1: %\documentclaLin_Yi-Fungss[]{article}
2: %\documentstyle[12pt,aasms4]{article}
3: \documentstyle[11pt,aaspp4,flushrt]{article}
4: %\documentstyle[emulateapj]{article}
5: %\usepackage{emulateapj}
6: \newcommand\beq{\begin{equation}}
7: \newcommand\eeq{\end{equation}}
8:
9: \begin{document}
10:
11: \title{Anomalous X-ray Pulsars and Soft $\gamma$-ray Repeaters: Spectral
12: Fits and the Magnetar Model}
13:
14: \author{Rosalba Perna,\altaffilmark{1} Jeremy S. Heyl,\altaffilmark{2}
15: and Lars E. Hernquist}
16: \affil{\small Harvard-Smithsonian Center for Astrophysics, 60 Garden Street,
17: Cambridge, MA 02138; rperna@cfa.harvard.edu, jheyl@cfa.harvard.edu,
18: lhernqui@kona.harvard.edu}
19:
20: \and
21:
22: \author{Adrienne M. Juett and Deepto Chakrabarty}
23: \affil{\small Department of Physics and Center for Space Research,
24: Massachusetts Institute of Technology, Cambridge, MA 02139;
25: ajuett@space.mit.edu, deepto@space.mit.edu}
26:
27: \altaffiltext{1}{Harvard Junior Fellow}
28: \altaffiltext{2}{{\em Chandra} Fellow}
29:
30:
31: \begin{abstract}
32: The energy source powering the X-ray emission from anomalous X-ray
33: pulsars (AXPs) and soft $\gamma$-ray repeaters (SGRs) is still
34: uncertain. In one scenario, the presence of an ultramagnetized
35: neutron star, or ``magnetar'', with $B \simeq 10^{14} - 10^{15}$ G is
36: invoked. To investigate this hypothesis, we have analyzed archival
37: {\em ASCA} data for several known AXPs and SGRs, and fitted them with
38: a model in which all or part of the X-ray flux originates as thermal
39: emission from a magnetar. Our magnetar spectral model includes the
40: effects of the anisotropy of the heat flow through an ultramagnetized
41: neutron star envelope, reprocessing by a light element atmosphere, and
42: general relativistic corrections to the observed spectrum. We obtain
43: good fits to the data with radii for the emitting areas which are generally
44: consistent with those expected for neutron stars, in contrast to
45: blackbody (BB) fits, which imply much smaller radii. Furthermore, the
46: inclusion of atmospheric effects results in inferred temperatures which
47: are lower than those implied by BB fits, but however still too high to
48: be accounted by thermal cooling alone. An extra source of heating
49: (possibly due to magnetic field decay) is needed. Despite the harder
50: tail in the spectrum produced by reprocessing of the outgoing flux
51: through the atmosphere, spectral fits still require a considerable
52: fraction of the flux to be in a power-law component.
53: \end{abstract}
54:
55: \keywords{stars: neutron --- $X$-rays: stars}
56:
57:
58: \section{Introduction}
59:
60: The roughly half dozen so-called anomalous X-ray pulsars (AXPs;
61: Mereghetti \& Stella 1995; van Paradijs, Taam, \& van den Heuvel 1995)
62: have recently emerged as a distinct class of objects. They are slow
63: ($P\sim 5$--10~s) rotators with no evidence of Doppler shifts from
64: binary motion, and several are associated with supernova remnants,
65: suggesting that they are young objects. Owing to their long periods,
66: their rotational energy loss is far too
67: low to power their observed X-ray luminosities. Models to account for the
68: properties of these sources fall into two broad categories. In the
69: first class of models, the X-ray emission is powered by accretion,
70: which could result from a binary companion of very low mass (Mereghetti \&
71: Stella 1995), the debris of a disrupted massive companion (van
72: Paradijs, Taam \& van den Heuvel 1995; Ghosh, Angelini \& White 1997),
73: or material falling back after a supernova explosion (Corbet et
74: al. 1995; Chatterjee, Hernquist \& Narayan 2000; Alpar 1999, 2000;
75: Marsden et al. 2001). In the second class of models, accretion is not
76: involved; instead, the AXPs are hypothesized to be either
77: ultramagnetized neutron stars (NSs; Thompson \& Duncan 1996) or remnants of
78: Thorne-${\dot {\rm Z}}$ytkow objects (van Paradijs et al. 1995). In
79: the former case, the X-ray luminosity could be powered either by
80: magnetic field decay (Thompson \& Duncan 1996; Heyl \& Kulkarni 1998)
81: or by residual thermal energy (Heyl \& Hernquist 1997a,b).
82:
83: If the AXPs are indeed ultramagnetized NSs (or ``magnetars''; Duncan
84: \& Thompson 1992, 1995), then they may be related to another class of
85: objects, the soft $\gamma$-ray repeaters (SGRs; e.g. Kouveliotou et
86: al. 1999). In their quiescent X-ray emission, the SGRs have slow
87: pulsations similar to those seen in the AXPs. In addition, the SGRs
88: sometimes show strong hard X-ray/soft gamma-ray bursts, which can be
89: distinguished from the classical $\gamma$-ray bursts (GRBs) by their recurrence
90: and their spectra, which are generally much softer than those of the GRBs.
91: According to Thompson \& Duncan
92: (1996), these bursts could be the result of sudden releases of energy
93: resulting from rearrangement of the magnetic fields
94: in the crusts of highly magnetized NSs.
95:
96: In the magnetar model for the AXPs and SGRs, whether the X-ray
97: luminosity is powered by cooling or magnetic field decay or by
98: a combination of both, the thermally emitting area must be consistent with
99: a neutron star surface. However, blackbody fits to the spectra of
100: these objects require emitting areas that are only a small fraction of
101: the total surface. This discrepancy may arise at least partially
102: because the thermal emission of NSs most likely does not have a true
103: blackbody spectrum. Atmospheric effects are known to distort NS
104: spectra. Fallback after the supernova explosion, and/or
105: accretion from the interstellar medium, will likely cover the surface of the
106: star with light elements. (Note that to cover the surface to an X-ray
107: optical depth of unity requires only $\sim 10^{14} {\rm g} -
108: 10^{-19} M_\odot$ of material.) Moreover, this material is likely to
109: suffer significant fractionization on short timescales (Alcock \& Illarionov 1980;
110: Romani 1987). Owing to the
111: enormous surface gravity of the star, the heavier elements will
112: settle out in of order 1$-$100 sec, leaving an atmosphere made of
113: light elements. Besides atmospheric effects, the emergent NS spectrum
114: is also dependent on the underlying temperature distribution on the
115: stellar surface (in blackbody fits this is assumed incorrectly to be
116: uniform),
117: and
118: on the general relativistic effect of light deflection owing to the
119: large surface gravity. A spectral analysis that takes into account all
120: these effects is the purpose of this work.
121:
122: More specifically, in this {\em Letter}, we analyze archival {\em
123: ASCA} data for some of the known AXPs and SGRs, and fit them with a
124: model in which the X-ray flux is produced by thermal emission from a
125: highly magnetized NS with an atmosphere made of light elements. We
126: take into account the anisotropy in the flow of heat through the
127: envelope of the NS owing to the intense magnetic field, as well as
128: general relativistic corrections to the observed spectrum. Our fits
129: are consistent with emission from the entire surface of a neutron
130: star, supporting the interpretation of AXPs and SGRs as magnetars.
131:
132: \section{The X-Ray Spectrum of a Cooling Magnetar}
133:
134: We consider a highly magnetized neutron star cooling through an
135: accreted envelope. Heyl \& Hernquist (1998b, 2000) showed that, if
136: $B_p\ga 10^{12}$ G, the flux transmitted through the envelope can be
137: well approximated by $F\propto \cos^2\psi$, where $\psi$ is the angle
138: between the local radial direction and the magnetic field. For a dipolar field,
139: \beq
140: \cos^2\psi=4\cos^2\theta_p/(3\cos^2\theta_p+1)
141: \label{eq:cospsi}
142: \eeq
143: (Greenstein \& Hartke 1983), where $\theta_p$ is the angle between the
144: radial direction at position ($\theta,\phi$) on the
145: surface of the star, and the magnetic pole. In spherical coordinates, it
146: is given by
147: \beq
148: \cos\theta_p=\cos\theta\cos\alpha+\sin\theta\sin\alpha\cos\phi\;,
149: \label{eq:tetap}
150: \eeq where $\alpha$ is the angle that the magnetic pole makes with the
151: line of sight. For this study, we consider an orthogonal
152: rotator.\footnote{We find that spectral fits are insensitive to this
153: choice when considering the average flux over the rotational period of the
154: star.} The angle $\alpha$ is then coincident with the phase angle $\Omega t$,
155: where $\Omega$ is the angular velocity of the star. For the local
156: emission from the NS surface, $n(E,T)$, we assume a blackbody spectrum
157: modified by the presence of an atmosphere made of light elements, for
158: which we adopt the semianalytical model of Heyl \&
159: Hernquist\footnote{This model yields spectral intensities very close
160: to those of Pavlov et al. (1994), who computed detailed atmospheres
161: for magnetic field strengths on the order of $10^{12}-10^{13}$~G.
162: Preliminary results (D. Lloyd et al. in preparation) suggest
163: furthermore that realistic spectra are similar to ours even for much
164: stronger fields, $B\sim 10^{14}-10^{15}$ G. } (1998a), but with the
165: inclusion of limb darkening. This is parameterized by an angle
166: dependence of the intensity $\propto \cos^\beta\delta$. The dependence
167: on $\beta$ is explored in the fits.
168:
169: Let $R$ be the radius of the NS star, $M$ its mass (for which we adopt
170: $M=1.4 M_\odot$), $R_s=2GM/c^2$ its Schwarzschild radius, and define
171: $e^{-\Lambda_s}\equiv\sqrt{1-R/R_s}$. If $D$ is the distance from the
172: star to the observer, the phase-averaged flux measured by an observer
173: at infinity (without including the reduction due to photoelectric
174: absorption by intervening material) is given by
175: \beq
176: f(E)=\frac{\pi R_\infty^2\;\sigma T^4_{p,\infty}}
177: {4\pi D^2}\frac{1}{k T_{p,\infty}}
178: \int_0^{2\pi}\frac{d\alpha}{2\pi}\int_0^1 2xdx\int_0^{2\pi}
179: \frac{d\phi}{2\pi}\;
180: I_0(\theta,\phi) \;n[Ee^{-\Lambda_s};T_{s}(\theta,\phi)]\;,
181: \label{eq:flux}
182: \eeq
183: in units of photons cm$^{-2}$ s$^{-1}$ keV$^{-1}$. Here $x=\sin\delta$
184: ($\delta$ being the angle between the normal to the NS surface and the
185: direction of the photon trajectory), $R_\infty\equiv Re^{\Lambda_s}$,
186: and $T_{p,\infty}\equiv T_{p}e^{-\Lambda_s}$, where $T_p$ is the
187: temperature at the pole. The general relativistic effects of light
188: deflection are taken into account through the ray-tracing function
189: (Page 1995)
190: \beq
191: \theta(\delta)=\int_0^{R_s/2R}x\;du\left/\sqrt{\left(1-\frac{R_s}{R}\right)
192: \left(\frac{R_s}{2R}\right)^2-(1-2u)u^2 x^2}\right.\;,
193: \label{eq:teta}
194: \eeq
195: where $u=R_s/2r$, with $r$ being the radial coordinate.
196: A photon emitted at an angle $\delta$
197: with respect to the normal to the surface comes from a colatitude
198: $\theta(\delta)$ on the star. The total flux at each point
199: $(\theta,\phi)$ of the NS surface is given by
200: \beq
201: I_0(\theta,\phi)=\frac{4\cos^2\theta_p}{3\cos^2\theta_p+1}\;
202: (0.75\;\cos^2\theta_p+0.25)^{0.2}\;.
203: \label{eq:I0}
204: \eeq
205: The first term in the right hand side is appropriate for a dipole, and
206: we have then assumed a further dependence of the flux on $B^{0.4}$, as
207: in Heyl \& Hernquist (1998b). Finally, the local temperature on the
208: stellar surface is
209: determined by
210: \beq
211: T_{s}(\theta,\phi)=T_p[I_0(\theta,\phi)]^{1/4}\;.
212: \label{tstar}
213: \eeq
214:
215: \section{Observations and Analysis}
216:
217: We analyzed archival observations of several AXPs and SGRs made by the {\em
218: Advanced Satellite for Cosmology and Astrophysics} ({\em ASCA};
219: Tanaka, Inoue, \& Holt 1994), obtained from the High Energy
220: Astrophysics Archival Research Center (HEASARC) at NASA Goddard Space
221: Flight Center. {\em ASCA} was launched in 1993 and continued to make
222: observations through 2000. It carried four identical grazing
223: incidence X-ray telescopes capable of imaging X-rays in the 0.5--10
224: keV range with a 24 arcmin (FWHM) field of view, a $\sim$1 arcmin
225: point spread function, and a total effective area of 1300~cm$^2$ at 1
226: keV. Each telescope had one dedicated focal plane instrument, and all
227: four instruments simultaneously recorded data for each observation.
228: The two CCD cameras (SIS0 and SIS1) had superior energy resolution
229: ($E/\Delta E\sim$20--50) and were sensitive down to about 0.5~keV.
230: The two gas scintillation imaging proportional counters (GIS2 and
231: GIS3) had more modest energy resolution ($E/\Delta E\sim$10) and
232: little sensitivity below 1 keV, but their effective area was
233: comparable to the CCDs around 2 keV and higher above $\sim$4 keV.
234: Thus, the GIS data were somewhat better suited for studying the
235: X-ray continuum spectrum of absorbed sources like the AXPs and SGRs.
236: However, we initially examined data from all four instruments in our
237: analysis.
238:
239: A summary of the {\em ASCA} observations that we analyzed is given in
240: Table~1. For simplicity, we confined ourselves to those sources where
241: there is no contaminating emission from a surrounding supernova
242: remnant to consider; this excludes the AXPs 1E 2259+586 and 1E 1841-045. We used the
243: standard screened events files provided through HEASARC by the {\em ASCA}
244: Guest Observer Facility (GOF). From these files, spectra were extracted
245: for the point sources. The extraction radii were 6 arcmin and 4arcmin for GIS and
246: SIS respectively. Background spectra were also extracted from the
247: event files from the area outside a 8 arcmin radius from the point source and
248: any other bright sources present in the observation. We fitted the
249: X-ray spectra for all the observations in Table~1 using the XSPEC
250: spectral analysis package (Arnaud 1996). In each case, we fit the
251: analytic magnetar model described in Equation~(\ref{eq:flux}). For
252: comparison, we also fit an ideal blackbody model (BB), assuming a uniform
253: temperature distribution over the emitting area. The free parameters of
254: the magnetar model are the pole temperature $T_p$ and the radius of the
255: star $R$. (For a non-relativistic star, the radius affects only the
256: overall normalization of the spectrum; however, when general
257: relativistic effects are taken into account, the radius also modifies
258: the shape of the spectrum through the dependence of the function
259: $\theta(x)$ in Equation~(\ref{eq:flux}) on $R$.)
260:
261: For both the magnetar and the BB models, we included the
262: multiplicative effect of interstellar photoelectric absorption
263: (Morrison \& McCammon 1983). We also allowed for the possibility of
264: an absorbed power law (PL) component, using the same absorption column
265: as for the thermal component. However, for each object, a fit using
266: only the absorbed thermal component alone was also attempted. A
267: spectrum produced by an atmosphere indeed has a harder tail compared
268: to simple blackbody emission at the same effective
269: temperature. Therefore, even if the blackbody fit always required an
270: extra power law component, this did not necessarily have to be the
271: case when processing by an atmosphere was included in the computation
272: of the spectrum.
273:
274: For each observation, we began by fitting the data for each instrument
275: individually, in order to evaluate the data quality separately. If the
276: data from one or more of the instruments had poor statistics (usually
277: in the case of the SIS data) or was otherwise problematic, it was
278: discarded. The remaining data were then fit jointly to maximize our
279: continuum sensitivity, with the overall normalization tied to that of
280: the GIS2 detector (which generally gave the best independent fits).
281: In the case of the SGR 1627$-$41 and SGR 1806$-20$, we found that
282: there were too few soft counts to allow a meaningful constraint on the
283: fit parameters for the thermal components. We do not consider these
284: sources further in this paper.
285:
286: The results of our fits are summarized in Table 2. For each source,
287: we show both the absorbed BB+PL fits and the absorbed magnetar+PL
288: fits. In the case of AX J1845-0258, however, the statistics were too poor
289: to constrain both components separately. For this object, the thermal
290: and non-thermal components in Table 2 represent alternate fits to
291: the same data. In a few cases (i.e. when the fit was acceptable), we
292: also show an absorbed magnetar-PL fit with the column density held
293: fixed at $N_{\rm H}=1\times 10^{22}$ cm$^{-2}$. For all our fits,
294: we also give
295: the fraction of the flux that is in the power law
296: component, computed in terms of the unabsorbed photon flux in the
297: 0.7--10 keV band. A comparison between the temperatures and radii
298: obtained in the BB fits and in the fits with the atmospheres is shown
299: in Figure 1 (except for the object RXS J1708$-$40, whose distance is still
300: largely uncertain, and for which a separate discussion will be made in
301: \S 4). The two dashed lines mark the region of NS radii which are
302: allowed by currently available models for the NS equation of state.
303:
304: \section{Discussion}
305:
306: We have analyzed archival {\em ASCA} data for several known AXPs and SGRs and
307: fitted their spectra with a model in which the X-ray emission
308: consists of thermal radiation from a highly magnetized neutron star,
309: as well as a power-law component at high energies. For the thermal
310: contribution, we have included distortions in the spectra due to the
311: presence of an atmosphere of light elements, accounted for the
312: anisotropic flow of heat through the envelope due to the intense
313: magnetic field, and included general relativistic corrections to the
314: observed flux. We find that the thermal emitting areas implied by our
315: model are always larger than those derived by spectral fits that use a
316: blackbody spectrum, and are generally consistent with those expected
317: for neutron stars. All the fits were made using a model with a
318: moderate beaming, i.e. $\beta\sim 0-1$. We have considered other
319: radiation patterns and find that a more intense beaming results in
320: slightly larger inferred areas and smaller temperatures, but the fits
321: are equally acceptable. Therefore, spectral fits alone made with the
322: phase averaged spectrum are not able to constrain the degree and type
323: of beaming, especially given the uncertainties in the other
324: parameters.
325:
326: In most of the cases we found that, despite the fact that processing
327: by an atmosphere leads to lower temperatures than those implied by the
328: BB fits, our inferred temperatures are still too high to be accounted
329: for by thermal cooling alone. However, if the magnetic field is
330: sufficiently strong and decaying, the energy from its decay may
331: augment the thermal emission from the surface (Heyl \& Kulkarni 1998).
332: The inferred surface temperatures of all of the objects with well
333: constrained fits are higher than would be expected for a neutron star
334: cooling through an iron envelope (Heyl \& Hernquist 2000); the results
335: for the AXP 4U 0142$+61$, RXS J1708$-$40 and SGR 1900$+$14
336: are marginally consistent with a neutron star cooling through a highly
337: magnetized iron envelope with a substantial contribution to the flux
338: from the decay of the magnetic field (Heyl \& Kulkarni 1998). If the
339: energy released by the magnetic field is deposited at high densities,
340: magnetic field decay alone cannot explain the high effective
341: temperatures of the AXPs 1E 1048.1$-5937$ and 1E 1845$-$0245. Thermal
342: emission through a light element envelope (Heyl \& Hernquist 1997b,
343: Potekhin, Chabrier \& Yakovlev 1997), however, can account for these
344: sources.
345:
346: Atmospheric effects play a crucial role in determining the emitted
347: spectrum (Romani 1987; Pavlov et al. 1994; Zampieri et al. 1995;
348: Rajagopal \& Romani 1996; Zavlin, Pavlov \& Shibanov 1996). For the
349: effective temperatures of these stars (i.e. $\sim$ a few $\times 10^6$
350: K), the spectra from hydrogen atmospheres depend only weakly on the
351: strength of the magnetic field (see e.g. Fig. 5 in Rajagopal, Romani
352: \& Miller 1997). However, the composition of the atmosphere may
353: dramatically affect the emitted spectrum (see, e.g. references above).
354: If the atmosphere were made of heavy elements such as iron, the
355: emitted spectrum would be much closer to a blackbody\footnote{Note
356: that the accretion model predicts that the opacity should be dominated
357: by heavy metals in SGRs and AXPs, because the settling time is
358: dominated by the rate of metal deposition for the accretion rates
359: implyed by the X-ray emission (Brown, Bildsten \& Rutledge 1998).}
360: (Rajagopal et al. 1997).
361:
362: Note that, given the mild dependence of the spectrum on the strength
363: of the magnetic field for $B\ga 10^{12}$ G, the model we have
364: developed could similarly be used to model the thermal emission
365: of neutron stars with magnetic fields $\sim 10^{12}- 10^{13}$ G and
366: light element atmospheres. However, in the case of the AXPs, the
367: additional heat (with respect to the predictions of standard cooling
368: scenarios) would be difficult to explain without the contribution from
369: magnetic field decay. A hypothetical contribution from accretion
370: would most likely result from matter channeled onto the magnetic
371: poles of the star by the presence of
372: the magnetic field lines. This would yield hot polar caps and the
373: total spectrum would show two thermal components, one at lower energy
374: being consistent with the total area of the star (due to thermal
375: cooling) and another at higher temperature but coming from a small
376: fraction of the star (due to the heated polar caps). Here
377: we find that spectral fitting requires only one thermal
378: component, and therefore this scenario does not seem to be
379: favored. However, it is also true that the high column density to
380: these sources would make a low-temperature thermal component difficult
381: to detect.
382:
383: The distances to the sources are still rather uncertain. As Figure 1
384: shows, relatively small variations with respect to the assumed values
385: would not affect any of the conclusions of this work. The value of the
386: inferred radius, $R$, roughly scales with the distance $D$. If any of
387: the distances turned out to be much smaller than the assumed value, so
388: that the radii were consequently much smaller than the
389: minimum allowed value by the NS equation of state, then an accretion
390: model where the X-ray emission is produced by a hot spot would be
391: favored with respect to the magnetar model, where it is produced by
392: the whole surface of the star. On the other hand, if any of the
393: distances turned out to be much larger than the assumed value, so that
394: the corresponding BB radius were already consistent with the emission
395: from the entire surface, then a light element atmosphere would not
396: be appropriate, as it would yield too large radii, while
397: an atmosphere made of heavy elements such as iron would be a viable
398: option, because it yields spectra very close to blackbody, as
399: discussed above. In such a case, it would be difficult to interpret
400: the X-ray emission as being powered by accretion unless the magnetic
401: field of the objects were very small, so that accretion could proceed
402: in a quasi-spherical fashion rather than being channeled through the
403: poles\footnote{Note that in models where accretion derives from
404: fallback disks (e.g. Chatterjee et al. 2000), a magnetic field of the
405: order of $10^{12}-10^{13}$ G is needed.}.
406:
407: Among all the objects that we considered, RXS J1708$-40$ is the one
408: which has the largest uncertainty in its distance. It lies in the
409: Galactic plane, and, along this direction, spiral arms are located at
410: a distance of $\sim 1$, $\sim 3$, and $\sim 4.5$ kpc (Taylor \& Cordes
411: 1993). The large column density to the source inferred from the X-ray
412: spectra suggests a likely distance in the 5-10 kpc range (Israel et
413: al. 1999). However, a smaller distance cannot be ruled out. Our
414: results suggest that this object, if it were a magnetar with a light
415: element atmosphere, would most likely reside in the middle spiral
416: arm. However, a magnetar with an atmosphere of heavy elements would be
417: consistent with the largest estimates of the distance.
418:
419: Besides the distance to the sources, the main uncertainty in the
420: inferred emitting areas from the spectral fits arises from the
421: uncertainty in the column density $N_{\rm H}$. This
422: conclusion is similar to
423: the findings of Rutledge et al. (1999) for their spectral fits of the
424: quiescent X-ray emission from accreting NS transients. This uncertainty may
425: be reduced if high signal-to-noise spectral data becomes available in a
426: passband that covers both the energy range where absorption is most
427: effective and where it is not. Such observations may be possible
428: with {\em Chandra X-Ray Observatory} and {\em XMM-Newton}.
429:
430: Despite the harder tail in the spectrum produced by reprocessing of
431: the outgoing flux through an atmosphere, spectral fits still require
432: in most cases a hard power law component
433: (even though its normalization is
434: often not well constrained). In the context of the magnetar model,
435: the origin of this power-law emission is not fully understood.
436: Thompson \& Duncan (1996) have proposed that the non-thermal
437: components of AXP and SGR spectra can be produced by magnetospheric
438: currents resulting from fracturing of the neutron star crust. An
439: important constraint on this type of model is provided by the fact
440: that the hard emission from AXPs appears to be pulsed, often with a
441: large amplitude. Although the angular distribution of the radiation
442: from these effects is not yet known, a magnetospheric origin for the
443: power-law component is appealing from the point of view of helping to
444: account for the observed pulsations, because gravitational bending
445: would be relatively less significant far from the stellar surface. A
446: preliminary analysis (C. Thompson, private communication) suggests
447: that photons produced by magnetospheric currents will preferentially
448: escape from regions near the magnetic poles, possibly resulting in a
449: large amplitude of pulsation. However, in order to be able to draw
450: firmer conclusions from a timing analysis, tighter limits on the power
451: law fraction $f_{\rm pl}$ are needed, possibly as a function of
452: energy. To such purpose, independent constraints on the column
453: densities would be highly desirable. In the fits in which $N_{\rm
454: H}$ is held fixed the normalization of the power law component is much
455: better constrained. Moreover, the column density itself influences the
456: inferred values of the pulsed fractions (Perna, Heyl \& Hernquist
457: 2000).
458:
459: In summary, our finding that plausible atmosphere models yield thermal
460: emitting areas consistent with a neutron star surface supports the
461: interpretation of AXPs and SGRs as magnetars. However, based on
462: currently available data, we cannot definitively discriminate between
463: the magnetar and accretion models for these objects. In particular,
464: the apparent requirement that a large fraction of the flux arises from
465: a hard power-law component significantly complicates efforts to infer
466: the true nature of these sources.
467:
468: A variety of objections have been raised against both the magnetar and
469: accretion models for AXPs and SGRs (e.g. Li 1999; Marsden, Rothschild
470: \& Lingenfelter 1999; DeDeo, Psaltis \& Narayan
471: 2000; Hulleman et al. 2000a; Marsden et al. 2001). In the case of the magnetar
472: interpretation, our results show that the thermal emission is
473: consistent with the surface area of a neutron star, provided that the
474: spectrum from this component is sufficiently distinct from a
475: blackbody. Moreover, the existence of a substantial non-thermal,
476: power-law component mitigates concerns that the large pulsed fractions
477: measured for some AXPs and SGRs might be inconsistent with the
478: magnetar hypothesis. For an object like 1E 1048.1-5937, for example, the
479: unusually large pulsed fraction of $\sim 70\%$ (e.g. Corbet \& Mihara
480: 1997; Oosterbroeck et al. 1998) may simply reflect the inferred
481: presence of a particularly important non-thermal process whose
482: contribution in this case approaches $\sim 80\%$ of the total flux.
483: It remains problematic, however, whether or not models
484: for the origin of the power-law emission in the context of
485: the magnetar model can account for such high relative fluxes.
486:
487: Perhaps the most challenging argument against the accretion model for
488: AXPs and SGRs comes from recent studies of optical and infrared
489: emission from these objects. In the work of Chatterjee et al. (2000),
490: for example, the presence of an extended disk is expected to yield
491: significant optical and infrared fluxes (e.g. Perna et al. 2000; Perna
492: \& Hernquist 2000), in disagreement with existing observational
493: measurements (e.g. Hulleman et al. 2000a,b). However, Menou et
494: al. (2001) have recently proposed an alternate accretion scenario
495: based on fallback in which the disk always remains geometrically thin
496: and radially compact and accretion is halted on timescales comparable
497: to the ages of AXPs and SGRs through a thermal ionization instability.
498: In this model, the optical and infrared emission from the disk is
499: greatly suppressed relative to that predicted by e.g. Chatterjee et
500: al. since the disk is limited in radial extent.
501:
502: Another argument against the accretion model for AXPs concerns the
503: unusually steady spindown of objects like RXS J1708$-40$ and 1E 2259$+586$
504: (Kaspi, Chakrabarty \& Steinberger 1999). Such behavior is not
505: characteristic of accreting, binary X-ray pulsars, but the
506: implication of this finding for isolated fallback disks is less
507: clear, particularly for models in which the disks are low in mass
508: (Menou et al. 2001) and considering the noisier behavior of AXP
509: 1E 1048.1$-$5937 (Kaspi et al. 2000).
510:
511: Thus, it appears that with existing observational constraints it is
512: not possible to conclusively rule out either the magnetar or accretion
513: scenario for AXPs and SGRs. In detail, however, the spectral energy
514: distributions expected in these two classes of models should exhibit
515: significant differences in many wavelength intervals. Hence, we are
516: optimistic that highly sensitive ongoing multi-wavelength observations
517: combining X-ray spectroscopy with deep optical and infrared searches
518: will eventually discriminate between the magnetar and accretion
519: pictures, definitively revealing the true nature of these sources.
520:
521: \acknowledgements{We thank Chris Thompson for enlightening discussion,
522: Jonathan McDowell for support with the XSPEC software, and an anonymous
523: referee for insightful comments.
524: Support for JSH was provided by the National Aeronautics and
525: Space Administration through Chandra Postdoctoral Fellowship Award
526: Number PF0-10015 issued by the Chandra X-ray Observatory Center, which
527: is operated by the Smithsonian Astrophysical Observatory for and on
528: behalf of NASA under contract NAS8-39073.
529: }
530:
531:
532: \begin{references}
533:
534: \reference{} Alcock, C., \& Illarionov, A. 1980, ApJ, 235, 534
535:
536: \reference{} Alpar, M. A. 1999, astro-ph/9912228
537:
538: \reference{} Alpar, M. A. 2000, astro-ph/0005211
539:
540: \reference{} Arnaud, K. A. 1996, in ASP Conf. Series 101, Astronomical
541: Data Analysis Software and Systems V, ed. G. Jacoby \& J. Barnes
542: (San Francisco: ASP), 17
543:
544: \reference{} Brown, E. F., Bildsten, L., \& Rutledge, R. E. 1998,
545: ApJL, 504, L195
546:
547: %\reference{} Chakrabarty, D., Pivovaroff, M.J., Hernquist, L.,
548: %Heyl, J.S. \& Narayan, R. 2000, ApJ, in press [astro-ph/0001026]
549:
550: \reference{} Chatterjee, P., Hernquist, L. \& Narayan, R. 2000, ApJ,
551: in press (CHN)
552:
553: %\reference{} Coe, M. J., Jones, L. R., \& Letho, H. 1994, MNRAS, 270, 178
554:
555: \reference{} Corbet, R. \& Mihara, T. 1997, ApJ, 475, L127
556:
557: \reference{} Corbet, R. et al. 1995, ApJ, 443, 786
558:
559: \reference{} DeDeo, S., Psaltis, D. \& Narayan, R., ApJ submitted, astro-ph/0004266
560:
561: \reference{} Duncan, R. C. \& Thompson, C. 1992, ApJ, 392, L9
562:
563: \reference{} Duncan, R. C. \& Thompson, C. 1995, in AIP
564: Conf. Proc. 366, High Velocity Neutron Stars and Gamma-Ray Bursts,
565: ed. R. E. Rothschild \& R. E. Lingenfelter (Woodbury: AIP), 111
566:
567: %\reference{} Fahlman, G. G. \& Gregory, P. C. 1981, Nature, 293, 202
568:
569: \reference{} Ghosh, P., Angelini, L. \& White, N. E. 1997, ApJ, 478, 713
570:
571: \reference{} Greenstein, G., \& Hartke, G. J. 1983, ApJ, 271, 283
572:
573: %\reference{} Greiveldinger, C. et~al. 1996, ApJ 465, L35.
574:
575: \reference{} Heyl, J. S. \& Hernquist, L. 1997a, ApJ, 491, L95
576:
577: \reference{} Heyl, J. S. \& Hernquist, L. 1997b, ApJ, 489, L67
578:
579: \reference{} Heyl, J. S. \& Hernquist, L. 1998a, MNRAS, 298, L17
580:
581: \reference{} Heyl, J. S. \& Hernquist, L. 1998b, MNRAS, 300, 599
582:
583: \reference{} Heyl, J. S. \& Hernquist, L. 2000, MNRAS, in press
584:
585: \reference{} Heyl, J. S. \& Kulkarni, S. R. 1998, ApJ, 506, L61
586:
587: \reference{} Hulleman, F., van Kerkwijk, M. H., Verbunt, F. W., \& Kulkarni, S. R.
588: 2000a, A\&A, 358, 60
589:
590: \reference{} Hulleman, F., van Kerkwijk, M. H. \& Kulkarni, S. R., Nature in
591: press,
592: preprint astro-ph/0011561
593:
594: \reference{} Hurley, K., et al. 1999, ApJ, 510, L107
595:
596: \reference{} Israel, G. L. et al. 1999, ApJ, 518, L110
597:
598: \reference{} Kaspi, V. Chakrabarty, D. \& Steinberger, J. 1999,
599: ApJ, 525, L33
600:
601: \reference{} Kaspi, V.M., Gavriil, F.P., Chakrabarty, D., Lackey, J.R.
602: \& Muno, M.P. 2000, ApJ, submitted (astro-ph/0011368)
603:
604:
605: \reference{} Kouveliotou, C. et al. 1999, ApJ, 510, L115
606:
607: \reference{} Li, X.-D. 1999, ApJ, 520, 271
608:
609: \reference{} Marsden, D., Rothschild, R. E., \& Lingenfelter, R. E.
610: 1999, ApJ, 520, L107
611:
612: \reference{} Marsden, D. Lingenfelter, R. E., Rothschild, R. E. \&
613: Higdon, J. C. 2001, ApJ in press (astro-ph/9912207)
614:
615: \reference{} Menou, K., Perna, R. \& Hernquist, L. 2001, in
616: preparation
617:
618: \reference{} Mereghetti, S. \& Stella, L. 1995, ApJ, 442, L17
619:
620: %\reference{} Mereghetti, S., Cremonesi, D., Feroci, M., \& Tavani, M.
621: %2000, astro-ph/0006249
622:
623: \reference{} Morrison, R., \& McCammon, D. 1983, ApJ, 270, 119
624:
625: \reference{} Oosterbroeck, T., Parmar, A.N., Mereghetti, S. \&
626: Israel, G.L. 1998, A\&A, 289, 837
627:
628: \reference{} Pandharipande, V. 1971, Nucl. Phys., A174, 641
629:
630: \reference{} Pandharipande, V. \& Smith, R. A. 1971, Nucl. Phys., A237, 507
631:
632: \reference{} Page, D. 1995, ApJ, 442, 273
633:
634: \reference{} Pavlov, G. G., Shibanov, Y. A., Ventura, J., \&
635: Zavlin, V. E. 1994, A\&A, 289, 837
636:
637: \reference{} Perna, R. \& Hernquist, L. 2000, ApJ, 544, L57
638:
639: \reference{} Perna, R., Hernquist, L. \& Narayan, R. 2000, ApJ,
640: 541, 344
641:
642: \reference{} Perna, R., Heyl, J. \& Hernquist, L. 2000, ApJ,
643: 538, L159
644:
645: \reference{} Potekhin, A. Y., Chabrier, G. \& Yakovlev, D. G. 1997,
646: A\&A, 323, 415.
647:
648: \reference{} Rajagopal, M., \& Romani, R. W. 1996, ApJ, 461, 327
649:
650: \reference{} Rajagopal, M., Romani, R. W. \& Miller, M. C. 1997,
651: ApJ, 479, 347
652:
653: \reference{} Romani, R. W. 1987, ApJ, 313, 718
654:
655: \reference{} Rutledge, R. E., Bildsten, L., Brown, E. F.,
656: Pavlov, G. G., \& Zavlin, V. E. 1999, 514, 945
657:
658: %\Reference{} Seward, F. D., Charles, P. A., \& Smale, A. P. 1986,
659: %ApJ, 305, 814
660:
661: %\reference{} Steinle, H., et al. 1987, Astroph, Space Sci. 131, 687
662:
663: \reference{} Tanaka, Y., Inoue, H., \& Holt, S.~S. 1994, PASJ, 46, L37
664:
665: \reference{} Taylor, J. H., \& Cordes, J. M. 1993, ApJ, 411, 674
666:
667: \reference{} Thompson, C. \& Duncan, R. C. 1996, ApJ, 473, 322
668:
669: \reference{} Torii, K., et al. 1998, ApJ, 503, 843
670:
671: \reference{} van Paradijs, J., Taam, R. E. \& van den Heuvel, E. P. J. 1995,
672: A\&A, 299, L41
673:
674: \reference{} White, N. E., Angelini, L., Ebisawa, K., Tanaka, Y., \& Ghosh, P. 1996,
675: ApJ, 463, L83
676:
677: \reference{} Zampieri, L., Turolla, R., Zane, S., \& Treves, A. 1995,
678: ApJ, 439, 849
679:
680: \reference{} Zavlin, V. E., Pavlov, G. G., \& Shibanov, Y. A. 1996, A\& A, 315, 141
681:
682: \end{references}
683:
684:
685: \clearpage
686:
687: \begin{deluxetable}{llc}
688: \tablewidth{0pt}
689: \tablecaption{{Log of \em ASCA} Observations Analyzed}
690: \tablehead{\colhead{Source} & \colhead{Date} & \colhead{Exposure (ks)}}
691: \startdata
692: \cutinhead{Anomalous X-ray pulsars}
693: 4U 0142$+$61 & 1998 August 21 & 37.7 \\
694: 1E 1048.1$-$5937 & 1998 July 26 & 117.7 \\
695: RXS J1708$-$40 & 1996 September 3 & 33.7 \\
696: AX J1845$-$0258 & 1993 October 12 & 89.0 \\
697: %1E 2259$+$586 & 1993 May 30 & 70.6 \\
698: \cutinhead{Soft $\gamma$-ray repeaters}
699: SGR 1627$-$41 & 1999 February 26 & 261.6 \\
700: SGR 1806$-$20 & 1993 October 10 & 108.7 \\
701: & 1993 October 20 & 133.3 \\
702: SGR 1900$+$14 & 1998 September 16 & 174.8 \\
703: \enddata
704: \end{deluxetable}
705:
706: \clearpage
707: \begin{deluxetable}{lccccccc}
708: \tablewidth{0pt}
709: \tablecaption{Spectral Fits}
710: \tablehead{
711: & & \multicolumn{2}{c}{Thermal component} &
712: \multicolumn{2}{c}{Power law component} & &
713: \\ \cline{3-4} \cline{5-6}
714: \colhead{Model} & \colhead{$N_{\rm H}$ ($10^{22}$ cm$^{-2}$)} &
715: \colhead{$kT$ (keV)} & \colhead{$R$ (km)} & \colhead{$\gamma$} &
716: \colhead{$C_1$\tablenotemark{a}} &
717: \colhead{$\chi^2_{\rm red}$/$N_{\rm dof}$}&\colhead{$f_{\rm pl}$\tablenotemark{b} $\;$(\%)} }
718: \startdata
719: \cutinhead{{\em 1E 1048.1$-$5937 (AXP)}, $d=10$ kpc\tablenotemark{(c)}}
720: BB+PL & 1.47$_{-0.2}^{+0.14}$ & 0.63$_{-0.04}^{+0.03}$ &
721: 1.8$_{-0.2}^{+0.3}$ & 3.9$_{-0.3}^{+0.5}$ &
722: 14$_{-10}^{+20}$& 0.89/771 & 80\\
723: Magnetar+PL & 1.62$_{-0.25}^{+0.34}$& 0.43$_{-0.02}^{+0.04}$& 15.2$_{-0.7}^{+4}$&
724: 4.9$_{-0.8}^{+1.2}$& 20$_{-9}^{+23}$& 0.89/771 & 78\\
725: Magnetar+PL & 1.0 (fixed)& 0.41$_{-0.01}^{+0.01}$& 16.5$_{-0.9}^{+1.3}$&
726: 3.4$_{-0.3}^{+0.3}$& 3.7$_{-0.5}^{+0.4}$& 0.91/772 & 45\\
727: \cutinhead{{\em 1E 1845$-$0258 (AXP)}, $d=8.5$ kpc\tablenotemark{(d)}}
728: BB & 5.16$_{-0.96}^{+1.08}$& 0.66$_{-0.06}^{+0.06}$& 2.0$_{-0.5}^{+0.8}$&
729: & & 0.94/177 & 0\\
730: PL only & 10.3$_{-1.5}^{+1.8}$& & &
731: 5.1$_{-0.6}^{+0.7}$& 974$_{-828}^{+43}$& 1.00/177 & 100\\
732: Magnetar & 6.3$_{-1.3}^{+2.1}$ & 0.41$_{-0.09}^{+0.07}$& 18$_{-7.2}^{+26}$&
733: & & 0.97/177 & 0\\
734: \cutinhead{{\em 4U 0142$+$61 (AXP)}, $d=1$ kpc\tablenotemark{(e)}}
735: BB+PL & 1.12$_{-0.08}^{+0.08}$& 0.41$_{-0.01}^{+0.02}$& 1.7$_{-0.3}^{+0.2}$&
736: 4.0$_{-0.1}^{+0.2}$& 333$_{-60}^{+64}$& 0.97/330 & 88\\
737: Magnetar+PL & 0.62$_{-0.14}^{+0.3}$& 0.28$_{-0.01}^{+0.01}$&
738: 16.1$_{-0.6}^{+0.9}$& 3.2$_{-0.6}^{+0.7}$& 44$_{-31}^{+130}$& 0.98/330 &35\\
739: Magnetar+PL & 1.0 (fixed)& 0.27$_{-0.01}^{+0.01}$& 14.4$_{-1.1}^{+1.0}$&
740: 4.0$_{-0.08}^{+0.06}$& 216$_{-10}^{+15}$& 0.99/331 & 75\\
741: \cutinhead{{\em RXS J1708$-$40 (AXP)}, $d=10$ kpc\tablenotemark{(f)}}
742: BB+PL & 1.47$_{-0.46}^{+0.5}$& 0.40$_{-0.08}^{+0.06}$& 11.5$_{-7}^{+6}$&
743: 2.8$_{-1}^{+0.7}$& 36.2$_{-30}^{+80}$& 0.82/128 & 73\\
744: Magnetar+PL & 1.24$_{-0.33}^{+0.66}$& 0.31$_{-0.11}^{+0.05}$&
745: 69$_{-33}^{+75}$& 2.1$_{-2}^{+1.4}$& 8.2$_{-8}^{+95}$& 0.93/128 &35\\
746: Magnetar+PL & 1.0 (fixed)& 0.34$_{-0.01}^{+0.02}$& 56$_{-8}^{+4}$&
747: 1.3$_{-1.5}^{+0.8}$& 1.7$_{-1.6}^{+7}$& 0.82/129 & 25\\
748: \cutinhead{{\em SGR 1900$+$14}, $d=5$ kpc\tablenotemark{(g)}}
749: BB+PL & 2.5$_{-0.3}^{+0.2}$& 0.53$_{-0.16}^{+0.08}$& 1.2$_{-0.7}^{+0.8}$&
750: 2.1$_{-0.2}^{+0.3}$& 105$_{-20}^{+80}$& 0.83/309 & 90\\
751: Magnetar+PL & 2.4$_{-0.3}^{+0.4}$& 0.34$_{-0.16}^{+0.05}$& 11.3$_{-0.3}^{+1.4}$&
752: 2.1$_{-0.3}^{+0.3}$& 110$_{-100}^{+70}$& 0.83/309 & 85\\
753: \enddata
754: \tablenotetext{a}{Power-law normalization at 1 keV in units of
755: $10^{-3}$ photons cm$^{2}$ s$^{-1}$ keV$^{-1}$.}
756: \tablenotetext{b}{Energy flux in the power law component for the 0.7-10 keV band.}
757: \tablenotetext{(c)}{Van Paradijs et al. (1995).}
758: \tablenotetext{(d)}{Torii et al. (1998).}
759: \tablenotetext{(e)}{Israel et al. (1999), but see also text.}
760: \tablenotetext{(f)}{White et al. (1996).}
761: \tablenotetext{(g)}{Hurley et al. (1999).}
762:
763: \end{deluxetable}
764:
765: \clearpage
766:
767: \begin{figure}[t]
768: \centerline{\epsfysize=5.7in\epsffile{fig1.ps}}
769: \caption{Inferred temperatures and radii for blackbody fits, and for fits
770: done with the magnetar model. The two dashed lines mark
771: the region of NS radii which are allowed by currently available models
772: for the NS equation of state for a NS of $M=1.4M_\odot$
773: (Pandharipande 1971; Pandharipande \& Smith 1975) .}
774: \label{fig:1}
775: \end{figure}
776:
777:
778:
779:
780:
781: \end{document}
782:
783: