1: \documentstyle[11pt,newpasp,twoside]{article}

2: \markboth{Author \& Co-author}{APS Conf. Ser. Style}

3: \pagestyle{myheadings}

4: \nofiles

5: 

6: % Some definitions I use in these instructions.

7: 

8: \def\emphasize#1{{\sl#1\/}}

9: \def\arg#1{{\it#1\/}}

10: \let\prog=\arg

11: 

12: \def\edcomment#1{\iffalse\marginpar{\raggedright\sl#1\/}\else\relax\fi}

13: \marginparwidth 1.25in

14: \marginparsep .125in

15: \marginparpush .25in

16: \reversemarginpar

17: 

18: \begin{document}

19: \title{Magnetically Driven Warping and Precession of Accretion Disks:

20: Implications for ``Exotic'' Stellar Variabilities}

21:  \author{Dong Lai}

22: \affil{Department of Astronomy, Cornell University, Ithaca, NY 14853, USA}

23: 

24: \begin{abstract}

25: The inner region of the accretion disk around a magnetized star is

26: subjected to magnetic torques that induce warping and precession of the

27: disk. These torques arise from interactions between the stellar field and

28: the induced electric currents in the disk.

29: These novel magnetic effects give rise to some

30: ``exotic'' stellar variabilities, and may play an important role

31: in explaining a number of puzzling behaviors related to disk accretion 

32: onto magnetic stars, such as mHz QPOs in X-ray pulsars,

33: long-term periodicities of X-ray binaries (including precession of jets),

34: low-Frequency (10-50~Hz) QPO's in low-mass X-ray binaries, and 

35: photometric variabilities of T Tauri stars.

36: \end{abstract}

37: 

38: 

39: \section{Introduction}

40: 

41: Disk accretion onto a magnetic star occurs in a variety of astrophysical 

42: contexts, including accreting neutron stars, white dwarfs and 

43: pre-main-sequence stars (e.g., Frank et al.~1992). The basic

44: picture of disk--magnetosphere interaction is well-known: at large radii the

45: disk is unaffected by the stellar magnetic field; a somewhat sudden

46: transition occurs when the stellar field disrupts the disk at the

47: magnetospheric boundary, and channels the plasma onto the polar caps of the

48: star. The magnetosphere boundary is located where 

49: the magnetic and plasma stresses balance, 

50: \begin{equation}

51: r_m=\eta\,\mu^{4/7}(GM\dot M^2)^{-1/7},

52: \end{equation}

53: where $M$ and $\mu$ are the mass and magnetic moment of the central star,

54: $\dot M$ is the mass accretion rate, $\eta$ is a dimensionless constant

55: of order unity.

56: 

57: Because of its intrinsic importance for a wide range of astrophysical 

58: systems, a large number of theoretical papers have been written on the 

59: subject of the interaction between accretion disks and magnetized stars

60: (see references in Lai 1999 and in Shirakawa \& Lai 2002a,b), 

61: and numerical study of this problem

62: is still in its infancy. Outstanding issues remain, including the 

63: efficiency of field dissipation in/outside the disk, whether the disk excludes

64: the stellar field by diamagnetic currents or the field can penetrate a large 

65: fraction of the disk, whether the threaded field remains closed (connecting the

66: star and the disk) or becomes open by differential shearing, and whether/how

67: magnetically driven wind is launched from the disk or the

68: magnetosphere/corotation boundary. 

69:  

70: Many previous theoretical papers have, for simplicity, adopted

71: the idealied geometry in which the magnetic axis, the spin axis and 

72: the disk angular momentum are aligned. However, in Lai (1999), it was 

73: shown that under quite general conditions, the stellar magnetic

74: field can induce warping in the inner disk and make the disk 

75: precess around the spin axis (see \S 2). Such magnetically driven warping 

76: and precession open up new possibilities for the dynamical behaviors of 

77: disk accretion onto magnetic stars, and may explain some of the observed

78: variabilities in different stars (including compact objects).  

79: 

80: \def\be{\begin{equation}}

81: \def\ee{\end{equation}}

82: \def\ba{\begin{eqnarray}}

83: \def\ea{\end{eqnarray}}

84: \def\go{\mathrel{\raise.3ex\hbox{$>$}\mkern-14mu

85:              \lower0.6ex\hbox{$\sim$}}}

86: \def\lo{\mathrel{\raise.3ex\hbox{$<$}\mkern-14mu

87:              \lower0.6ex\hbox{$\sim$}}}

88: \def\cJ{{\cal J}}

89: \def\cQ{{\cal Q}}

90: \def\cO{{\cal O}}

91: 

92: 

93: \section{Magnetically Driven Warping/Precession}

94: 

95: Lai (1999) shows that the inner region of the accretion disk onto a rotating

96: magnetized central star is subjected to magnetic torques which induce 

97: warping and precession of the disk. The origin of these torques lies in

98: These magnetic torques result from the interactions between

99: the accretion disk and the stellar magnetic field.

100: Depending on how the disk responds to the stellar field,

101: two different kinds of torque arise:

102: (i) If the vertical stellar magnetic field $B_z$ penetrates the disk,

103: it gets twisted by the disk rotation

104: to produce an azimuthal field $\Delta B_\phi=\mp\zeta B_z$ that has different

105: signs above and below the disk ($\zeta$ is the azimuthal pitch of the field

106: line and depends on the dissipation in the disk), and a radial surface current

107: $K_r$ results. The interaction between $K_r$ and the stellar $B_\phi$ gives

108: rise to a vertical force. While the mean force (averaging over the azimuthal

109: direction) is zero, the uneven distribution of the force induces a net

110: {\it warping torque} which tends to misalign the angular momentum of the disk

111: with the stellar spin axis.

112: (ii) If the disk does not allow the vertical stellar field

113: (e.g., the rapidly varying component of $B_z$ due to stellar rotation)

114: to penetrate, an azimuthal screening current $K_\phi$ will be induced on the

115: disk. This $K_\phi$ interacts with the radial magnetic field $B_r$

116: and produces a vertical force. The resulting {\it precessional torque}

117: tends to drive the disk into retrograde precession around the stellar spin

118: axis.

119: 

120: In general, both the magnetic warping torque and the precessional torque are

121: present. For small disk tilt angle $\beta$ (the angle between the disk normal

122: and the spin axis), the precession angular frequency and warping rate

123: at radius $r$ are given by 

124: \ba

125: &&\Omega_p (r)=\frac{\mu^2}{\pi^2 r^7\Omega(r)\Sigma(r) D(r)}F(\theta),

126: \label{eqn:Omega_p}\\

127: &&\Gamma_w (r)=\frac{\zeta\mu^2}{4\pi r^7\Omega(r)\Sigma(r)}\cos^2\theta,

128: \label{eqn:Gamma_w}

129: \ea

130: where $\mu$ is the stellar magnetic dipole moment, $\theta$ is

131: the angle between the magnetic dipole axis and the spin axis,

132: $\Omega(r)$ is the orbital angular frequency, and $\Sigma(r)$ is the surface

133: density of the disk. [Note that the stellar spin

134: frequency $\Omega_s$ does not appear in eqs.~(2) \& (3) 

135: since the variation of the field geometry due to the spin has been averaged

136: out; this is justified because $\Omega_s\gg |\Omega_p|,~|\Gamma_w|$.]

137: The dimensionless function $D(r)$ is given by

138: \be

139: D(r)={\rm max}~\left(\sqrt{r^2/r^2_{\rm in}-1}, \sqrt{2H(r)/r_{\rm in}}\right)

140: \label{eqn:D(r)},

141: \ee

142: where $H(r)$ is the half-thickness and $r_{\rm in}$ is the inner radius of the

143: disk. The function $F(\theta)$ depends on the dielectric property of the

144: disk. We can write

145: \be

146: F(\theta)=2f\cos^2\theta-\sin^2\theta,

147: \ee

148: so that $F(\theta)=-\sin^2\theta$ if only the spin-variable vertical field is

149: screened out by the disk ($f=0$), and $F(\theta)=3\cos^2\theta-1$ if all

150: vertical field is screened out ($f=1$). In reality, $f$ lies between 0 and 1.

151: For concreteness, we shall set $F(\theta)=-\sin^2\theta$ in the following.

152: 

153: We also note the effect of {\it magnetically driven resonances}. 

154: For a general magnetic field--disk geometry,

155: the vertical magnetic force on a disk element varies

156: with the stellar rotation period. This gives rise to a number of 

157: {\it vertical resonances} in the disk. Similarly. there exist 

158: {\it epicyclic resonances} due to the time-dependent radial magnetic

159: force. Although the force expressions are model-dependent,

160: the existence of the resonances appears to be inevitable.

161: These magnetically driven resonances are somewhat similar to the corotation

162: resonance and Lindblad resonances in gravitational systems. 

163: The resonances may act as an extra source (in addition to the non-resonant

164: precessional and warping torques discussed above) for generating bending 

165: waves and spiral waves in the disk. Near the resonances, fluid elements 

166: undergo large out-of-plane and radial excursions, which may lead to

167: thickening of the disk. This may be analogous to the 

168: Lorentz resonances (which occur when charged particles

169: move around a rotating magnetic field) in the jovian ring

170: (e.g., Schaffer \& Burns 1992). However, because of the

171: fluid nature of the disk, the resonances may not lead to sharp edges 

172: in the disk.

173: 

174: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

175: \section{Dynamics of Warped Disks, Effects of Viscosity, Global Warping Modes

176: and Nonlinear Evolution}

177: 

178: %%%%%%%%%%%%%

179: \subsection{Effects of Viscosity}

180: 

181: Since the magnetic torque drives the disk tilt,

182: while the viscosity reduces the tilt, one can derive the 

183: criterion for the warping instability. Roughly speaking, the disk warp

184: can grow if the timescale associated with the warping torque is shorter than

185: the viscous time $r^2/\nu_2$ [where

186: $\nu_2$ is the viscosity (measuring the $r$-$z$ stress)

187: associated with reducing disk tilt].

188: Since the warping torque is a steep function

189: of $r$, the warping instability occurs only inside

190: a critical radius $r_w$. Our analysis (Lai 1999) shows that 

191: local warping torque can overcome viscous damping when 

192: \be

193: \Gamma_w >{2\pi^2}{\nu_2\over r^2}\Longleftrightarrow

194: {\rm Instability}.

195: \label{criterion}\ee

196: Assuming that $\nu_2/\nu_1$ is independent of $r$, the above equation reduces

197: to

198: \be

199: r<r_w=\left({3\,\zeta\cos^2\theta\over 8\pi^2\cJ}{\nu_1\over\nu_2}

200: \right)^{2/7}\left({\mu^4\over GM\dot M^2}\right)^{1/7},

201: \ee

202: where we have used

203: $\Sigma=({\dot M/3\pi\nu_1}){\cal J}$, 

204: and ${\cal J}$ is a dimensionless function of $r$ which approaches unity in the

205: region far from the inner edge of the disk.

206: Thus $r_w$ is typically a few times the canonical Alf\'ven radius

207: (the magnetosphere boundary). Therefore, 

208: as the disk approaches the magnetosphere,

209: its normal vector $\hat l$ will tend to be tilted with respect to the stellar

210: spin even if at large radii it is aligned with the spin axis. 

211: 

212: Another aspect of the viscous effect is what we call 

213: ``{\it Magnetic Bardeen-Petterson Effect}''.

214: Because of the magnetic precessional torque,

215: the tilted disk will be driven into differential precession (with the

216: precession rate dependent on $r$). By analogy with the

217: well-known Bardeen-Petterson effect (i.e., the inner region of an accretion 

218: disk undergoing Lense-Thirring precession around a rotating black hole

219: tends to align itself with the equatorial plane of the black hole; 

220: see Bardeen \& Petterson 1975),

221: we expect that the magnetically driven

222: precession also tends to damp the tilt of the inner disk through the action of

223: viscosity. Setting $\Omega_p$ equal to $\nu_2/r^2$, we obtain 

224: the magnetic Bardeen-Petterson radius:

225: \be

226: r_{\rm MBP}=\left({3\sin^2\theta\over\pi\cJ D}{\nu_1\over\nu_2}\right)^{2/7}

227: \left({\mu^4\over GM\dot M^2}\right)^{1/7}.

228: \ee

229: Inside $r_{\rm MBP}$, the combined effect of viscosity and precession

230: tends to align the disk normal with the spin axis. 

231: We see that typically $r_{\rm MBP}$ is of the same order as $r_w$ (the

232: warping radius) and $r_m$

233: (the magnetosphere radius). Thus the precessional torque 

234: has an opposite effect on the

235: disk tilt as the warping torque. However, 

236: because of the broad warp-alignment transition expected for the magnetic

237: Bardeen-Petterson effect and the long timescale involved, we expect that the 

238: precession-induced alignment will be overwhelmed by the warping instability.

239: 

240: %%%%%%%%%%%%%%%%%%%%%%5

241: \subsection{Global Warping Modes and Nonlinear Evolution}

242: 

243: Since the precession rate $\Omega_p(r)$ depends strongly on $r$,

244: coupling between different rings is needed to produce a global coherent

245: precession. The coupling can be achieved either by viscous stress or

246: through bending waves (e.g., Papaloizou\& Pringle~1983; Papaloizou \&

247: Terquem~1995). In the viscosity dominated regime

248: (i.e., the dimensionless viscosity parameter $\alpha$ greater than $H/r$),

249: the dynamics of the warps can be studied using the formalism of

250: Papaloizou \& Pringle (1983) (see also Ogilvie~1999;

251: Ogilvie \& Dubus~2001). We model the disk as a collection of

252: rings which interact with each other via viscous stresses.

253: Each ring at radius $r$ has the unit normal vector ${\bf\hat l}(r,t)$.

254: In the Cartesian coordinates, with the $z$-axis along the neutron star spin,

255: we write

256: ${\hat{\bf l}}=(\sin\beta\cos\gamma,\sin\beta\sin\gamma,\cos\beta)$,

257: with $\beta(r,t)$ the tilt angle and $\gamma(r,t)$ the twist angle.

258: For $\beta\ll 1$, the dynamical warp equation for ${\hat{\bf l}}$

259: (Lai 1999; see Papaloizou \& Pringle 1983; Pringle 1992) reduces to

260: an equation for $W(r,t)\equiv \beta(r,t)e^{i\gamma(r,t)}$:

261: \ba

262: &&\frac{\partial W}{\partial t}-

263: \left[\frac{3\nu_2}{4r}\left(1+\frac{2r\cJ'}{3\cJ}\right)

264: +\frac{3\nu_1}{2r}(\cJ^{-1}-1)\right]

265: \frac{\partial W}{\partial r}\nonumber\\

266: &&\qquad\qquad =\frac{1}{2}\nu_{2}\frac{\partial^2 W}{\partial r^2}

267: +i\Omega_pW+\Gamma_wW,\label{eqn:evolution}

268: \ea

269: where $\cJ'=d\cJ/dr$ (we assume that the ratio of $\nu_2$ to $\nu_1$ is

270: constant). In deriving the above equation, we have used the relations for the

271: radial velocity and surface density: $v_r=-3\nu_1\cJ^{-1}/2r$ and

272: $\Sigma={\dot M}\cJ/3\pi\nu_1$.

273: The values and functional forms of $\nu_1$, $\nu_2$, $\Omega_p$,

274: $\Gamma_w$ and the dimensionless function $\cJ(r)$ depend on disk models

275: (see Shirakawa \& Lai 2002a,b for details).

276: 

277: Shirakawa \& Lai (2002) carried out a global analysis

278: of warping/precession modes in a viscous accretion disk,

279: and show that under a wide range of conditions, the magnetic warping torque 

280: can overcome viscous damping and make the mode grow. The warping/precession

281: modes are concentrated near the inner edge of the disk (at the

282: magnetosphere-disk boundary), and can give rise to variabilities or

283: quasi-periodic oscillations (QPOs) in the X-ray/UV/optical fluxes from X-ray

284: pulsars (see \S 4). Pfeiffer \& Lai (2003) studied the nonlinear evolution

285: of the warping-precession modes, and found that the mode tends to saturate

286: at a large amplitude (its value depends on the parameters of the system).

287: The implications of such nonlinear behavior remain to be understood.

288: 

289: 

290: \section{Applications}

291: 

292: The magnetically driven warping instability and precession 

293: help explaining a number of observational puzzles related to 

294: stellar variabilities (for more details, see Lai 1999; 

295: Shirakawa \& Lai 2002ab; Pfeiffer \& Lai 2003).

296: 

297: \smallskip

298: {\bf (i) Milli-Hertz QPO's in Accreting X-ray Pulsars}:

299: Quasi-Periodic Oscillations (QPO's)  with frequencies $1-100$~mHz have been

300: detected in at least 11 accreting X-ray pulsars.

301: These mHz QPOs are often interpreted in terms of the

302: beat frequency model (BFM), in which the observed QPO frequency represents the

303: beat between the Keplerian frequency $\nu_K$ at the inner disk radius

304: and the NS spin frequency $\nu_s$, 

305: or in terms of the Keplerian frequency model (KFM), in which

306: the QPOs arise from the modulation of the X-rays by some inhomogeneities in the

307: inner disk at the Keplerian frequency. However for several sources, more than

308: one QPOs have been detected and the difference in the QPO frequencies is not

309: equal to the spin frequency. Thus KFM and/or the BFM cannot be the whole story.

310: Also note that in both the KFM and the BFM, it is always postulated that the

311: inner disk contain some blobs or inhomogeneities, whose physical origin is

312: unclear.

313: In Shirakawa \& Lai (2002) we suggest a ``Magnetic Disk Precession Model''

314: for the mHz variabilities and QPOs of accreting X-ray pulsars.

315: The magnetically driven precession of the warped inner disk (outside

316: but close to the magnetosphere boundary) can modulate

317: X-ray/UV/optical flux in several ways. We identify $\nu_{\rm QPO}$

318: with the global precession frequency driven by the magnetic

319: torques. Our calculations show that under a wide range of conditions,

320: the warping/precession mode is concentrated near the disk inner edge, and

321: the global mode frequency is equal to $A=0.3-0.85$ (depending on details of

322: the disk structure) times the magnetically driven precession frequency at

323: $r_{\rm in}=r_m$. An examination of the observed properties

324: of mHz QPOs in several systems (such as 4U 1626-67) suggests that

325: some hitherto unexplained QPOs are likely to be results of magnetically

326: driven disk warping/precession (see Chakrabarty et al.~2001).

327: 

328: \smallskip

329: {\bf (ii) Spin evolution of accreting X-ray pulsars}: Recent long-term,

330: continuous monitoring of X-ray pulsars with the BATSE instrument on the Compton

331: Gamma Ray Observatory (CGRO) has revealed a number of puzzling behaviors of the

332: spins of these objects (see Bildsten et al.~1997 and references therein). 

333: Several well-measured disk-fed systems (e.g., Cen X-3, GX 1+4 and 4U 1626-67) 

334: display sudden transitions between episodes of steady

335: spin-up and spin-down, with the absolute values of spin torques approximately

336: equal (to within a factor of a few). The transition timescale

337: ranges from days to years. It is likely that the  magnetically driven disk

338: warping may be a crucial ingredient in determining the spin behaviors of

339: accreting X-ray pulsars. With the magnetic warping torque, the perpendicular

340: state is an ``attractor''. The observed sign switching of $\dot\omega_s$ (spin

341: derivative) in several X-ray pulsars may be associated with the ``wandering''

342: of the inner disk around this ``attractor''. Rough estimate based on the

343: magnetic torque indicates that the switching timescale (which depends

344: on the stellar field strength, the disk parameters and geometry)

345: ranges from days to years, in agreement with observations. 

346: 

347: \smallskip

348: {\bf (iii) Quasi-periodic oscillations in low-mass X-ray binaries}: 

349: Rapid variability in low-mass X-ray binaries, containing

350: weakly magnetized ($B\sim 10^8$~G) neutron stars,

351: has been studied since the discovery of the so-called 

352: horizontal-branch oscillations (HBOs) (see van der Klis 1998).

353: The HBOs are quasi-periodic oscillations

354: (QPOs) (with $Q$-value $\nu/\Delta\nu$ of order a few)

355: which manifest as broad Lorentzian peaks in the X-ray power spectra 

356: with centroid frequencies in the range of 15--60~Hz which are positively

357: correlated with the inferred mass accretion rate. Stella and Vietri (1998)

358: suggested that HBOs (and other low-frequency QPOs)

359: are associated with Lense-Thirring precession of the

360: inner accretion disk around the rotating NS. 

361: For this interpretation to be viable, the inner 

362: disk must be tilted with respect to the stellar spin axis. Since the

363: Bardeen-Petterson effect tends to keep the inner region of the disk 

364: (typically within 100--1000 Schwarzschild radii) co-planar with the star 

365: and radiation-driven warping is only effective at large disk radii

366: (Pringle 1996), another mechanism to drive warping in the inner disk 

367: is needed. The magnetic warping torque provides

368: a natural source for inducing disk tilt. Moreover, 

369: the magnetically driven (retrograde) precession 

370: rate is not negligible compared to the Lense-Thirring precession rate,

371: and will contribute to the total precession (Shirakawa \& Lai 2002a).

372: 

373: Other possible applications include:

374: 

375: {\bf (iv) Long-term (super-orbital) variabilities in X-ray binaries

376: (including precession of jets):} The well-known examples include

377: Her X-1 (35 days), LMC X-4 (30.4 days) and SS433 (164 days). It has always

378: been thought that these super-orbital periods are caused by 

379: precession of accretion disks, perhaps driven by the binary companion. 

380: However, the tidal torque from the companion is relevant only if the disk is

381: warped. Without any extra driver for the disk warp, the disk would be

382: flat. Magnetic field may play a role here.

383: 

384: {\bf (v) Photometric period variations of T Tauri stars:} 

385: T Tauri stars have magnetic fields of order 1~kG. Being magnetic,

386: they are variable. Most of the variabilities can be explained by rotating

387: cold spots or hot spots on the stellar surface. However, some of the

388: variabilities in classical T Tauri stars are not easy to understand

389: in this picture. For example, AA Tauri shows photometric variability

390: (by 1 mag) in different bands on timescales of 8.5~days, but there is

391: no clear color variation (see Bouvier et al.~1999). 

392: This and some other features can be naturally

393: explained by a warped inner disk which causes occultation of the photosphere

394: (see Carpenter et al.~2001 for possibly other examples). 

395: 

396: 

397: \smallskip

398: I thank Akiko Shirakawa and Harald Pfeiffer for their important 

399: contributions. This work has been supported in part 

400: by NSF AST 9986740 and NASA 

401: NAG 5-8484, and by the Alfred P. Sloan foundation.

402: 

403: \begin{thebibliography}{}

404: 

405: \bibitem[]{} 

406: Bardeen, J.~M., \& Petterson, J.~A. 1975, ApJ, 195, L65

407: 

408: \bibitem[]{} 

409: \vskip -0.2cm

410: Bildsten, L., et al.~1997, ApJS, 113, 367

411: 

412: \bibitem[]{} 

413: \vskip -0.2cm

414: Bouvier, J. et al.~1999, A\&A, 349, 619

415: %Magnetospheric accretion onto the T Tauri star AA Tauri. I. Constraints from

416: %multisite spectrophotometric monitoring

417: 

418: \bibitem[]{}

419: \vskip -0.2cm

420: Chakrabarty, D. et al.~2001, ApJ, 562, 985

421: 

422: \bibitem[]{}

423: \vskip -0.2cm

424: Carpenter, J.M. et al.~2001, AJ, 121, 3160

425: 

426: \bibitem[]{}

427: \vskip -0.2cm

428: Frank, J., et al.

429: %King, A., \& Raine, D. 

430: 1992, Accretion Power in Astrophysics

431: (Cambridge Univ. Press)

432: 

433: \bibitem[]{}

434: \vskip -0.2cm

435: Lai, D. 1999, ApJ, 524, 1030

436: 

437: \bibitem[]{}

438: \vskip -0.2cm

439: Ogilvie, G.~I.~1999, MNRAS, 304, 557

440: 

441: \bibitem[]{}

442: \vskip -0.2cm

443: Ogilvie, G.~I., \& Dubus, G. 2001, MNRAS, 320, 485

444: 

445: \bibitem[]{}

446: \vskip -0.2cm

447: Papaloizou, J.~C., \& Pringle, J.~E. 1983, MNRAS, 202, 1181

448: %disk eqn

449: 

450: \bibitem[]{}

451: \vskip -0.2cm

452: Papaloizou, J.~C., \& Terquem, C. 1995, MNRAS, 274, 987

453: 

454: \bibitem[]{}

455: \vskip -0.2cm

456: Pfeiffer, H., \& Lai, D. 2002, ApJ, in preparation

457: 

458: \bibitem[]{} 

459: \vskip -0.2cm

460: Pringle, J.~E. 1996, MNRAS, 281, 857

461: 

462: \bibitem[]{} 

463: \vskip -0.2cm

464: Schaffer, L., \& Burns, J.~A. 1992, ICARUS, 96, 65

465: 

466: \bibitem[]{}

467: \vskip -0.2cm

468: Shirakawa, A., \& Lai, D.~2002a, ApJ, 564, 361

469: 

470: \bibitem[]{}

471: \vskip -0.2cm

472: Shirakawa, A., \& Lai, D.~2002b, ApJ, 565, 1134

473: 

474: \bibitem[]{} 

475: \vskip -0.2cm

476: Stella, L., \& Vietri, M. 1998, ApJ, 492, L59

477: 

478: 

479: 

480: \end{thebibliography}

481: 

482: \end{document}

483: