities of the

226: associated inviscid solutions show up through the existence of shear

227: layers. As shown by fig.~\ref{in_attr}, the shape of inertial modes is deeply

228: influenced by the underlying singularity of the inviscid solution. We

229: have shown \citep{RGV99} that these shear layers are in fact nested

230: layers with different scales for their inner part scales as

231: $E^{1/3}$ and their outer part seems to scale with $E^{1/4}$.

232: Because of these internal shear layers, these modes are strongly damped.

233: 

234: We therefore see that according to whether a neutron star has a central

235: core or not, the damping of inertial modes will be extremely different.

236: If there is a central core,  the only regular modes are the inertial

237: r-modes which will be by far the least damped; if there isn't any

238: core then a dense spectrum exists \citep{Green69,LF99} but inertial

239: r-modes  remain the most unstable because of their simple structure.

240: 

241: 

242: \section{Damping of toroidal inertial modes in a sphere or a shell}

243: 

244: We shall now give the expression of the viscous damping of

245: inertial r-modes when one of the boundary is solid therefore when the

246: dissipation is due to Ekman boundary layers; we shall thus complete the

247: works of \cite{BU00} and \cite{AJKS00} by giving the rigorous estimate

248: of the damping rate; the method which we use here has been outlined in

249: \cite{Green69}.

250: 

251: The damping rate is given by:

252: 

253: \beq \gamma = \Re e(\lambda)= -E\frac{\int (\na : \vu)^2 dV}{\int \vu^2 dV} \eeq

254: where $(\na : \vu)^2$ stands for the squared rate-of-strain tensor (see

255: below). The velocity field of r-modes is

256: 

257: \[ \bu_\theta = A r^m (\sin\theta)^{m-1}\sin(m\phi+\omega_m t) \]

258: \[ \bu_\varphi = A r^m (\sin\theta)^{m-1}\cos\theta\cos(m\phi+\omega_m t) \]

259: 

260: The kinetic energy integral may be evaluated explicitly

261: 

262: \beq \int \vu^2 dV = \pi A^2 \lp

263: 1-\eta^{2m+3}\rp\frac{2^{m+1}(m+1)!}{m(2m+3)!!} \eeqn{ke}

264: where $\eta$ is the ratio of the radius of the inner boundary to the one

265: of the outer boundary.

266: 

267: The dissipation integral need more work if one of the boundary is

268: no-slip. In this case, dissipation is essentially coming from the Ekman layers

269: and thus we need to derive the flow in these layers. The method has been

270: given by Greenspan from whom we know that the boundary layer correction

271: $\tu$, is related to the interior solution $\bu$ by

272: 

273: \beq \tu_\theta+i\tu_\varphi = -(\bu_\theta+i\bu_\varphi)_{r=r_b}

274: e^{-\zeta\sqrt{i\cth\pm i\omega}} \eeq

275: where $\zeta$ is the radial scaled variable $(r-r_b)/\sqrt{E}$ with

276: $r_b$ as the radius of the boundary (1 or $\eta$). The complete solution is then $\vu=\tu+\bu$; setting $\beta=\omega t + m\varphi$  we have

277: 

278: \beqa \bu_\theta+i\bu_\varphi = \frac{A r^m (\sth)^{m-1}}{2i}\hspace*{2cm} \\

279: \qquad\times\lac (1-\cth)e^{i\beta} - (1+\cth)e^{-i\beta}\rac \eeqa

280: from which it follows that

281: 

282: \beqa \tu_\theta+i\tu_\varphi = \frac{A r_b^m (\sth)^{m-1}}{2i}

283: \lac (1+\cth)e^{-i\beta-\zeta\sqrt{i\cth-i\omega}}\right.\\ -

284: \left. (1-\cth)e^{i\beta-\zeta\sqrt{i\cth+i\omega}}\rac \eeqa

285: 

286: Now we need the expression of the square of the rate-of-strain tensor

287: $s_{ij}=\partial_iv_j+\partial_jv_i$ in spherical coordinates, viz

288: 

289: \[ (\na : \vu)^2= s_{rr}^2 +

290: s_{\theta \theta }^2 +s_{\phi\phi}^2 + 2(s_{r\theta }^2 +

291: s_{r\phi}^2 +s_{\theta \phi}^2) \]

292: 

293: Since the radial derivatives dominate, this expression reduces to the

294: contribution of the tangential stresses. Using the scaled coordinate,

295: $\zeta=|r-r_b|/\sqrt{E}$, we have

296: 

297: \[ (\na : \vu)^2= \frac{2}{E}\lac \lp\dzeta{u_\theta}\rp^2 +

298: \lp\dzeta{u_\varphi}\rp^2\rac_{r=r_b} \]

299: 

300: We now set $p=\cth -\omega$ and $q=\cth+\omega$. We thus get

301: 

302: \beqa (\na : \vu)^2= \frac{A^2 r_b^{2m}(\sth)^{2m-2}}{2E}\lac

303: (1+\cth)^2|p|e^{-\zeta\sqrt{2|p|}}\right. \\

304: +(1-\cth)^2|q|e^{-\zeta\sqrt{2|q|}}  \\

305: \left. -2\sin^2\theta\sqrt{pq}\;\Re e\lp

306: e^{2i\beta-\zeta(\sqrt{iq}+\sqrt{-ip})}\rp\rac \eeqa

307: Integrating over the $\varphi$-variable yields

308: 

309: \beqan \int_0^{2\pi} (\na : \vu)^2 d\varphi = \frac{\pi A^2

310: r_b^{2m}(\sth)^{2m-2}}{E}\hspace*{20mm}\nonumber \\

311: \times\lac

312: (1+\cth)^2|p|e^{-\zeta\sqrt{2|p|}}+(1-\cth)^2|q|e^{-\zeta\sqrt{2|q|}}\rac

313: \eeqan{eeqq}

314: We now integrate over the radial variable:

315: 

316: \beqa \int_\eta^1\int_0^{2\pi} (\na : \vu)^2 d\varphi\; r^2dr = \hspace*{20mm}\\

317: \frac{\pi A^2(\sth)^{2m-2}}{E} \int_\eta^1 r^{2m+2} f(\zeta) dr \eeqa

318: with

319: $f(\zeta)=(1+\cth)^2|p|e^{-\zeta\sqrt{2|p|}}+

320: (1-\cth)^2|q|e^{-\zeta\sqrt{2|q|}}$; since $r=\eta+\sqrt{E}\zeta$ or

321: $r=1-\sqrt{E} \zeta$ according to which side of the integral is chosen,

322: it turns out that

323: 

324: \beqa & &\hspace*{-15mm}\int_\eta^1\!\int_0^{2\pi} (\na : \vu)^2d\varphi\; r^2dr \\

325: &=& K\int_0^\infty \lc

326: (1+\cth)^2|p|e^{-\zeta\sqrt{2|p|}}\right. \\

327: &&+\left.(1-\cth)^2|q|e^{-\zeta\sqrt{2|q|}}\rc d\zeta \\

328: &=& \frac{K}{\sqrt{2}} \lc (1+\cth)^2\sqrt{|\cth-\omega|} \right. \\

329: &&+\left. (1-\cth)^2\sqrt{|\cth+\omega|} \rc

330: \eeqa

331: with $K=\pi A^2(\sth)^{2m-2}P(\eta)/\sqrt{E}$, where $P(\eta)$ is a

332: function depending on the boundary conditions (see table~\ref{bcp}). Finally

333: integrating over $\theta$, we find

334: 

335: \beq \int (\na : \vu)^2 dV = \frac{2\pi A^2P(\eta)}{\sqrt{2E}}

336: {\cal I}_m \eeqn{diss}

337: with

338: 

339: \beq  {\cal I}_m = \int_0^\pi

340: (1+\cth)^2\sqrt{|\cth-\omega|}\sin^{2m-1}\theta d\theta \eeq

341: 

342: Finally grouping \eq{ke} and \eq{diss}, we find the damping rate

343: 

344: \beq \gamma = -\frac{m(2m+3)!!}{2^{m+3/2}(m+1)!}Q(\eta)

345: {\cal I}_m\sqrt{E}

346: \eeqn{lf}

347: where $Q(\eta)=P(\eta)/\lp1-\eta^{2m+3}\rp$.

348: 

349: \begin{table}[t]

350: \centerline{

351: \begin{tabular}{ccc}

352:  Outer B.C. & no-slip & stress-free\\

353:  \hline \\

354:  Inner B.C. & & \\

355: no-slip & $1+\eta^{2m+2}$ & $\eta^{2m+2}$ \\

356: stress-free & 1 & No Ekman layer\\

357: \hline

358: \end{tabular}

359: }

360: \caption[]{Expression of the $P(\eta)$ function according to boundary

361: conditions.}

362: \label{bcp}

363: \end{table}

364: 

365: For the cases $m=1$ and $m=2$ we evaluated the expression of ${\cal

366: I}_m$, viz

367: 

368: \[ {\cal I}_1 = \frac{\sqrt{2}}{35}\lp 3^{5/2}+19\rp \]

369: \[{\cal I}_2 =

370: 4\lp\frac{2}{3}\rp^{11/2}\frac{3401+2176\sqrt{2}}{5\times7\times9\times11}

371: \]

372: Other values are computed numerically and given in table~\ref{integ}.

373: 

374: The values given by \eq{lf} may be compared to other derivations, in

375: particular that of \cite{Green69} for $m=1$ who finds $\gamma/\sqrt{E} =

376: -2.62/\sqrt{2} = -1.8526$ ! For $m=2$, a direct numerical calculation,

377: similar to that of \cite{RV97}, gives $-2.482\sqrt{E}$ at $E=10^{-8}$

378: which is in good agreement with the analytical formula.

379: 

380: \begin{table}[t]

381: \centerline{

382: \begin{tabular}{ccccc} \hline

383: \\

384:  & $m=1$ & $m=2$ & $m=3$ & $m=4$ \\

385: \\

386: ${\cal I}_m$ & 1.3976 & 0.80411 & 0.58075 & 0.46155 \\

387: ${\gamma\over\sqrt{E}}$ & -1.8526 & -2.4876 & -3.0318 & -3.5339\\

388: \\

389: \hline

390: \end{tabular}

391: }

392: \caption[]{Values of the first integrals ${\cal I}_m$ and the

393: corresponding values of the scaled damping rates; this latter value

394: equals that of \cite{Green69} when multiplied by $\sqrt{2}$ because of

395: our choice of the time scale.}

396: \label{integ}

397: \end{table}

398: 

399: \section{Application to neutron stars and conclusions}

400: 

401: Let us now apply these results to the case of rapidly rotating neutron

402: stars. We take the viscosity from \cite{BU00}, $\nu=1.8\,f/T^2_8$ m$^2$/s

403: where $f$ is a dimensionless parameter taking into account the different

404: transport mechanisms in the fluid (superfluid phases for instance) and

405: $T_8$ is the temperature in 10$^8$K unit.

406: Using a radius of 12.53~km and an angular frequency of

407: $2\pi\times$1kHz, we find an Ekman number $\sim 10^{-12}$ which is indeed very

408: small and thus boundary layer theory applies.

409: 

410: \begin{figure}[t]

411: \centerline{\psfig{figure=./critical.ps,width=8cm}}

412: \caption[]{Curves of critical angular velocity, normalized by

413: $\Omega_K=\frac{2}{3}\sqrt{\pi G\overline{\rho}}$, for different models.

414: The solid line shows the result of the present work, the dashed-dotted

415: one is that of \cite{AJKS00} and the dashed one is for \cite{BU00}. The

416: dotted line is the critical curve for a non-crusted star. No core has been

417: included ($\eta=0$).}

418: \label{crit} \end{figure}

419: 

420: We may now estimate the charateristic time scale for the damping of the

421: $m=2$-mode. We find

422: 

423: \beq T_{d} = 26.7 {\rm s}\; \frac{T_8}{\sqrt{f}}\lp\frac{R}{10\;\rm

424: km}\rp \lp\frac{1\;\rm kHz}{\nu_s}\rp^{1/2} \eeq

425: which is a somewhat smaller value than the previous estimate of

426: \cite{BU00} and \cite{AJKS00} who find a characteristic time of 100~s

427: and 200~s respectively. Our disagreement with these authors comes

428: from their approximate evaluation of the boundary layer dissipation and

429: from the resulting functional dependence with respect to mass and

430: density. Let us first evaluate the damping rate according to

431: \cite{LL7189}; it turns out that

432: 

433: \beq 2\gamma = -\lp\frac{\omega E}{2}\rp^{1/2}\frac{\lp \int_{4\pi}

434: \vu^2\sth d\theta d\varphi\rp(r=1)}{\intvol \vu^2 dV} \eeq

435: where we used our non-dimensional units. Since the radial dependence of

436: the modes is in $r^m$ and $\omega=1/(m+1)$, we easily find that

437: 

438: \[ \gamma = -\frac{2m+3}{2\sqrt{2m+2}} \sqrt{E} \]

439: When this expression is applied to the $m=2$-mode, we find that

440: $\gamma=-1.429\sqrt{E}$ which is a factor $1.74$ weaker than the correct

441: result. 

442: 

443: If we use, as previous authors, a step function for describing the

444: density difference between that of the crust and the mean density, we

445: find that the damping rate reads

446: 

447: \beq \gamma_{Ek} = -2.4876\sqrt{E}\frac{\rho_b}{\overline{\rho}}\,

448: 2\Omega = -0.0346 \frac{\rho_b}{\overline{\rho}}

449: \frac{\sqrt{f\Omega_\star}}{T_8} {\rm s}^{-1}\eeqn{damp_rate}

450: where $\rho_b$ is the density of the fluid just below the crust and

451: $\Omega_\star=\Omega/\sqrt{\pi G\overline{\rho}}$.

452: 

453: 

454: Our calculation therefore shows that the window

455: of instability in the $\Omega, T$ plane is smaller than previously

456: estimated for crusted neutron stars.

457: 

458: Considering a 1.4~M${_\odot}$ neutron star with a radius of 12.53~km as a

459: test case, the growth rate of the mode due to gravitational radiation is

460: $\gamma_{gw} = 0.658 {\rm s}^{-1} \Omega_\star^6$

461: \cite[we use the expression given in][]{LOM98}; although, it is not

462: relevant for an incompressible fluid, we take into account the damping

463: rate due to bulk viscosity in order to ease comparison with previous

464: work; from \cite{LMO99}, we find $\gamma_{bulk} = -2.2\, 10^{-12}\;{\rm

465: s}^{-1}\; T_9^6\Omega_\star^2$. From \eq{damp_rate}, we have

466: $\gamma_{Ek}= -1.53\, 10^{-3}\;{\rm s}^{-1}\; \Omega_\star^{1/2}/T_9$ where

467: we took $\rho_b=1.5\,10^{17}$kg/m$^3$; solving the equation

468: 

469: \[ \gamma_{gw}+\gamma_{Ek}+\gamma_{Bulk} = 0 \]

470: for different values of the temperature yields the curves displayed in

471: figure~\ref{crit}.

472: 

473: As expected, we see that the window of instability narrows compared to

474: \cite{AJKS00}: for a given temperature, the critical angular velocity

475: raises by $\sim$10\% typically.

476: 

477: 

478: Another interesting conclusion of this work is that the presence of a

479: solid inner core does not change the damping rates very much unless its

480: radius is close to unity. The reason for that is to be found in the

481: shape of the inertial r-modes whose amplitudes are concentrated near the

482: outer boundary. Therefore, the rotating instability of rapidly rotating

483: stars is quite insensitive to the presence of a solid core and more

484: generally to any phase transition which does not occur close to the

485: surface.

486: 

487: \acknowledgments

488: 

489: I am very grateful to S.~Bonazzola and E.~Gourgoulhon for drawing my

490: attention on these questions and for helpful discussions. I am also

491: very grateful to Ian Jones for his note about the density profile used

492: in models of neutron stars in previous work.  Part of the calculations

493: have been carried out on the Nec SX5 of IDRIS at Orsay and on the

494: CalMip machine of CICT in Toulouse which are gratefully acknowledged.

495: 

496: 

497: \bibliography{../../biblio/bibnew}

498: 

499: 

500: 

501: \end{document}

502: