1: \documentclass[preprint]{aastex}

2: 

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

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

5: \newcommand{\bea}{\begin{eqnarray}}

6: \newcommand{\eea}{\end{eqnarray}}

7: \newcommand{\nn}{\nonumber}

8: \newcommand{\um}{^{\mu}}

9: \newcommand{\un}{^{\nu}}

10: \newcommand{\dm}{_{\mu}}

11: \newcommand{\dn}{_{\nu}}

12: \newcommand{\dmn}{_{\mu\nu}}

13: \newcommand{\umn}{^{\mu\nu}}

14: \newcommand{\cP}{{\cal P}}

15: \newcommand{\cF}{{\cal F}}

16: \newcommand{\cA}{{\cal A}}

17: \newcommand{\cJ}{{\cal J}}

18: \def\pa{\partial}

19: \def\bB{\mbox{\boldmath$B$}}

20: \def\bE{\mbox{\boldmath$E$}}

21: \def\bJ{\mbox{\boldmath$J$}}

22: \def\bA{\mbox{\boldmath$A$}}

23: \def\bP{\mbox{\boldmath$P$}}

24: \def\bQ{\mbox{\boldmath$Q$}}

25: \def\bV{\mbox{\boldmath$V$}}

26: \def\bbe{\mbox{\boldmath$\beta$}}

27: \def\rg{{r_g\over r}}

28: \def\rs{r_{\ast}}

29: %

30: \shorttitle{General Relativistic Particle Acceleration in Pulsar Polar Cap}

31: \shortauthors{Sakai \& Shibata}

32: %

33: \begin{document}

34: 

35: \title{General Relativistic Electromagnetism and Particle Acceleration 

36: in Pulsar Polar Cap}

37: \author{Nobuyuki Sakai}

38: \affil{Faculty of Education, Yamagata University, Yamagata 990-8560, Japan}

39: \email{sakai@ke-sci.kj.yamagata-u.ac.jp}

40: \and

41: \author{Shinpei Shibata}

42: \affil{Faculty of Science, Yamagata University, Yamagata 990-8560, Japan}

43: \email{shibata@sci.kj.yamagata-u.ac.jp}

44: 

45: \begin{abstract}

46: We reconstruct a 3+1 formalism of general relativistic 

47: electromagnetism, and derive the equations of motion of charged 

48: particles in the pulsar magnetosphere, taking account of the inclination 

49: between the rotation axis and the magnetic axis. Apart from the 

50: previous works where space charge is evaluated by assuming the flow 

51: velocity being the light speed, we analyze particle 

52: motion in the polar cap, finding that gravity changes significantly its 

53: dynamics and the condition for acceleration.

54: \end{abstract}

55: 

56: \keywords{acceleration of particles, gravitation, magnetic fields, pulsars}

57: 

58: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

59: \section{Introduction}

60: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

61: 

62: The origin of radio emission from pulsars remains a great mystery. One of 

63: the likely scenarios is that particles are accelerated along open magnetic 

64: field lines and emit $\gamma$-rays which subsequently convert into 

65: electron-positron pairs under a strong magnetic field. A combination 

66: of the primary beam and pair plasma provide the radio emission 

67: mechanism \cite{Mel}.

68: The process of particle acceleration has been intensively studied in flat 

69: spacetime \cite{FAS,AS,Mes81,Shi91,Shi95,Shi97}.

70: The field-aligned electric field is driven by deviation of the space 

71: charge from Goldreich-Julian charge density \cite{GJ}, which is determined by 

72: the magnetic field geometry. Therefore, general relativistic (GR) effects

73: on the field geometry are crucial for the formation of the 

74: field-aligned electric field and particle acceleration.

75: 

76: Muslimov \& Tsygan (1992) initiated the GR analysis of electromagnetic 

77: fields around a pulsar; they solved Maxwell equations on the 

78: assumption that the particles move with the light speed. Shibata 

79: showed, however, that their assumption of constancy of particle speed was not 

80: always true \cite{Shi91,Shi97}: 

81: the particle motion and space charge show oscillatory behavior even 

82: on field lines curving toward the rotation axis if the assumed current 

83: density is sub-critical.

84: Mestel (1996) extended Shibata's equations of motion to include GR 

85: effects, though his analysis is not complete.

86: Recently Harding and Muslimov (2001) extended Muslimov-Tsygan's 

87: analysis to include the pair-creation dynamics.

88: 

89: In this paper we derive the basic equations more rigorously and generally:

90: we depart Muslimov-Tsygan (and Harding-Muslimov) model by solving equations 

91: of motion for the particles, and correct Mestel's equation and extend it 

92: to include the inclination angle between the magnetic axis and the 

93: rotation angle. Next we solve the electric field together with the 

94: particle motion for the region near the magnetic pole just above the surface. 

95: It turns out that the effect of gravity on particle dynamics cannot be 

96: ignored, and hence the condition for acceleration is also modified. 

97: As pointed by Arons (1997), local deficit of space charge takes place 

98: on most of the field lines. In particular, particle dynamics on the 

99: ``away'' field lines from the rotation axis is changed qualitatively by 

100: gravity if the current density is less than the critical value. It is 

101: also true that oscillatory solution exists even with GR effect. We 

102: find that some oscillatory solutions disappear above a certain height. 

103: 

104: This paper is organized as follows. In section 2, we reconstruct a 3+1 

105: formalism of GR electromagnetism. In section 3, we derive 

106: equations of motion of charged particles around a magnetized 

107: rotating star. In section 4, solving the equations of motion, we analyze 

108: particle acceleration in the polar cap. Conclusions and discussions are given in section 5.

109: We use the unit $c=1$.

110: 

111: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

112: \section{3+1 Electromagnetic Equations: General Formalism}

113: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

114: 

115: A 3+1 split of Maxwell equations were made by Landau \& Lifshitz 

116: (1975) and Thorne \& Macdonald (1982). Landau \& Lifshitz derived 3+1 Maxwell equations 

117: in a straightforward manner, which we essentially follow. 

118: They introduced a metric

119: \footnote{In this paper Greek letters run from 0 to 3, and Latin letters run from 1 to 3, 

120: contrary to the convention of Landau \& Lifshitz (1975).},

121: \beq\label{LLds}

122: ds^2=g\dmn dx\um dx\un

123: =-h(dt-g_i dx^i)^2+\gamma_{ij}dx^idx^j,

124: \eeq\beq

125: h\equiv-g_{00},~~ g_i\equiv{g_{0i}/h},~~ \gamma^{ij}\equiv g^{ij},

126: \eeq

127: which is formally in a 3+1 form.

128: However, it is hard to understand how the entire 4D spacetime ($x\um,g\dmn$) is 

129: decomposed into 3D spatial hypersurfaces ($x^i,q_{ij}$) because each hypersurface 

130: does not correspond to $t=$constant.

131: 

132: On the other hand, Thorne \& Macdonald adopted the metric used in the ADM formalism:

133: \beq\label{ds}

134: ds^2=g\dmn dx\um dx\un=-\alpha^2dt^2+q_{ij}(dx^i+\beta^idt)(dx^j+\beta^jdt),

135: \eeq\beq

136: \alpha^2\equiv-{1/g_{00}},~~ \beta_i\equiv g_{0i},~~ 

137: q_{ij}\equiv g_{ij}, 

138: \eeq

139: where $\alpha$ and $\beta$ are called the lapse function and the shift vector, 

140: respectively. With this metric one can understand the concept of the 3+1 

141: decomposition easily, as we describe below. However, they adopted the ``congruence'' 

142: approach and wrote down the equation with 

143: coordinate-free language such as expansion and shear of the fiducial 

144: observers. Although their equations are powerful for some theoretical arguments, 

145: one cannot avoid using coordinates to solve them in most cases.

146: 

147: For the purpose of our study, we combine and modify the two methods: we adopt the 

148: ADM formalism to decompose a spacetime and electromagnetic fields into 

149: a 3+1 form, and write down 3+1 electromagnetic equations in a straightforward manner.

150: 

151: First, with the metric (\ref{ds}), we introduce the fiducial observers with the 4-velocity,

152: \beq\label{u}

153: u\dm=(-\alpha,0,0,0), 

154: \eeq

155: and the spacelike hypersurface $\Sigma$ which is orthogonal to $u\um$ 

156: at each time $t$. Because $\Sigma(t)$ is characterized by a $t=$ constant 

157: hypersurface, $x^i$ and $q_{ij}$ are 

158: the 3-coordinates and the 3-metric of $\Sigma(t)$, respectively. 

159: We define the projection tensor, which projects any 4-vector or tensor 

160: into $\Sigma(t)$, as

161: \beq

162: h\umn\equiv g\umn+u\um u\un.

163: \eeq

164: 

165: Electromagnetic quantities are originally in the 3+1 form, i.e., 

166: 3-vectors and scalars: electric field $\bE$, magnetic field $\bB$, 

167: charge density $\rho$, current density $\bJ$, scalar potential $\phi$, 

168: and vector potential $\bA$. We define those quantities on $\Sigma(t)$, 

169: i.e., those quantities are supposed to be measured by the 

170: observers with the velocity $u\um$.

171: Note that the ``ordinary'' components of a vector $\bP$ are neither the covariant 

172: components $P_i$ nor the contravariant components $P^i$ \cite{Wei}.

173: If the 3-coordinates $x^i$ are orthogonal, i.e., 

174: $q_{ij}=h_{(i)}\delta_{ij}$, the ordinary components $P_{(i)}$ are

175: \beq

176: P_{(i)}=h_{(i)}P^i={P_i/h_{(i)}}.

177: \eeq

178: As long as we adopt the 4-velocity (\ref{u}), which specifies 

179: $\Sigma$, we can convert any 3-vector $\bP$ on $\Sigma$ is converted into 

180: a 4-vector $P\um$ in a natural way,

181: \beq

182: P\um=(0,~P^i).

183: \eeq

184: It is easy to confirm that $P\um$ is the vector on $\Sigma$: $P\um u\dm=0$ and $P\um h\dm^i=P^i$.

185: 

186: Let us reconstruct the electromagnetic field $\cF\umn$, the charge-current 

187: vector ${\cJ\um}$ and the 4-potential $\cA\um$ from the quantities defined 

188: on $\Sigma$ and $u\um$ \cite{TM}:

189: \bea\label{defF}

190: &&\cF\umn=u\um E\un-u\un E\um

191: +\epsilon^{\mu\nu\lambda\sigma}u_{\lambda}B_{\sigma}, \\

192: &&\cJ\um=\rho u\um+J\um, ~~

193: \cA\um=\phi u\um+A\um,

194: \label{defJA}\eea

195: where $\epsilon^{\mu\nu\lambda\sigma}$ is the Levi-Civita tensor with

196: $\epsilon^{0123}=1/\sqrt{-g}$. We can invert these relations as

197: \bea

198: E^i=\cF^{i\nu} u\dn, &~~& 

199: B^i=-\frac12\epsilon^{i\nu\lambda\sigma}u\dn \cF_{\lambda\sigma}, \\

200: \rho=-\cJ\um u\dm, &~~& J^i=h^i\dn\cJ\un, \\

201: \phi=-\cA\um u\dm, &~~& A^i=h^i\dn \cA\un.

202: \eea

203: 

204: The 4D expressions of Maxwell equations and the continuity equation 

205: are

206: \beq\label{4DMax}

207: \cF\umn_{~~;\nu}=4\pi\cJ\um,~~ \cF_{[\mu\nu;\lambda]}=0,~~

208: \cJ\um_{~;\mu}=0.

209: \eeq

210: Substituting (\ref{defF}) and (\ref{defJA}) into (\ref{4DMax}), we 

211: obtain 3+1 Maxwell equations:

212: \bea\label{divB}&&

213: {\rm div}\bB=0,

214: \\\label{rotE}&&

215: {1\over\sqrt{q}}\pa_t(\sqrt{q}B^i)

216: +[{\rm rot}(\alpha\bE+\bbe\times\bB)]^i=0, 

217: \\\label{divE}&&

218: {\rm div}\bE=4\pi\rho,

219: \\\label{rotB}&&

220: -{1\over\sqrt{q}}\pa_t(\sqrt{q}E^i)

221: +[{\rm rot}(\alpha\bB+\bbe\times\bE)]^i=4\pi\alpha J^i,\label{M34}

222: \\\label{CC}&&

223: {1\over\sqrt{q}}\pa_t(\sqrt{q}\rho)+{\rm div}(\alpha\bJ)=0,

224: \eea

225: where cross product, divergence and rotation are defined as

226: \beq

227: [\bQ\times\bP]^i \equiv \epsilon^{ijk} Q_j P_k, ~~

228: {\rm div}\bP \equiv{1\over\sqrt{q}}\pa_i(\sqrt{q}P^i), ~~

229: [{\rm rot}\bP]^i\equiv\epsilon^{ijk}\pa_jP_k,

230: \eeq

231: and $\epsilon^{ijk}$ is the spatial Levi-Civita tensor with 

232: $\epsilon^{123}=1/\sqrt{q}$.

233: 

234: To complete 3+1 electromagnetic equations, we rewrite 

235: the equation of motion of a charged particle. We denotes its mass 

236: by $m$, its charge by $e$, its proper time by $\tau$ and its 4-velocity by 

237: $v\um\equiv dx\um/d\tau$. Then the equation of motion is written as

238: \beq\label{4eom}

239: m\left({dv\um\over d\tau}+\Gamma\um_{\nu\lambda}v\un v^{\lambda}\right)

240: =e\cF\umn v\dn,

241: \eeq

242: where $\Gamma$ is the affine connection.

243: We define a scalar $\gamma$, the 3-velocity vector $V^i$ and its 

244: norm $V$ as

245: \beq\label{defV}

246: \gamma\equiv-v\um u\dm=\alpha v^0,~~

247: V^i\equiv{v\um h^i\dm\over\gamma}={v^i+\beta^iv^0\over\gamma},~~

248: V^2\equiv q_{ij}V^iV^j.

249: \eeq

250: From the normalization $v\dm v\um=-1$ and (\ref{defV}), we find

251: $\gamma={1/\sqrt{1-V^2}}$,

252: which implies that the scalar $\gamma$ is nothing but the Lorentz factor.

253: With these quantities and the 3+1 electromagnetic quantities we rewrite 

254: the the equation of motion (\ref{4eom}) as

255: \beq\label{eom1}

256: {d\over d\tau}\left\{\gamma(-\alpha+\bbe\cdot\bV)\right\}

257: -{\gamma^2\over\alpha}\left(-\pa_0\alpha+{\alpha\over2}V^iV^j\pa_0 q_{ij}

258: +\bV\cdot\pa_0\bbe\right)

259: ={e\gamma\over m\alpha}\bE\cdot\bV,

260: \eeq\beq

261: {d\over d\tau}(\gamma V_i)

262: -{\gamma^2\over\alpha}\left(-\pa_i\alpha+{\alpha\over2}V^jV^k\pa_i q_{jk}

263: +\bV\cdot\pa_i\bbe\right)

264: ={e\gamma\over m\alpha}\left\{\alpha\bE+\bV\times\bB-(\bE\cdot\bV)\bbe\right\}_i. 

265: \label{eom2}\eeq

266: 

267: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

268: \section {Electromagnetic Equations in Pulsar Magnetosphere}

269: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

270: 

271: We assume that the spacetime outside a neutron star is stationary and axisymmetric.

272: Because gravity of charged particles is negligible, the outer spacetime is 

273: described by Kerr metric. We adopt the spherical coordinates

274: $(\hat t,~\hat r,~\hat\theta,~\hat\varphi)$ so that $\hat\theta=0$ accords 

275: with the rotation axis. In the slow-rotation where we take up to the first order

276: of the angular momentum of the star $L$, the spacetime is described by the metric,

277: \beq\label{ds2}

278: \hat\alpha^2=F(\hat r)\equiv1-{\hat r_g\over\hat r},~~

279: \hat\beta^{\hat\varphi}=-\omega(\hat r)\equiv-{2L\over\hat r^3},~~

280: \hat q_{ij}={\rm diag}(F^{-1}(\hat r),~\hat r^2,~\hat r^2\sin^2\hat\theta),

281: \eeq

282: where $r_g$ is the gravitational radius.

283: 

284: We consider the two-step coordinate transformations. First, we 

285: introduce the ``corotating coordinates'' as 

286: \beq

287: \hat\varphi\longrightarrow\hat\varphi-\Omega\hat t,

288: \eeq

289: where $\Omega$ is the angular velocity of the star \cite{LL}. The

290: shift vector is, accordingly, transformed as

291: \beq

292: \hat\beta^{\hat\varphi}\longrightarrow\hat\beta^{\hat\varphi}=\Omega-\omega.

293: \eeq

294: Second, we move on to the ``magnetic coordinates'' $(t,~r,~\theta,~\varphi)$, 

295: where the magnetic dipole axis accords with $\theta=0$.

296: Supposing the magnetic axis is at $\hat\theta=\chi,~\hat\varphi=0$, the 

297: coordinate transformation from the corotating coordinate to the 

298: (corotating) magnetic coordinate is given by

299: \bea

300: &&\hat x=\hat r\sin\hat\theta\cos\hat\varphi,~~

301: \hat y=\hat r\sin\hat\theta\sin\hat\varphi,~~

302: \hat z=\hat r\cos\hat\theta,~~

303: \nn\\

304: &&x=\hat x\cos\chi+\hat z\sin\chi,~~

305: y=\hat y,~~

306: z=-\hat x\sin\chi+\hat z\cos\chi,

307: \\

308: &&x=r\sin\theta\cos\varphi,~~

309: y=r\sin\theta\sin\varphi,~~

310: z=r\cos\theta.

311: \nn\eea

312: The shift vector in (\ref{ds2}) is, accordingly, transformed into

313: \beq\label{ds31}

314: \beta^r=0,~~ \beta^{\theta}=-\sin\chi\sin\varphi(\Omega-\omega),~~

315: \beta^{\varphi}=(\cos\chi-\cot\theta\cos\varphi\sin\chi)(\Omega-\omega),

316: \eeq

317: while the lapse function and the spatial metric remain unchanged:

318: \beq\label{ds32}

319: \alpha^2(r)=F(r),~~ q_{ij}={\rm diag}(F^{-1}(r),~r^2,~r^2\sin\hat\theta).

320: \eeq

321: 

322: In this (and any) axisymmetric spacetime the observers with the velocity (\ref{u}) 

323: is called ``zero-angular-momentum observers'' (ZAMOs) because of 

324: vanishing their angular momentum per unit mass: $u\dm({\pa/\pa\varphi})\um=0$

325: \cite{TM}.

326: 

327: Now we analyze the basic equations (\ref{divB})-(\ref{CC}), (\ref{eom1}) 

328: and (\ref{eom2}) with the metric (\ref{ds31}) and (\ref{ds32}). First,

329: we suppose that the magnetic field distortion due to the external 

330: currents is negligibly small. Then the magnetic field is 

331: governed by (\ref{divB}) and (\ref{rotB}) with $\bJ=0$:

332: \beq\label{Bsol}

333: {\rm div}\bB=0,~~ {\rm rot}(\alpha\bB)=0.

334: \eeq

335: We adopt the dipole-like solution of (\ref{Bsol}) \cite{GO}:

336: \bea\label{Br}

337: B_{(r)}&\equiv&{B^r\over\sqrt{F}}

338: =B_0{f(r)\over f(\rs)}\left({\rs\over r}\right)^3\cos\theta

339: \approx B_0\left[1+\frac34\left(\rg-{r_g\over\rs}\right)\right]

340: \left({\rs\over r}\right)^3\cos\theta,

341: \\

342: B_{(\theta)}&\equiv&r^2B^{\theta}={B_0\over2}{\sqrt{F}\over f(\rs)}

343: \left[-2f(r)+3F^{-1}(r)\right]\left({\rs\over r}\right)^3\sin\theta

344: \nn\\

345: &\approx&B_0\left(1+{r_g\over2r}-{3r_g\over4\rs}\right)

346: \left({\rs\over r}\right)^3\sin\theta, \label{Bth}

347: \\

348: f(r)&\equiv&-3\left({r\over r_g}\right)^3

349: \left[\ln F(r)+\rg\left(1+{r_g\over2r}\right)\right]

350: \approx 1+{3r_g\over 4r},

351: \eea

352: where $r=\rs$ denotes the star surface, and the 

353: approximate expressions are obtained by expanding in powers of $r_g/r$

354: or $r_g/\rs$ \cite{KK}.

355: 

356: Equation (\ref{rotE}) is satisfied if there exists a potential $\Phi$ such as

357: \beq\label{Phi}

358: \alpha\bE+\bbe\times\bB=-\nabla\Phi.

359: \eeq

360: Substituting (\ref{Phi}) into the other of Maxwell equations (\ref{divE}), we obtain

361: \beq\label{divEax}

362: -{\rm div}\left({\nabla\Phi\over\alpha}\right)=4\pi(\rho-\rho_{GJ}),~~

363: 4\pi\rho_{GJ}\equiv-{\rm div}\left({\bbe\over\alpha}\times\bB\right),

364: \eeq

365: where $\rho_{GJ}$ is the Goldreich-Julian density.

366: Each term is written explicitly as

367: \bea

368: {\rm div}\left({\nabla\Phi\over\alpha}\right)&=&

369: {\sqrt{F}\over r^2}{\pa\over\pa r}

370: \left(r^2{\pa\Phi\over\pa r}\right)+{1\over\sqrt{F}r^2}

371: \left[{1\over\sin\theta}{\pa\over\pa\theta}

372:          \left(\sin\theta{\pa\Phi\over\pa\theta}\right)

373:        +{1\over\sin^2\theta}{\pa^2\Phi\over\pa\varphi^2}\right],

374: \\

375: 4\pi\rho_{GJ}&=&-\alpha\bB\cdot{\rm rot}{\bbe\over\alpha^2} \nn\\

376: &=&-{B_{(r)}\over\sqrt{F}\sin\theta}

377: \left\{ {\pa(\sin^2\theta\beta^{\varphi})\over\pa\theta}

378: +{\pa\beta^{\theta}\over\pa\varphi} \right\}

379: +B_{(\theta)}{F\sin\theta\over r}{\pa\over\pa r}

380: \left({r^2\beta^{\varphi}\over F}\right).

381: \eea

382: 

383: Let us consider the flow of charged particles. In the case under 

384: consideration, magnetic field dominates over electric field, where

385: particles are at the lowest Landau level and the inertial drift 

386: motion across the magnetic field is negligible. We may, therefore, 

387: regard that particles go along with magnetic field lines:

388: \beq\label{Vj}

389: \bV=\kappa\bB,~~ \bJ=\rho\bV=\rho\kappa\bB,

390: \eeq

391: where $\kappa=V/B\equiv|\bV|/|\bB|$ is a scalar function. 

392: 

393: Then continuity equation (\ref{CC}) implies \cite{MT,Mes96}

394: \beq

395: 0={\rm div}(\alpha\bJ)={\rm 

396: div}(\alpha\rho\kappa\bB)=\bB\cdot\nabla(\alpha\rho\kappa),

397: \eeq

398: which reads

399: \beq

400: \alpha\rho\kappa={\sqrt{F}J\over B}={\rm constant~on~magnetic~field~

401: lines.}

402: \eeq

403: 

404: Equations (\ref{eom1}) and (\ref{eom2}) give the motion of 

405: charged particles. 

406: If we assume $B_{\varphi}=E_{\varphi}=0$, equation (\ref{Phi}) and 

407: (\ref{Vj}) give $V_{\varphi}=\pa_{\varphi}\Phi=0$. Using the 

408: relation

409: \beq\label{EV}

410: \bE\cdot\bV=-{\nabla\Phi\over\alpha}\cdot\bV

411: =-{1\over\alpha}V^i\pa_i\Phi

412: =-{1\over\alpha}\left({1\over\Gamma}{dx^i\over d\tau}+{\beta^i\over\alpha}

413: \right){\pa\Phi\over\pa x^i}

414: =-{1\over\sqrt{F}\Gamma}{d\Phi\over d\tau},

415: \eeq

416: we rewrite one of the equations of motion (\ref{eom1}) to

417: \beq\label{eom}

418: {d\over d\tau}(\sqrt{F}\Gamma)

419: =\frac em{1\over F}{d\Phi\over d\tau}. 

420: \eeq

421: The other equation (\ref{eom2}) need not be solved because the 

422: spatial trajectory is determined by (\ref{Vj}).

423: 

424: Finally, let us consider boundary conditions. We assume that the star crust 

425: is a perfect conductor, and hence particles on the surface do not suffer 

426: Lorentz force. Equations (\ref{eom2}), (\ref{Phi}), (\ref{Vj}) and 

427: (\ref{EV}) reduces this ideal-MHD condition to $\nabla\Phi=0$.

428: We also assume that there exist closed magnetic lines, where the ideal-MHD 

429: condition holds. 

430: Thus open field lines in the polar region have another boundary, 

431: defined by ``last open field lines'', $\theta=\theta_c(r)$. 

432: Therefore, our boundary conditions are $\Phi=\nabla\Phi=0$ on 

433: $r=r_{\ast}$ and on $\theta=\theta_c(r)$.

434: 

435: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

436: \section{Particle Acceleration in the Polar Cap}

437: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

438: 

439: We are interested in the ``polar cap'' region, $\theta<\theta_c(r)$, where particle 

440: acceleration may occur. For simplicity, we restrict ourselves to the region near 

441: the magnetic pole just above the surface, where the following approximations 

442: hold: (i) $\theta_c(r)\approx$ constant, (ii) $\theta\ll1$ (we take 

443: its first order), and (iii) $d/dr$ along the particle trajectory 

444: $\approx\pa/\pa r$. Therefore, we can apply our equation below only to

445: the region elongated less than the polar cap radius: $r-r_*<r\theta_c$.

446: 

447: As for the non-radial direction, we expand $\Phi$ as

448: \beq\label{Ylm}

449: \Phi=\sum_{l,m}\bar\Phi(r)Y_{lm}(\theta,\varphi).

450: \eeq

451: We only take the mode of the polar cap scale, $l\approx\pi/\theta_c$.

452: 

453: The above assumptions simplify (\ref{divEax}) as

454: \bea\label{divEpc}

455: &&-{\sqrt{F}\over r^2}{d\over dr}\left(r^2{d\Phi\over dr}\right)

456: +{l(l+1)\over\sqrt{F}r^2}\Phi

457: =4\pi\left({J\over V}-\rho_{GJ}\right), \\

458: &&4\pi\rho_{GJ}={2B\Omega\over\sqrt{F}}

459: \left\{\left(1-{\omega\over\Omega}\right)\cos\chi+\theta Q\cos\varphi\sin\chi\right\},

460: \label{GJ}\\

461: &&Q\equiv{r_g\over 2r}-{3\omega\over 2\Omega}+{r_g\omega\over r\Omega}

462: +{3\over 2fF}\left(1-{3r_g\over 2r}+{\omega\over 2\Omega}\right)

463: \approx\frac32-{11r_g\over r}-{3\omega\over 4\Omega},

464: \label{Q}\eea

465: where $B\equiv|\bB|=B_{(r)}+O(\theta^2)$

466: \footnote{

467: Equation (\ref{Q}) is slightly different from the expression obtained by 

468: Muslimov and Tsygan (1992) and Harding and Muslimov (2001):

469: \beq

470: Q={3r_g\over 2r}-{3\omega\over 2\Omega}

471: +{3\over 2fF}\left(1-{3r_g\over 2r}+{\omega\over 2\Omega}\right).

472: \eeq

473: We believe that there is an error in their calculation. 

474: Nevertheless this difference does not change results practically; 

475: because we have assumed $\theta\ll1$, all terms but 2/3 (the dominant 

476: one) in $Q$ are negligible in equation (\ref{GJ}).

477: }.

478: Equation (\ref{eom}) describes the 

479: time-evolution of particles; however, if particles make a stationary flow, we can read their spatial 

480: configuration by replacing the time $\tau$ with a spatial valuable $r$:

481: \beq\label{eompc}

482: {d\over dr}(\sqrt{F}\gamma)=\frac em{1\over F}{d\Phi\over dr}.

483: \eeq

484: 

485: Using the normalized variables \cite{Shi97},

486: \bea

487: j&\equiv&-{2\pi\sqrt{F(s)}J(s)\over\Omega B(s)}={\rm constant}, ~~

488: \phi(s)\equiv\frac em\Phi(s), ~~

489: s\equiv\sqrt{{2\Omega B_0 e\over mc^2}}r,\\

490: \bar j(s)&\equiv&-{2\pi\sqrt{F(s)}\rho_{GJ}(s)\over\Omega(s)B(s)}

491: =\left(1-{\omega(s)\over\Omega}\right)\cos\chi+\theta(s)Q(s)\cos\varphi\sin\chi,

492: \eea

493: we rewrite (\ref{divEpc}) and (\ref{eompc}) as

494: \bea\label{eq1}

495: {\sqrt{F}\over s^2}{d\over ds}\left(s^2{d\phi\over ds}\right)

496: -{l(l+1)\over\sqrt{F}s^2}\phi

497: &=&{B\over B_0\sqrt{F}}\left({j\over V}-\bar j\right), 

498: \\\label{eq2}

499: {d\over ds}(\sqrt{F}\gamma)&=&{1\over F}{d\phi\over ds}.

500: \eea

501: The particle trajectory is given by integrating 

502: $d\theta/dr=B^r/B^{\theta}$, resulting in

503: \beq

504: \theta(s)\approx\theta_{\ast}\sqrt{{s\over s_{\ast}}} ~~ {\rm for} ~~

505: \theta\ll1,

506: \eeq

507: where $\ast$ denotes a value at the star surface $r=r_{\ast}$ for any 

508: variable.

509: In the following analysis we fix some of the parameters: $r_{\ast}=10$km, 

510: $r_g=2$km, $\Omega=2\pi/0.1{\rm s},~ \omega_{\ast}/\Omega=0.1,~

511: \theta_c=\sqrt{\Omega r_{\ast}/c}\approx 0.046~(\pi/\theta_c\approx=69),~

512: \chi=30^{\circ}$ and $\gamma_{\ast}=1.00001$.

513: 

514: We can argue equation (\ref{eq1}) on the analogy of particle motion by regarding 

515: $\phi$ as position and $s$ as time.

516: At a given point (or time) $j>\bar j(s)$, the ``force term'' (LHS) of (\ref{eq1}) always 

517: positive; $\phi$ and $\gamma$ increase acceleratingly. If $j<\bar j(s)$, on the other hand, the 

518: force term takes both positive and negative, depending on $V$, and hence oscillating 

519: behavior is expected. If we set $r_g=\omega=0$ (no gravity) and $\chi=0$ (no inclination), 

520: we reproduce Shibata's result, $\bar j=1$ \cite{Shi97}. The expression 

521: of $\bar j$ in (\ref{eq1}) indicates that both effects of gravity and 

522: of inclination reduce $\bar j$, or equivalently, the critical value of $j$.

523: 

524: Numerical solutions of (\ref{eq1}) and (\ref{eq2}) in Figs.\ 1-3

525: verify the above arguments. It is apparent that gravity reduces the critical 

526: value of $j$:  monotonic increase of $\gamma$ appears with lower values of $j$ 

527: that in non-GR case.

528: 

529: Let us consider the dynamics for $j\approx\bar j_{\ast}$ carefully. To 

530: see the behavior analytically, we expand $\bar j$ with

531: $\xi\equiv s/s_{\ast}-1$ up to the first order, resulting in

532: \beq

533: \bar j(\xi)=\bar j_{\ast}\left[1+C\xi\right],

534: ~~ C\approx 

535: 3\left({\omega_{\ast}\over\Omega}+{\theta_{\ast}\over4}\cos\varphi\tan\chi\right).

536: \eeq

537: This implies that $\bar j$ increases or decreases according to the sign of $C$. 

538: For the particles on the magnetic axis ($\theta=0$), $\bar j$ is an increasing 

539: function due to the term $\omega_{\ast}/\Omega$ in GR case while it 

540: is constant in non-GR case.

541: This gravitational effect is illustrated in Fig.\ 1:

542: contrary to non-GR case (a), in GR case (b) oscillation for 

543: $j<\bar j_{\ast}$ does not continue to infinity 

544: but the particle velocity becomes $V=0$ at some point. Although the dynamics 

545: is independent of $\gamma_{\ast}$ in non-GR case, the 

546: ending point $V=0 ~(\gamma=1)$ depends on $\gamma_*$ in GR case: it becomes further for 

547: larger $\gamma_{\ast}$.

548: 

549: Dynamics of particles off the magnetic axis ($\theta\ne 0$) is more 

550: interesting. In the absence of gravity ($\omega_{\ast}=0$), the sign 

551: of $C$ depends simply on whether particles move ``toward'' to 

552: ($\cos\varphi>0$) or ``away'' from ($\cos\varphi<0$) the rotation axis, 

553: as claimed by Shibata (1997). He showed, for example, that 

554: particles are accelerated after oscillation along ``away'' field 

555: lines if $j$ is slightly less than 1. This behavior is reproduced in 

556: Fig.\ 2(a). Gravity changes this behavior drastically. If we take 

557: $\omega_{\ast}/\Omega=0.1$ typically, $C$ becomes positive even on ``away'' 

558: field lines unless $\theta_{\ast}>0.4|\sec\varphi|\cot\chi>0.4\cot\chi$. 

559: Unless $\chi$ is large enough, in most small-$\theta$ region, $\bar j$ 

560: simply increases and acceleration after oscillation 

561: cannot occur. This argument is confirmed by the numerical result in 

562: Fig.\ 2(b).

563: 

564: For the particles moving ``toward'' the rotation axis, $\cos\varphi$ is 

565: positive, and therefore $\bar j$ always increases along the field lines. 

566: The results in Fig.\ 3 are also understandable. Among the solutions 

567: in Fig.\ 3(b), the solutions with $j=1.003\bar j_{\ast}$ could be 

568: realistic because particle energy becomes so large that 

569: electron-positron pairs are created with finite electric field.

570: 

571: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

572: \section {Conclusions and Discussions}

573: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

574: 

575: We have reconstructed a 3+1 formalism of general relativistic

576: electromagnetism, and derived the equations of motion of charged 

577: particles in pulsar magnetosphere. Our basic equations are the 

578: correct and generalized version of those of Muslimov \& Tsygan (1992) and 

579: of Mestel (1996) in the sense that the equations include arbitrary current

580: density, arbitrary velocity which satisfies the equation of motion, 

581: and arbitrary inclination angle between the rotation axis and the magnetic axis. 

582: 

583: We have solved our equations of motion together with the electric 

584: field structure along the magnetic field for the region near the magnetic 

585: pole just above the surface in our approximate method. We have found 

586: that the effect of gravity on particle dynamics cannot be ignored and hence the condition for 

587: acceleration is also modified. In particular, particle dynamics on the ``away'' field lines 

588: is changed qualitatively by gravity if the current density is sub-critical.

589: 

590: Monotonic increase of $\gamma$ with the boundary condition

591: $\bE\cdot\bB=0$ at the stellar surface and at infinity

592: is found with GR effect, and the achieved Lorentz factor

593: is significantly larger than the case of non-GR (Muslimov \&

594: Tsygan 1992). With sub-critical current densities, we find

595: oscillatory solutions even with GR effect. Since we take the

596: current densities as a free parameter, the local accelerator

597: model can be linked with the global model by adjusting the 

598: current density.

599: 

600: We have also found that some oscillatory solutions disappear above 

601: a certain height. This may indicate an intermittent flow of

602: charged particles rather than a steady outflow if the imposed

603: current density is smaller than a threshold. This may cause

604: pulse nulling.

605: 

606: As pointed by Arons (1997), an important GR effect is local increase of $\bar{j}$, 

607: which appears on most of field lines, no matter whether large scale

608: field curvature of magnetic field lines is "away" or "toward".

609: He suggested that this fact can be responsible for the axisymmetric

610: radio emission about the magnetic axis. If the current density

611: is taken as a free parameter, the super-critical current density

612: in the polar annuli can also be responsible for the distribution 

613: of radio emitting region.

614: We shall extend our analysis to a more global region to construct a pulsar model in 

615: the subsequent work.

616: 

617: \acknowledgements

618: N.S.\ thanks Yasufumi Kojima for discussions, and also Yukawa 

619: Institute for Theoretical Physics, where a part of this work has been 

620: done as a visitor program. 

621: This work was partially supported by the Grant-in-aid for Scientific 

622: Research Fund of Monbukagakusho (Nos.\ 13740139 and 12640229).

623: 

624: \newpage

625: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

626: \begin{thebibliography}{}

627: \bibitem[Arons 1997]{Aro}

628: Arons, J. 1997, in "Neutron Stars and Pulsars" eds. Shibazaki, N.,Kawai, N.,

629: Shibata, S., \& Kifune, T. (Universal Academy Press, Inc.), 339

630: \bibitem[Arons \& Scharlemann 1979]{AS}Arons, J., Scharlemann, E.T. 1979, \apj, 231, 854

631: \bibitem[Ginzburg \& Ozenoi 1965]{GO}

632: Ginzburg, V.L., \& Ozenoi, L.M. 1965, Sov. Phys. JETP, 20, 689

633: \bibitem[Goldreich \& Julian 1969]{GJ}

634: Goldreich, P., \& Julian, W.H. 1969, \apj, 157, 869

635: \bibitem[Fawley et al. 1977]{FAS}

636: Fawley, W.M., Arons, J., Scharlemann, E.T. 1977, \apj, 217, 227

637: Fawley, W.M., Arons, J., Scharlemann, E.T. 1977, \apj, 217, 227

638: \bibitem[Harding \& Muslimov 2001]{HM}

639:  Harding, A.K., \& Muslimov, A.G. 2001, \apj, 556, 987

640: \bibitem[Konno \& Kojima 2000]{KK}

641: Konno, K., \& Kojima, Y. 2000, Prog. Theor. Phys., 104, 1117

642: \bibitem[Landau \& Lifshitz 1975]{LL}

643: Landau, L.D., \& Lifshitz, E.M. 1975, ``The Classical Theory of Fields'' (Pergamon)

644: \bibitem[Melrose 2000]{Mel}

645: Melrose, D.B. 2000, in ASP Conf. Ser. 202, ``Pulsar Astronomy: 2000 and beyond", 

646: eds. M. Kramer, N. Wex, \& R. Wielebinski, 721

647: \bibitem[Mestel 1981]{Mes81}

648: Mestel, L. 1981, in IAU Symposium 95, eds. W. Sieber \& R. Wielebinski (Dordrecht, Reidel), 9

649: \bibitem[Mestel 1996]{Mes96}

650: Mestel, L. 1996, in ASP Conf. Ser. 105, "Pulsars: Problems and Progress", 

651: eds. S. Jonston, M.A. Walker, \& M. Bailes (San Francisco: ASP), 417

652: \bibitem[Muslimov \& Tsygan 1992]{MT}Muslimov, A.G., \& Tsygan, A.I. 1992, MNRAS, 255, 61

653: \bibitem[Scharlemann, Arons \& Fawley 1978]{SAF}

654: Scharlemann, E.T., Arons, J., Fawley, W.M. 1978, \apj, 378, 239

655: \bibitem[Shibata 1991]{Shi91}Shibata, S. 1991, \apj, 276, 537

656: \bibitem[Shibata 1995]{Shi95}Shibata, S. 1995, MNRAS, 276, 537

657: \bibitem[Shibata 1997]{Shi97}Shibata, S. 1997, MNRAS, 287, 262

658: \bibitem[Thorne \& Macdonald 1982]{TM}

659: Thorne, K.S., \& Macdonald, D. 1982, MNRAS, 198, 339

660: \bibitem[Weinberg 1972]{Wei}

661: Weinberg, S. 1972, ``Gravitation and Cosmology'' (John Wiley \& Sons)

662: \end{thebibliography}

663: \clearpage

664: 

665: \begin{figure}

666: \figurenum{1}

667: \epsscale{.9}

668: \plotone{f1a.eps}

669: \epsscale{.9}

670: \plotone{f1b.eps}

671: \caption{Spatial distribution of $\gamma$ along magnetic field lines. We choose

672: $\theta=0$. (a) represents solutions for the flat spacetime with $r_g=\omega=0$.

673: $\gamma$ oscillates or increases monotonically according to $j$. (b) represents GR

674: solutions. In the case of $j<j_{\ast}$ oscillation 

675: eventually stops with $V$ approaching zero, in contrast with the non-GR case.}

676: \end{figure}

677: 

678: \begin{figure}

679: \figurenum{2}

680: \epsscale{.9}

681: \plotone{f2a.eps}

682: \epsscale{.9}

683: \plotone{f2b.eps}

684: \caption{Spatial distribution of $\gamma$ along ``away'' field lines. 

685: We choose $\theta_{\ast}=5^{\circ},~\varphi=180^{\circ}$. 

686: (a) and (b) represent flat and GR solutions, respectively.

687: Gravity changes particle dynamics for $j<\bar j_{\ast}$: acceleration after oscillation 

688: does not occur.}

689: \end{figure}

690: 

691: \begin{figure}

692: \figurenum{3}

693: \epsscale{.9}

694: \plotone{f3a.eps}

695: \epsscale{.9}

696: \plotone{f3b.eps}

697: \caption{Spatial distribution of $\gamma$ along ``toward'' field lines. 

698: We choose $\theta_{\ast}=5^{\circ},~\varphi=0^{\circ}$.

699: (a) and (b) represent flat and GR solutions, respectively.}

700: \end{figure}

701: 

702: \end{document}

703: