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

2: \usepackage{emulateapj5}

3: 

4: \shorttitle{Possible evidence for pulsed X-rays from outer gap}

5: \shortauthors{Wang et al.}

6: 

7: \input psfig.sty

8: %huai

9: \begin{document}

10: 

11: \title{Possible evidence for the pulsed X-ray emission from outer gap in PSR B1937+21?}

12: \author{H.G. Wang, R.X. Xu and G.J. Qiao}

13: \affil{Department of Astronomy, Peking University, Beijing 100871,

14: China}

15: 

16: %\email{wanghg@bac.pku.edu.cn}

17: 

18: \begin{abstract}

19: 

20: The fastest millisecond pulsar PSR B1937+21 presents an interpulse

21: separated from the main pulse by nealy $180^{\rm o}$ at radio

22: frequencies. Recently, the ASCA observations (Takahashi et al.

23: 2001) detected pulsed X-ray emission from this pulsar. Only a

24: single narrow X-ray pulse is observed, which is coincident with

25: the radio interpulse in phase. We investigate the possible origin

26: of the pulsed X-rays from the polar cap (PC) accelerators or the

27: outer gap (OG) accelerators in the frame of PC model and OG model,

28: respectively, by assuming a dipolar magnetic field structure and

29: the same radio emission pattern from its poles for the pulsar. The

30: OG model can naturally explain the main observational facts. For

31: the PC model, the coincidence between the X-ray pulse and the

32: radio interpulse can not be reproduced in the assumed case.

33: However when considering possible deviation from our assumption,

34: PC model may still be valid for this pulsar in some cases.

35: 

36: %The uncertainty of the maximum rate of position angle sweep and

37: %the implication of large values of this rate are discussed.

38: 

39: %We investigate the validity of the

40: %polar cap (PC) model and outer gap model (OG) for the X-ray

41: %emission from PSR B1937+21

42: %It shows that the PC model can be almost ruled out while the OG

43: %model can explain the main observational facts.

44: 

45: \end{abstract}

46: 

47: \keywords{pulsar: PSR B1937+21: X-rays: emission mechanism}

48: 

49: \section{Introduction}

50: 

51: Even more than 30 years after the discovering of high energy

52: pulsars, the theoretical reproduction of X-ray and $\gamma$-ray

53: emission from such pulsars is still a matter of debate. It is

54: commonly agreed that there are two scenarios on modelling X-ray

55: and $\gamma$-ray creation: the outer gap (OG) model (e.g., Cheng,

56: Ho \& Ruderman 1986a,b, Romani 1996, Cheng \& Zhang 1999, Hirotani

57: \& Shibata 2001) and the polar cap (PC) model (e.g., Harding 1981,

58: Sturner \& Dermer 1994, Daugherty \& Harding 1996, Luo et al.

59: 2000, Zhang \& Harding 2000). The fundamental difference between

60: these two types of models is the location of the regions where

61: particles are accelerated to relativistic energies and emit high

62: energy photons. The early PC model (Harding 1981) assumed that the

63: emission is produced just above the PC surface. In the present

64: versions of PC model (e.g. Daugherty \& Harding 1996, Harding \&

65: Muslimov 1998), it is proposed that the particle acceleration

66: region may extend from the PC surface to several stellar radius

67: above due to free charge flow and inertial frame dragging, so that

68: wide double-peak X-ray and $\gamma$-ray light curves can be

69: reproduced, given the inclination angle is not large. In contrast,

70: the OG model (Cheng et al. 1986a, CHR) presumed that the gaps can

71: exist in the outer magnetosphere between the null charge surface

72: and the light cylinder. Later, CHR model was developed to the

73: single OG models (e.g., Chiang \& Romani 1994, Romani \&

74: Yadigaroglu 1995, Cheng et al. 2000), which claimed that the three

75: dimension extents of OGs are constrained by the pair cascade

76: processes, and a single OG can produce wide double-peak high

77: energy light curves. Whereas theoretical considerations of more

78: detailed physical processes for particle acceleration and photon

79: emission are necessary, it is urgent and interesting to find new

80: observational evidence for these models.

81: %

82: 

83: PSR B1937+21, with the period of 1.56 ms, is the fastest

84: millisecond pulsar (MSP) known. At radio bands, it exhibits an

85: interpulse emission that is roughly equal to the main pulse in

86: intensity and separated from it by a phase of about $180^{\rm o}$.

87: Recently, the ASCA observations detected pulsed X-ray emission

88: from this pulsar (Takahashi et al. 2001). Only one nonthermal

89: narrow pulse was observed, which is coincident with the radio

90: interpulse in phase within the timing errors. The pulse width is

91: about $100\pm 61$ $\mu$s ($23^{\rm o}\pm 14^{\rm o}$). Besides the

92: narrow pulse, the light curve reveals two additional wide

93: Gaussian-shaped bulges above the background level, with each phase

94: interval being about 0.5 rotation period and each peak intensity

95: $\sim1/4$ of the nonthermal pulse peak.

96: 

97: 

98: Where does the X-ray emission of the fastest MSP come from, the PC

99: or the OG? It has not been extensively studied as the other high

100: energy pulsars such as the Crab pulsar and Vela. The previous work

101: was done by Luo et al (2000), in which the PC model was modified

102: for MSPs. They applied the theory to PSR B1937+21 and suggested

103: that the X-ray emission probably originates from the location of

104: one stellar radius above the PC.

105: 

106: In this paper, efforts on modelling the observational data at

107: radio and X-ray bands are made for both the PC and OG models. The

108: inclination angle between the rotation and magnetic axes is

109: constrained in section 2, which is necessary for the modelling.

110: The modelling is carried out in section 3. Conclusions and

111: discussions are placed in section 4.

112: 

113: \section{The inclination angle}

114: 

115: The inclination angle ($\alpha$) between the rotation and magnetic

116: axes is a necessary parameter for both the PC and OG models to

117: give various high energy emission beams. Unfortunately, there is

118: no agreement on the value of inclination angle of PSR B1937+21. In

119: this section we reinvestigate the value of $\alpha$ under the

120: double-pole model, viz., the radio interpulse and main pulse are

121: considered to be from the opposite magnetic poles of a dipolar

122: field, and the result is used in the calculation in section 3.

123: 

124: We assume that the radio emission beams from double poles are

125: axisymmetric around the magnetic axis and have the same radius,

126: namely, $\rho_{1}=\rho_{2}$. According to the geometry model (Gil

127: et al. 1984,  Lyne \& Manchester 1988, LM88) one has

128: %

129: %------------ Eq. 1------------------

130: \begin{equation}

131: \sin^{2}{{\rho_{1}}\over{2}}=\sin^{2}{{W_{1}}\over{4}}\sin\alpha_{1}\sin(\alpha_{1}+\beta_{1})+\sin^{2}{{\beta_{1}}\over{2}}

132: \label{eq:rou1}

133: \end{equation}

134: %

135: %-----------  Eq. 2--------------

136: \begin{equation}

137: \sin^{2}{{\rho_{2}}\over{2}}=\sin^{2}{{W_{2}}\over{4}}\sin\alpha_{2}\sin(\alpha_{2}+\beta_{2})+\sin^{2}{{\beta_{2}}\over{2}},

138: %\label{eq:rou2}

139: \end{equation}

140: %

141: where $\beta$ is impact angle between the line of sight (LOS) and

142: the magnetic axis, $W$ is the pulse width of the average profile,

143: the subscripts `1' and `2' denote the main pulse and the

144: interpulse, respectively. There are two simple geometrical

145: relations between the inclination angles and the impact angles,

146: namely, $\alpha_{2}=\pi-\alpha_{1}$, and

147: $\beta_{2}=\alpha_{1}+\beta_{1}-\alpha_{2}$.

148: 

149: >From the above relations $\beta_{1}$ can be derived as

150: %

151: %------------- Eq. 3-----------------

152: \begin{equation}

153: \beta_{1}=\tan^{-1}\left[{{A-\tan^{2}\alpha_{1}}\over{(A+1)\tan\alpha_{1}}}\right],

154: \label{eq:bt1}

155: \end{equation}

156: %

157: where

158: %

159: %------------ Eq. 4----------------

160: \begin{equation}

161: A={{1}\over{\sin^{2}{{W_{1}}\over{4}}}-\sin^{2}{{W_{2}}\over{4}}}.

162: \end{equation}

163: %

164: In the following we neglect the subscript `1', so that all the

165: $\alpha$, $\beta$ and $\rho$ below are referred to the main pulse

166: except for special declaration.

167: 

168: To figure out the value of $A$, the profile at 1.5 GHz is chosen

169: (downloaded from EPN) for its high time resolution and low

170: dispersion smearing (Kramer et al. 1998). The pulse widths are

171: measured at the level of 10\% of their peak intensities, which are

172: $19^{\rm o}.4\pm 0^{\rm o}.4$ and $17^{\rm o}.8 \pm 0^{\rm o}.4$

173: for the main pulse and the interpulse respectively. So that

174: $\beta(\alpha)$ and $\rho(\alpha)$ can be calculated, as shown by

175: the solid curve and the curve dotted by circles respectively in

176: Fig.1. The $\beta(\alpha)$ curve approximates to a linear function

177: of $\alpha+\beta\thickapprox90^{\rm o}$ when

178: $\alpha\lesssim85^{\rm o}$, which is determined by the fact that

179: the main pulse and interpulse are both narrow and differ only a

180: little in pulse width.

181: 

182: For a dipolar field, the shape of the polar cap is found to change

183: from a circle (for $\alpha=0^{\rm o}$) to an ellipse of which the

184: longitudinal radius is about 1.6 times of the latitudinal radius

185: (for $\alpha=90^{\rm o}$, Cheng et al. 2000). Since the deviation

186: from circular shape is not essentially significant, we simply

187: regard the polar cap as a circle in this section. So that the

188: opening angle (between the magnetic axis and the tangent of the

189: magnetic field line) of the polar cap $\rho_{\rm PC}$ can be

190: determined by the last open field line on the plane containing the

191: rotation and magnetic axes (${\bf \Omega}-{\bf \mu}$ plane). The

192: radius $\rho_{\rm PC}$ is a function of the stellar radius R and

193: the inclination angle $\alpha$, as shown by the dash curves in

194: Fig.1 for $R=3$ km and 10 km.

195: 

196: We further assume that the boundary of the radio beam is defined

197: by the last open field lines, then the beam radius should be

198: greater than $\rho_{\rm PC}$. According to this criteria, the

199: inclination angle is constrained to be $\alpha\lesssim 63^{\rm o}$

200: for $R=10$ km and $\alpha\lesssim 77^{\rm o}$ for $R=3$ km. As the

201: radii of neutron stars (NSs) are currently believed to be about 10

202: km, we accept that $\alpha\lesssim 63^{\rm o}$.

203: 

204: Can we determine the exact value of $\alpha$? It is sure if the

205: maximum rate of position angle $({\rm d}\psi/{\rm d}\phi)_{\rm

206: max}$ is exactly known, where $\psi$ is the position angle and

207: $\phi$ the pulse longitude. The maximum rate presents the second

208: relationship between $\alpha$ and $\beta$, which reads

209: %

210: %------------- Eq.5---------------

211: \begin{equation}

212: \left({{\rm d}\psi}\over{{\rm d}\phi}\right)_{\rm

213: max}={{\sin\alpha}\over{\sin\beta}}  . \label{eq:ps}

214: \end{equation}

215: %

216: Combining with Eq.s~\ref{eq:bt1} and ~\ref{eq:ps}, $\alpha$ and

217: $\beta$ can be solved for a given $({\rm d}\psi/{\rm d}\phi)_{\rm

218: max}$.

219: 

220: Although recent polarization observations present flat position

221: angle sweeps (Thorsett \& Stinebring 1990, Stairs et al. 1999), it

222: can not be asserted that the real value of $({\rm d}\psi/{\rm

223: d}\phi)_{\rm max}$ is small, because observations may give a less

224: steep position angle gradient due to smearing of finite sampling

225: time, to the frequency dispersion in pulse arrival time (Liu \&

226: Wu, 1999), and to the interstellar scattering (Gil 1985a). The

227: $\beta^{\prime}(\alpha)$ curves derived from Eq.~\ref{eq:ps} for

228: $({\rm d}\psi/{\rm d}\phi)_{\rm max}= 1, 3$ and 20 respectively

229: are presented by the dotted curves in Fig.1. The intersections of

230: $\beta^{\prime}(\alpha)$ and $\beta(\alpha)$ shows that a larger

231: $({\rm d}\psi/{\rm d}\phi)_{\rm max}$ results in a larger $\alpha$

232: and smaller $\beta$, which means the LOS sweeps across the radio

233: beam more closely to the beam center.

234: 

235: One may find that when the real value of $({\rm d}\psi/{\rm

236: d}\phi)_{\rm max}$ is large enough, $\alpha$ would exceed $63^{\rm

237: o}$ (for example, taken $({\rm d}\psi/{\rm d}\phi)_{\rm max}=3$ as

238: proposed by Gil (1985a), $\alpha$ is $71^{\rm o}$) and hence

239: contradict against $\alpha\lesssim 63^{\rm o}$. However, if the

240: radius is smaller, for example, $R=3$ km, this inconsistency would

241: cancel. In fact, Xu et al. (2001) suggested that PSR B1937+21 is

242: probably a strange star (SS) with low mass and small radius. The

243: detailed discussion is placed in section 4.

244: 

245: The range of $\alpha$ presented above is different from the

246: conventional consideration in double-pole model that $\alpha$

247: should be close to 90$^{\rm o}$ (Stairs et al. 1999).

248: Alternatively, there is another kind of so-called single pole

249: model to interpret the interpulse, which suggests that the

250: interpulse emission comes from the same pole as the main pulse. In

251: the single pole model proposed by Gil (1985a) for PSR B1937+21,

252: $\alpha$ only need to be 20$^{\rm o}$. The single pole model

253: predicts that the separation between the main pulse and interpulse

254: may be close to 180$^{\rm o}$ and is frequency independent (Gil

255: 1983, 1985b), which is coincident with the observation of PSR

256: B1937+21 (Hankins \& Fowler 1986). However, observations with high

257: time resolution (e.g. Kramer et al. 1998, Stairs et al. 1999)

258: failed to find the emission components between the main pulse and

259: interpulse, which was reported by Stinebring et al. (1984) and was

260: suggested to be a strong support to the single pole model (Gil

261: 1985a). Therefore, in this paper the radio emission is considered

262: to come from double poles. More confirmative estimations of $({\rm

263: d}\psi/{\rm d}\phi)_{\rm max}$ are expected to determine $\alpha$

264: and $\beta$.

265: 

266: 

267: \section{The origin of nonthermal X-ray emission from PSR B1937+21}

268: 

269: In this section we calculate the X-ray beams in the frames of both

270: the PC and OG models to find out whether they are able to

271: reproduce the observational facts. The facts used here are:

272: 

273: (1) at 1.4 GHz the separation between the peaks of interpulse and

274: main pulse is $174^{\rm o}$ (measured from the profile presented

275: by Takahashi et al. (2001));

276: 

277: (2) at 1.5 GHz (EPN data) the 10\% widths of the main pulse and

278: interpulse are $19.4^{\rm o}$ and $17.7^{\rm o}$, respectively;

279: 

280: (3) the nonthermal X-ray pulse is nearly coincident with the

281: interpulse, the X-ray pulse width is about $23^{\rm o}$ (Takahashi

282: et al. 2001).

283: 

284: \subsection{Origin from the PC accelerators?}

285: 

286: Luo et al. (2000) discussed the viability of PC models for high

287: energy emission from MSPs. They found that the maximum Lorenz

288: factor of particles is limited by curvature radiation and not

289: sensitive to the specific acceleration model, but the height where

290: the Lorenz factor achieves the maximum is model dependent, which

291: may be between $0.01R$ (for the inner vacuum gap) and above $0.1R$

292: (for the space-charge limited gap) from PC surface for pulsar

293: period $P=2$ ms and a surface magnetic field $B_{\rm

294: s}=7.5\times10^{8}$ G. Assuming a space-charge limited flow, the

295: pair cascades can occur at the typical distance (to the star

296: center) of $r\simeq (1.5-2.5)R$, and high energy emission is

297: radiated from this region. With respect to $P=1.56$ ms and $B_{\rm

298: s}=4.1\times10^{8}$ G for PSR B1937+21, their analysis applies to

299: this pulsar.

300: 

301: Since the radio emission is also radiated from the region near the

302: PC, homocentric radio and X-ray beams are produced (Fig.2a).

303: Quantitatively, we assume the radio emission arises from the PC up

304: to the distance $r=2.5R$, the X-rays may arise from the PC to a

305: (a) relatively higher distance, e.g., $r=3R$, or (b) lower

306: distance, e.g., $r=2R$. For simplicity we assume the emission

307: regions are bounded by the last open field lines. In case (a) the

308: X-ray beams are wider than the radio beams, therefore, when the

309: line of sight sweeps across the both radio beams (to reproduce the

310: main pulse and the interpulse) it would sweep across the both

311: X-ray beams either, and hence gives double X-ray pulse, which is

312: inconsistent with the observation. In case (b) the X-ray beams are

313: narrower than the radio beams, and then, for a proper viewing

314: angle (between LOS and the rotation axis), only one X-ray beam

315: could be observed. In this case, could the fact (3) be explained

316: as well? We analyze this issue by using the figure of the

317: (phase-viewing angle) plane on which the emission beams are

318: projected.

319: 

320: In Fig.3, the horizon axis is pulse phase, the vertical axis is is

321: the viewing angle $\zeta$ ($\zeta=\alpha+\beta$). The dotted

322: curves represent the boundaries of X-ray beams, the solid curves

323: are the boundaries of radio beams, and the PCs are shown by the

324: dashed curves. Retardation (due to distinct emission heights) and

325: abberation effects (due to emission sources co-rotating with the

326: pulsar) have been taken into account, both of which make the lower

327: beam move towards the trailing edge of the higher beam. To obtain

328: the figure, a moderate inclination angle is taken, $\alpha=60^{\rm

329: o}$. Although there is $\alpha\lesssim 63^{\rm o}$ as discussed in

330: section 2, the angle should not be too small, or it would give too

331: wide radio beams (Fig.1), which must be emitted from unbelievably

332: high distances near the light cylinder.

333: 

334: According to Fig.3, when $\zeta\simeq80^{\rm o}$ the X-ray peak is

335: coincident with the radio pulse centered on phase about 0.9; when

336: $\zeta\simeq100^{\rm o}$ it is coincident with the radio pulse on

337: phase about 0.4.

338: %

339: However in both cases,  the radio pulse associated with the X-ray

340: pulse is the main pulse which is wider and more intense than the

341: other one.

342: 

343: However, we should point out that the above analysis is based on a

344: simple assumption that the radio (X-ray) beams from both poles

345: have the same width and the emission pattern in the beams is

346: symmetrical about the magnetic axis. Since the magnetosphere

347: structure of pulsars and the detailed radio emission process are

348: still unknown exactly, possibly significant deviation from the

349: assumption can not be ruled out, and the validity of the PC model

350: need to be considered further. We discuss such possibility in

351: section 4.

352: 

353: \subsection{Origin from the OG accelerators}

354: 

355: In the original OG model CHR suggested that a global current flow

356: through the magnetosphere can result in large regions of OGs

357: between the null charge surface and the light cylinder along the

358: last open field lines. Within the OGs particles with one kind of

359: charge are accelerated outward from the star and give an outward

360: emission beam, while those with the opposite charge are

361: accelerated towards the star and give an inward beam. The high

362: energy photons were proposed to be emitted from two OGs associated

363: with the two poles so that double-peaked $\gamma$-ray pulse

364: profile can be reproduced, of which one peak corresponds to the

365: outward beam from one OG and the other peak to the inward beam

366: from the opposite OG. CHR assumed that the OGs are active only

367: near the ${\bf \Omega}-{\bf \mu}$ plane. However, this assumption

368: is merely valid for large inclination angles.

369: 

370: The currently prevalent OG models are the single OG models (e.g.

371: Chiang \& Romani 1992, Chiang \& Romani 1994, Cheng et al. 2000).

372: Generally the inward emission is not important in these models for

373: the reason that the inward high energy photons can not pass freely

374: through the inner magnetosphere due to magnetic pair production.

375: The outward emission from the OG associated with a single pole can

376: produce a broad, irregularly-shaped emission beam which is

377: particularly dense near the edge.

378: %

379: The OG regions can be supported along all the last open field

380: lines, but the three dimension scales of OGs are limited by the

381: pair production processes.

382: 

383: In the latest version of this type of model (Cheng et al. 2000),

384: the fraction size ($f\equiv h/R_{\rm L}$) of the gap is $f\simeq

385: 5.5P^{26/21}B_{12}^{-4/7}\xi^{-1/7}$, which can be estimated by

386: the threshold of $\gamma-\gamma$ pair production, $E_{\rm

387: X}(f)E_{\gamma}(f)\geqslant(m_{\rm e}c^{2})^{2}$, where $h$ is the

388: mean vertical extension perpendicular to the magnetic field,

389: $R_{\rm L}$ is the radius of light cylinder,

390: $\xi=\Delta\phi/2\pi$, $\Delta\phi$ is the transverse extension of

391: the gap, $E_{\rm x}$ is the energy of the X-ray photons emitted

392: from hot PCs, and $E_{\gamma}$ is the characteristic photon energy

393: emitted by the relativistic particles. The radial scale of pair

394: production is limited to a range between $r_{\rm in}$ and $r_{\rm

395: lim}\sim 6r_{\rm in}(\xi=0)$, where $r_{\rm in}$ (the subscript

396: `in' means the inner boundary of the OG) is the distance of null

397: charge surface, $\xi=0$ corresponds to the ${\bf \Omega}-{\bf

398: \mu}$ plane.

399: 

400: In the following our modelling is in the frame of single OG model.

401: Only the outward emission beams from two OGs are considered, as

402: illustrated by Fig.2b. For PSR B1937+21, we have a thin OG with

403: $f=0.16$ $\xi^{-1/7}$, so that the X-rays can be simply regarded

404: as being radiated from the last open field lines unless $\xi$ is

405: too small. The radial scale $r_{\rm lim}/r_{\rm in}$ and the

406: transverse scale $\xi$ are free parameters in calculating the

407: X-ray beams.

408: 

409: First we consider a general situation of the OG scenario to test

410: if the observation facts can be hopefully reproduced. From Fig.2b

411: one can see that provided the observer's viewing angle is not just

412: $90^{\rm o}$, e.g., $\zeta=83^{\rm o}$, the LOS can sweep across

413: both of the radio beams and only one X-ray beam. By assuming a

414: group of reasonable values of parameters (see the first line of

415: Table 1), namely, the inclination angle, the stellar radius, the

416: distance of radio emission, and the radial and transverse

417: extensions of the OGs, the X-ray beams are calculated, which are

418: demonstrated by the line-shadowed areas in Fig.3. Retardation and

419: abberation effects are also included. It shows clearly that the

420: X-ray pulse could be associated with the radio interpulse (which

421: has a smaller width than the other one), and the narrow X-ray

422: pulse width may be obtained if proper extension of the OG is

423: assumed.

424: 

425: Then we model the observational data. The parameters listed in the

426: second line of Table 1 are found out to be able to reproduce the

427: narrow X-ray pulse width, the radio pulse widths, and the

428: coincidence between the X-ray pulse and the radio interpulse,

429: which are in good agreement with the observational data. In order

430: to simulate the observational profiles, we simply assume that the

431: X-ray and radio pulses are Gaussian shapes, and assume

432: additionally wide, weak, hot X-ray emission from both of the PCs.

433: The theoretical profiles are plotted in Fig.4, together with the

434: observational profiles for comparison.

435: 

436: It should be pointed out that modelling the X-ray pulse width is

437: not sensitive to the value of $r_{\rm lim}/r_{\rm in}$ but to

438: $\xi$, thereby the range of $r_{\rm lim}/r_{\rm in}$ is relaxed

439: and a reasonable value is chosen. Other groups of values are also

440: tried. It is found that for the moderate inclination angles

441: $40^{\rm o}\lesssim\alpha\lesssim63^{\rm o}$ the observational

442: data can be reproduced with proper choice of the gap size

443: ($\xi=40^{\rm o}$ in Table 1 is approximately the maximum

444: transverse scale). Therefore, our calculation suggests that the

445: nonthermal X-rays of PSR B1937+21 may be emitted from the OGs.

446: %and the pulsar may be an oblique rotator with a moderate inclination

447: %angle.

448: 

449: \section{Conclusions and Discussions}

450: 

451: The discovery of pulsed X-ray emission from the MSP PSR B1937+21

452: and the phase alignment between the X-ray pulse and the radio

453: interpulse provide valuable information on the emission mechanism.

454: In this paper we investigate the possible origin of the X-rays

455: from both the PC and OG accelerators. In the frame of the

456: prevailing OG model (e.g. Cheng et al. 2000), the X-rays from

457: outer gap accelerators could account for the main observational

458: facts by assuming proper size of the OGs for an oblique rotator.

459: In the frame of PC model, by assuming symmetric geometry for radio

460: and X-ray emission, the X-rays from polar cap accelerators would

461: produce an X-ray peak aligned with the main radio pulse in phase,

462: which is contradictive against the observation.

463: 

464: Some more discussions for both the OG and PC models are presented

465: as follows.

466: %

467: First we refer to the OG model.

468: 

469: There is a slight inconsistency as shown in Fig.4, i.e., in the OG

470: model, the calculated separation between the interpulse and main

471: pulse is $180^{\rm o}$, $6^{\rm o}$ greater than the observational

472: value. This may be due to the retardation effect. The phase shift

473: for different heights can be roughly estimated as $\Delta s=\Delta

474: r/(Pc)$. A difference of $\Delta r=0.8R$ is enough to produce the

475: phase shift of $6^{\rm o}$.

476: 

477: A moderate inclination angle $55^{\rm o}$ is used in the

478: modelling. But $\alpha$ could be larger if the real gradient of

479: position angle is steeper than the present observations as

480: suggested in section 2. Could the OG model be still valid for

481: large value of $\alpha$?

482: 

483: As shown in Fig.1, when $({\rm d}\psi/{\rm d}\phi)_{\rm max} \ga

484: 4$ it would result in a puzzling problem that the derived radio

485: beam radius is considerably smaller than the PC radius, if $R=10$

486: km and a magnetic dipole are assumed. A much smaller stellar

487: radius could cancel the problem, but this requires an SS scenario,

488: because the smallest radius of NSs is generally believed to be

489: $\sim$9 km while SSs can have much smaller radius due to their

490: different equations of state. In fact, according to the

491: observational limits on the radius and mass derived from the pulse

492: width and $({\rm d}\psi/{\rm d}\phi)_{\rm max}$, Xu et al. (2001)

493: suggested that PSR B1937+21 is probably a strange star with much

494: low mass, small radius and weak magnetic moment. If PSR B1937+21

495: is an SS with small radius, for example, $R=2$ km, a group of

496: parameters is found out to be able to model the observational data

497: by OG model, which is listed in the third line in Table 1.

498: 

499: Then we turn to the PC model. When the emission geometry is not

500: symmetrical, the PC model could be able to explain the observation

501: as well. The possible asymmetry is discussed for the radio and the

502: X-ray emission, respectively.

503: 

504: (1) It has been assumed that the radio emission pattern from the

505: two poles is the same in our modelling; the different behaviors of

506: the main pulse and the interpulse are thus geometrical origin.

507: However, their difference may be caused intrinsically since the

508: mechanism for the radio emission is still poorly understood. For

509: example, a possible reason may be that the pulsar has a

510: non-dipolar magnetic field, thus the distribution of emission

511: intensity in the radio beams may be different from each other,

512: which could lead to that for one beam only part of it is observed

513: while for the other a larger part or the whole is observed.

514: 

515: (2) When PSR B1937+21 is an NS or an SS with crust and its

516: inclination angle is near $90^{\rm o}$, the X-ray emission pattern

517: may be significantly asymmetric about the magnetic axis. In this

518: case, the space charge is negative on the side toward the rotation

519: axis (where ${\bf \Omega\cdot B}>0$, hereafter side I) and

520: positive on the side away from the axis (where ${\bf \Omega\cdot

521: B}<0$, hereafter side II). On side II the ions can be pulled away

522: from the surface by strong electric field if the binding energy is

523: small enough. Therefore, the half beam from side II could be much

524: less luminous than that from side I due to the much smaller

525: Lorentz factors of the ions, and thus may be too weak to be

526: observed. Notice that, if side I is above the equator on one pole,

527: it should be below the equator on the other pole.

528: %

529: Therefore, only single X-ray peak is observed which may coincide

530: with the radio interpulse.

531: 

532: In recent years, it is suggested that pulsars may be bare strange

533: stars (BSSs) (e.g., Xu 2002 and references therein). If PSR

534: B1937+21 is also a BSS, the positive charge on side II is carried

535: by positrons but not ions, then the emission pattern on both sides

536: should be the same, and each beam may be symmetric around the

537: magnetic axis. In this case, when symmetrical geometry is assumed

538: for the radio emission, PC model could not account for the

539: observational facts, otherwise, PC model may still be valid.

540: 

541: In general, further research on the pulsar's magnetosphere

542: structure and emission mechanisms will be helpful to understand

543: the origin of its X-rays for PSR B1937+21. We are expecting that

544: future polarization observations could provide more confirmative

545: value of $({\rm d}\psi/{\rm d}\phi)_{\rm max}$, which is

546: meaningful not only for constraining whether this pulsar is an NS

547: or an SS but also for a better understanding of its X-ray

548: emission.

549: 

550: 

551: \vspace{0.2cm} %

552: \noindent {\it Acknowledgments}:

553: 

554: We are grateful to Prof. J.A. Gil and Dr. B. Zhang for their

555: helpful comments and discussions. The valuable suggestions from an

556: anonymous referee are also sincerely acknowledged. This work is

557: supported by National Nature Science Foundation of China

558: (10173002), and by the Special Funds for Major State Basic

559: Research Projects of China.

560: 

561: \begin{thebibliography}{}

562: 

563: \bibitem[Cheng et al.(1986)]{CHR86a} Cheng, K.S., Ho, C. \& Ruderman, M.A., 1986a, ApJ, 300, 500

564: \bibitem[Cheng et al.(1986)]{CHR86b} -------- 1986b, ApJ, 300, 522

565: \bibitem[Cheng et al.(2000)]{C00} Cheng, K.S., Ruderman, M.A. \& Zhang, L., 2000, ApJ, 537, 964

566: \bibitem[Cheng \& Zhang 1999]{CZ99} Cheng, K.S., \& Zhang, L., 1999, ApJ, 515,

567: 337

568: \bibitem[Chiang \& Romani(1992)]{CR92} Chiang, J. \& Romani, R.W., 1992, ApJ, 400,

569: 629

570: \bibitem[Chiang \& Romani(1994)]{CR94} -------- 1994, ApJ, 436,

571: 754

572: \bibitem[Daugherty \& Harding(1996)]{DH96} Daugherty, J.K. \& Harding, A.K., 1996, ApJ, 458, 278

573: \bibitem[Gil(1983)]{G83} Gil, J., 1983, A\&A, 127, 267

574: \bibitem[Gil(1985)]{G85a} -------- 1985a, A\&A, 143, 443

575: \bibitem[Gil(1985)]{G85b} -------- 1985b, ApJ, 299, 154

576: \bibitem[Gil et al. 1984]{GGR84} Gil, J., Gronkowski, P. \& Rudnicki, W., 1984, A\&A, 132,

577: 312

578: \bibitem[Hankins \& Fowler 1986]{HF86} Hankins, T.H. \& Fowler, L.A., 1986, ApJ, 304,

579: 256

580: \bibitem[Harding 1981]{H81} Harding, A.K., 1981, ApJ, 245, 267

581: \bibitem[Harding et al.(1998)]{HM98} Harding, A.K. \& Muslimov, A.G., 1998, ApJ,

582: 508, 328

583: \bibitem[Hirotani \& Shibata 2001]{HS01} Hirotani, K. \& Shibata,

584: S., 2001, ApJ, 558, 216

585: \bibitem[Kramer et al.(1998)]{k98} Kramer, M. et al., 1998, ApJ,

586: 501, 270

587: \bibitem[Liu et al.(1999)]{L99} Liu, X.F. \& Wu, X.J., 1999, Acta Astrophys. Sin.,

588: 19, 68

589: \bibitem[Luo et al.(2000)]{L00} Luo, Q., Shibata, S. \& Melrose,

590: D.B., 2000, MNRAS, 318, 943

591: \bibitem[Lyne \& Manchester.(1988)]{LM88} Lyne, A.G. \& Manchester, R.N., 1988, MNRAS, 234, 477

592: \bibitem[Romani 1986)]{R96}Romani, R.W., 1996, ApJ, 470, 469

593: \bibitem[Romani \& Yadigaroglu 1995]{RY95} Romani, R.W. \& Yadigaroglu, I.-A., 1995, ApJ,

594: 438, 318

595: \bibitem[Stairs et al.(1999)]{S99} Stairs, I.H., Thorsett, S.E. \& Camilo, F., 1999, ApJs, 123, 627

596: \bibitem[Stinebring et al. 1984)]{SBCWD84} Stinebring, D.R., Boriakoff, V., Cordes, J.M., Wolszczan, A. \& Deich,

597: W.T.S.:1984, Millisecond Pulsar Workshop, Green Bank, West

598: Virginia, eds. S. Reynolds, D. Stinebring

599: \bibitem[Sturner \& Dermer 1994)]{SD94} Sturner, S.J. \& Dermer, C.D., 1994, ApJ, 420,

600: L79

601: \bibitem[Takahashi et al.(2001)]{T01} Takahashi, M. et al., 2001, \apj,

602: 554, 316

603: \bibitem[Thorsett(1990)]{TS90} Thorsett, S.E. \&

604: Stinebring, D.R., 1990, ApJ, 361, 644

605: \bibitem[Xu (2002)]{X02} Xu, R.X., 2002, ApJ, 570, L65

606: \bibitem[Xu et al.(2001)]{X01} Xu, R.X., Xu, X.B. \& Wu X.J., 2001,

607: Chin.Phys.Lett, 18, 837

608: \bibitem[Zhang \& Harding (2000)]{XH00} Zhang, B. \& Harding, A.K., 2000,

609: ApJ, 532, 1150

610: 

611: \end{thebibliography}

612: 

613: 

614: %\clearpage

615: 

616: \begin{deluxetable}{cccccccc}

617: \tablecaption{Parameters for calculating the radio and X-ray

618: beams/profiles}%

619: \tablewidth{0pt}%

620: \tablehead{ \colhead{No.} & \colhead{$\alpha$ ($^{\rm o}$)} &

621: \colhead{$\zeta$ ($^{\rm o}$)} & \colhead{$R$ (km)} &

622: \colhead{$r\tablenotemark{a}$ $$ ($R$)} & \colhead{$r_{\rm

623: in}(0)$\tablenotemark{b} $$ ($R$)}& \colhead{$r_{\rm lim}/r_{\rm

624: in}$} &

625: \colhead{$\Delta\phi$ ($^{\rm o}$)} }%

626: 

627: \startdata

628: 1 & 60 & - & 10 & 2.5 & 1.3 & 3.5 & 100 \\

629: 2 & 55 & 89.5 & 10 & 1.7 & 1.5 & 2.0 & 40 \\

630: 3 & 75 & 89.5 & 2 & 2.9 & 2.1 & 2.5 & 66 \\

631: 

632: \enddata

633: \tablenotetext{a}{the distance of radio emission, in unit of

634: stellar radius $R$.}

635: %\tablecomments{b}

636: \tablenotetext{b}{the distance of the null charge surface on the

637: ${\bf \Omega}-{\bf \mu}$ plane.}

638: \end{deluxetable}

639: 

640: \clearpage

641: 

642: %Figure 1 ---------------------------------------

643: \begin{figure}

644: 

645: \centerline{\psfig{figure=f1.eps,height=80mm,width=100mm,angle=0}}

646: 

647: \caption{Plot of $\beta$ and $\rho$ as functions of inclination

648: angle $\alpha$. The curves of $\beta(\alpha)$ and $\rho(\alpha)$

649: are derived from the observational pulse widths. The dashed curves

650: are $\beta^{\prime}(\alpha)$ derived from Eq.~\ref{eq:ps}, given

651: $({\rm d}\psi/{\rm d}\phi)_{\rm max}=1, 3$ and $20$, respectively.

652: The dotted curves are the opening angle of the polar cap

653: ($\rho_{\rm PC}$), given $R=3$ km and $10$ km.\label{Figure1}}

654: \end{figure}

655: 

656: %\clearpage

657: 

658: %Figure 2 ---------------------------------------

659: \begin{figure}

660: 

661: \centerline{\psfig{figure=f2a.eps,height=80mm,width=80mm,angle=0}

662: \psfig{figure=f2b.eps,height=80mm,width=80mm,angle=0}}

663: 

664: 

665: \caption{(a) Scheme for the X-ray beams in the frame of PC model.

666: (b) Scheme for the X-ray beams produced by the OG model. Radio

667: beams are also plotted. \label{Figure2}}

668: \end{figure}

669: 

670: \clearpage

671: 

672: %Figure 3 ---------------------------------------

673: \begin{figure}

674: 

675: \centerline{\psfig{figure=f3.eps,height=80mm,width=100mm,angle=0}}

676: 

677: \caption{Emission beams projected onto the (phase-viewing angle)

678: plane for $\alpha=60^{\rm o}$. The solid curves are the boundaries

679: of radio beams, the dotted curves are the boundaries of X-ray

680: beams from the extended acceleration zones above the PCs suggested

681: by the PC model, the line-shadowed areas are the X-ray beams from

682: the outer gaps, and the dashed curves represent the PCs (See text

683: and the first line of Table 1 for the related parameters).

684: \label{Figure3}}

685: \end{figure}

686: 

687: %\clearpage

688: 

689: %Figure 4 ---------------------------------------

690: \begin{figure}

691: 

692: \centerline{\psfig{figure=f4.eps,height=80mm,width=100mm,angle=0}}

693: 

694: \caption{Theoretical X-ray and radio profiles (the dashed curves)

695: together with the observational profiles (Takahashi et al. 2001).

696: The parameters in the second line of Table 1 are used to obtain

697: the theoretical profiles. \label{Figure4}}

698: \end{figure}

699: 

700: \end{document}

701: