1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[12pt,preprint]{emulateapj}
3:
4:
5: \begin{document}
6:
7: \shorttitle{Thermal X-rays from Millisecond Pulsars}
8: \shortauthors{Bogdanov et al.}
9:
10: \title{Thermal X-rays from Millisecond Pulsars: \\
11: Constraining the Fundamental Properties of Neutron Stars}
12:
13: \author{Slavko Bogdanov, Jonathan E. Grindlay, and George B. Rybicki}
14:
15: \affil{Harvard-Smithsonian Center for Astrophysics, 60 Garden Street,
16: Cambridge, MA 02138; \\ sbogdanov@cfa.harvard.edu, josh@cfa.harvard.edu}
17:
18:
19: \begin{abstract}
20:
21: We model the X-ray properties of millisecond pulsars (MSPs) by
22: considering hot spot emission from a weakly magnetized neutron star
23: (NS) covered by a hydrogen atmosphere. We investigate the
24: limitations of using the thermal X-ray pulse profiles of MSPs to
25: constrain the mass-to-radius ($M/R$) ratio of the NS. The accuracy
26: is strongly dependent on the viewing angle and magnetic inclination
27: but is ultimately limited only by photon statistics. We demonstrate
28: that valuable information regarding NSs can be extracted even from
29: data of fairly limited photon statistics through modeling of
30: archival observations of the nearby isolated PSRs J0030+0451 and
31: J2124--3358. The X-ray emission from these pulsars is consistent
32: with the presence of an atmosphere and a dipolar field
33: configuration. For both MSPs, the favorable geometry allows us to
34: place limits on the allowed $M/R$ of NSs. Assuming 1.4 M$_{\odot}$,
35: the stellar radius is constrained to be $R > 9.4$ km and $R > 7.8$
36: km (68\% confidence) for PSRs J0030+0451 and J2124--3358,
37: respectively. We explore the prospects of using future observatories
38: such as \textit{Constellation-X} and \textit{XEUS} to conduct X-ray
39: timing searches for MSPs not detectable at radio wavelengths due to
40: unfavorable viewing geometry. We are also able to place strong
41: constraints on the magnetic field evolution model proposed by
42: Ruderman. The pulse profiles indicate that the magnetic field of an
43: MSP does not have a tendency to align itself with the spin axis nor
44: migrate towards one of the spin poles during the low-mass X-ray
45: binary phase.
46:
47:
48: \end{abstract}
49:
50: \keywords{pulsars: general --- pulsars: individual (PSR J0030+0451,
51: PSR J2124--3358) --- stars: neutron --- X-rays: stars}
52:
53:
54: \section{Introduction}
55:
56: Recent X-ray studies have revealed that a number of known
57: rotation-powered millisecond pulsars (MSPs) exhibit predominantly
58: thermal soft X-ray emission
59: \citep{Grind02,Zavlin06,Bog06a,Zavlin07}. The infered effective
60: emission radii $R_{\rm eff}$ indicate that this radiation is localized
61: in regions on the neutron star (NS) surface much smaller than the
62: expected stellar radius ($R_{\rm eff}\ll R$) but comparable to the
63: classical radius of the pulsar magnetic polar cap $R_{pc}=(2\pi
64: R/cP)^{1/2}R$. This finding is in agreement with theoretical models of
65: pulsars, which indicate that the conditions in the magnetosphere of a
66: typical MSP favor heating of the polar caps to $\sim$$10^6$ K by a
67: return flow of energetic particles along the open magnetic field lines
68: \citep[see, e.g.,][for details]{Hard02,Zhang03}. As this heat is
69: restricted to a small fraction of the NS, study of the X-ray spectra
70: and pulse profiles of MSPs can reveal important information about the
71: star such as the radiative properties of the NS surface, magnetic
72: field geometry, and NS compactness ($R/R_S$, where
73: $R_S=2GM/c^2$). This approach, originally proposed by \citet{Pavlov97}
74: in the context of radio MSPs, can serve as a valuable probe of key NS
75: properties that are inaccessible by other observational means
76: (e.g.~radio pulse timing). As shown by \citet{Pavlov97},
77: \citet{Zavlin98}, and \citet{Bog07}, a model of polar cap thermal
78: emission from an optically-thick hydrogen (H) atmosphere provides an
79: excellent description of the X-ray pulse profiles of PSR J0437--4715,
80: the nearest known MSP. On the other hand, a blackbody model is
81: inconsistent with the X-ray timing data and can be definitively ruled
82: out. Furthermore, there is compelling evidence for a magnetic dipole
83: axis offset from the NS center \citep{Bog07}. Finally, the compactness
84: of PSR J0437--4715 is constrained to be $R/R_S>1.6$ (99.9\%
85: confidence). Thus, modeling of X-ray data of MSPs appears to be a
86: very promising approach towards answering long-standing questions
87: regarding the fundamental properties of MSPs and NSs, in general.
88:
89: The present paper represents an extension of the work described in
90: \citet{Bog07}. Herein we explore the detailed properties of the MSP
91: X-ray emission model with particular emphasis on its use for
92: constraints on the NS equation of state (EOS). The work is organized
93: as follows. In \S2 we examine the properties of our model; in \S3 we
94: discuss an application of our model to archival X-ray observations of
95: PSRs J0030+0451 and J2124--3358. In \S4 we present a discussion and
96: end with conclusions in \S5.
97:
98:
99:
100: %
101: % FIGURE 1
102: %
103: \begin{figure*}[t!]
104: \includegraphics[width=0.98\textwidth]{f1.ps}
105: \caption{(\textit{Left}) Representative synthetic hydrogen atmosphere
106: lightcurves for a rotating $M=1.4$ M$_\odot$, $R=10$ km NS with two
107: antipodal hot spots for the four lightcurve classes (I-IV, from top
108: to bottom, respectively), defined by Beloborodov (2002). The dashed
109: lines show the individual flux contribution from the two hot spots
110: while the solid line shows the total observed flux. All fluxes are
111: normalized to the value corresponding to a face-on hot spot
112: ($\alpha=\zeta=0$). Two rotational cycles are shown for
113: clarity. (\textit{Right}) Orthographic map projection of the MSP
114: surface for the four pulse profiles. The dashed line shows the
115: magnetic axis while the dotted line shows the line of sight to the
116: observer. The hatched region corresponds to the portion of the star
117: not visible to the observer.}
118: \end{figure*}
119:
120:
121:
122:
123: \section{MODEL ANALYSIS}
124:
125: The specifics of the theoretical model employed in this work are
126: outlined in \citet{Bog07}. In brief, the model considers a
127: relativistic rotating compact star with two identical X-ray emitting
128: hot spots, which in the case of MSPs (presumably) correspond to the
129: magnetic polar caps. The position of each spot relative to the
130: observer is given by the angle $\psi$ defined as
131: %
132: \begin{equation}
133: \cos\psi(t)=\sin \alpha \sin \zeta \cos \phi (t) + \cos \alpha \cos \zeta
134: \end{equation}
135: %
136: where $\alpha$ is the pulsar obliquity (i.e.~angle between the spin
137: and magnetic axes), $\zeta$ is the angle between the pulsar spin axis
138: and the line of sight to the distant observer, and $\phi(t)$ is the
139: spin phase. We take a non-rotating Schwarzschild metric as a
140: description of the space-time in the vicinity of the NS and include
141: Doppler boosting and propagation time delays. This formalism is
142: remarkably accurate as long as the spin period of the star is
143: $\gtrsim$3 ms where rotation-induced oblateness is unimportant
144: \citep{Cad07}. The NS surface is assumed to be covered by an
145: unmagnetized ($B<10^9$ G), optically-thick H atmosphere
146: \citep[][]{Zavlin96,McC04}. The angle-dependent emission pattern of
147: this atmosphere differs substantially from the isotropic one of a
148: blackbody, which is particularly evident in the rotation-induced
149: modulations of the X-ray flux \citep[see Fig.~1 of][]{Bog07}.
150:
151:
152: %
153: % FIGURE 2
154: %
155: \begin{figure*}[t]
156: \begin{center}
157: \includegraphics[width=0.82\textwidth]{f2.eps}
158: \end{center}
159: \caption{Lightcurves for a rotating $M=1.4$ M$_\odot$, $R=10$ km NS
160: with two antipodal hot spots with values of $\alpha$ and $\zeta$ as
161: in Figure 1. The \textit{dashed line} shows the idealized case of
162: point-like hot spots while the \textit{solid line} is for the case
163: of a filled circular hot spot (\textit{left} set of plots) and a
164: thin annulus (\textit{right} set of plots), both of radius 2 km and
165: $T_{\rm eff}=2\times 10^6$ K. The bottom panel for each class shows
166: the ratio between the two lightcurves shown in the upper panel. All
167: fluxes are normalized to the value corresponding to
168: $\alpha=\zeta=0$. Two rotational cycles are shown for clarity.}
169: \end{figure*}
170:
171:
172: Using this model we generate both synthetic spectra and pulse profiles
173: (Fig.~1) for the following parameters: the effective temperatures and
174: radii of the emission region(s) $T_{1}$, $T_{2}$, $R_1$ and $R_{2}$,
175: $\alpha$, $\zeta$, and $M/R$. Unless noted otherwise, we fix the mass
176: to the canonical NS value $M=1.4$ M$_{\odot}$ and vary only $R$. We
177: also allow for an off-center magnetic axis by including offsets in the
178: position of the secondary hot spot in latitude and longitude, $\Delta
179: \alpha$ and $\Delta \phi$, respectively, from the antipodal position.
180: The corresponding net offset of the secondary hot spot from the
181: antipodal position across the NS surface is\footnote{Note that
182: equation (12) in \citet{Bog07} is valid only if the angle $\alpha$ is
183: reckoned from the equator towards the spin pole. However, as defined
184: by convention, the pulsar obliquity $\alpha$ is measured from the spin
185: pole towards the equator. Thus, the correct expression for the
186: seondary hot spot offset is given by Equation (2) in this present
187: paper.}
188: %
189: \begin{equation}
190: \Delta s = R \cos ^{-1}[\cos\alpha\cos(\alpha+\Delta\alpha)+\sin\alpha\sin(\alpha+\Delta\alpha)\cos\Delta\phi]
191: \end{equation}
192: %
193: while the total displacement (i.e. impact parameter) of the magnetic
194: axis from the stellar center is
195: %
196: \begin{equation}
197: \Delta x = R\sin\left(\frac{\Delta s}{2R}\right)
198: \end{equation}
199: %
200: We conduct fits to both the spectrum and pulse profile. As pointed out
201: by \citet{Pavlov97}, this is of importance because the spectral fits
202: provide tight constraints on $T_{\rm eff}$ and $R_{\rm eff}$ but do
203: not provide useful information regarding the system geometry or
204: $M/R$. Conversely, the pulse profiles provide constraints on the
205: compactness and geometry of the NS but are less sensitive to the
206: parameters that define the emission spectrum ($T_{\rm eff}$ and
207: $R_{\rm eff}$).
208:
209: %
210: % FIGURE 3
211: %
212: \begin{figure}[t]
213: \begin{center}
214: \includegraphics[angle=270,width=0.49\textwidth]{f3.ps}
215: \caption{The location of pulse profile classes I--IV on the
216: $\alpha-\zeta$ plane for MSPs with $R=10$ (\textit{solid}), $12$
217: (\textit{dashed}), and $14$ (\textit{dotted}) km assuming $M=1.4$
218: M$_{\odot}$. See Figure 1 and text for a definition of each class.}
219: \end{center}
220: \end{figure}
221:
222: \subsection{Extended versus Point-like Emission}
223:
224: In \citet{Bog07} only point-like hot spots were considered, which was
225: sufficient for application to the available data for J0437--4715.
226: Here, we also consider more realistic extended hot spots in order to
227: ascertain the effect of the uncertain physical size and shape of the
228: hot polar caps on the observed pulse profiles. To accomplish this, we
229: consider a grid of emission spots across the NS surface, which allows
230: us to construct X-ray emission regions of arbitrary size and shape.
231: Figure 2 shows the resulting pulse profiles for two extreme cases: 1.~a
232: uniformly heated, circular polar cap of radius 2 km and 2.~a thin annulus
233: (with thickness much smaller than its radius) also with a 2 km
234: radius. The value $R_{pc}=2$ km corresponds to the expected polar cap
235: radius of a $P=5$ ms pulsar. Surprisingly, the physical extent and
236: exact shape of the emission regions do not significantly affect the
237: observed pulse profile, and only differs from the point-like case by
238: $\lesssim$1\%, except for class IV where it differs by as much as
239: $\sim$15\%. The largest differences occur at the turning points of
240: the lightcurves.
241:
242: In reality, due to the rotation of the pulsar the polar cap shape is
243: probably distorted \citep[see, e.g.,][]{Nar83,Biggs90}, with
244: elongation or compression in the meridional (north-south) direction on
245: the NS surface. However, even in such instances the differences from a
246: circular hot spot with the same effective area are relatively small
247: (few \%). This also means that for emission regions on the NS surface
248: comparable in area to that expected for a circular pulsar polar cap,
249: the X-ray pulse profiles do not provide useful information regarding
250: the details of the region geometry. Moreover, the weak dependence of
251: the pulse profile on the polar cap size implies that the distance to
252: the pulsar, which is covariant with the emission area through the flux
253: normalization ($R_{\rm eff}^2/D^2$), does not significantly affect the
254: pulse profile shape and pulsed fraction. Thus, unlike alternative
255: methods for measuring NS compactness (e.g., using quiescent low-mass
256: X-ray binaries), fits to the X-ray pulse profiles of MSPs are not
257: strongly affected by uncertainties in the distance. Finally, Figure 2
258: implies that the pulse profiles are weakly sensitive to broad-band
259: calibration uncertainties in the detector effective area as
260: well. Narrow-band deviations from the true effective area of the
261: instrument (for instance, near absorption edges) are also neglegible
262: as the pulse profiles cover relatively wide energy intervals (see
263: \S2.2).
264:
265: Note that the ambiguity in the exact shape of the polar caps may
266: ultimately limit the accuracy of the constraint on the desires NS
267: parameters using the approach described here. Nonetheless, for the
268: observed spectra of MSPs \citep[see, e.g.,][]{Zavlin06}, the effective
269: radius of the hotter emission region is found to be of order a few
270: hundred meters, which compared to the size of the star ($\sim$10 km)
271: is effectively point-like. Thus, the uncertainty in the true shape of
272: the emission region can be overcome by considering only the hotter
273: emission component.
274:
275:
276: %
277: % FIGURE 4
278: %
279: \begin{figure*}[t]
280: \begin{center}
281: \includegraphics[width=0.52\textwidth]{f4a.ps}
282: \includegraphics[width=0.43\textwidth]{f4b.ps}
283: \caption{(\textit{Left}) Simulated 1-Ms spectroscopic and timing
284: observation of PSR J0437--4715 with \textit{Constellation-X} XMS
285: assuming $R=12$ km and $M=1.4$ M$_{\odot}$. The choice of phase zero
286: is arbitrary. (\textit{Right}) Best fit confidence intervals for $R$
287: (assuming $M=1.4$ M$_{\odot}$) from pulse profile fits to simulated
288: 250-ks and 1-Ms \textit{Constellation-X} XMS observations of PSR
289: J0437--4715.}
290: \end{center}
291: \end{figure*}
292:
293:
294: \subsection{Limiting Accuracy of $M/R$ Constraints Using MSPs}
295:
296: It has been shown that combined X-ray spectroscopic and timing studies
297: of rotation-powered MSPs can potentially be used to infer $M/R$ of the
298: NS \citep{Pavlov97,Zavlin98,Bog07}. Here, we wish to determine
299: whether with further improvement in data quality, $M/R$ can be
300: determined to significantly better accuracy. This is essential if
301: this method is to be used for reliable measurements of the NS
302: compactness and ultimately the NS EOS. In addition, it is important to
303: ascertain how the viewing and magnetic geometries affect the
304: constraints on the desired NS parameters. For this purpose, we have
305: carried out a series of fits to simulated X-ray data of MSPs with
306: greatly improved photon statistics than currently available. A major
307: advantage of the greatly increased data quality is the possibility of
308: high signal-to-noise ratio energy-resolved pulse profiles in multiple
309: energy bands and phase-resolved spectroscopy for several phase
310: windows. As an illustrative example, we consider the current version
311: of the proposed effective area curve of the \textit{Constellation-X}
312: observatory X-ray Microcalorimeter Spectrometer (XMS)
313: detector\footnote{See http://constellation.gsfc.nasa.gov/} and assume
314: a range of exposure times, spanning from 250 ks to 1 Ms. We generate
315: representative lightcurves for each of the four distinct types defined
316: by \citet{Belo02} based on the visibility of the two antipodal hot
317: spots (see Figs.~1 and 3)\footnote{Note that these are defined for the
318: antipodal hot spot case, but the regions are similar for offset hot
319: spots for small displacements.}, assuming both $R=10$ km and $M=1.4$
320: $M_{\odot}$, representative of strange equations of state, and $R=12$ km
321: and $M=1.4$ $M_{\odot}$ expected from conventional nucleonic equations of state
322: \citep{Latt01}.
323:
324: Class I pulse profiles occur in cases when only the primary hot spot
325: is observable at all times, while the secondary hot spot is in the
326: invisible portion of the star for the entire period (see Fig.~1).
327: Note that for two antipodal polar caps, this can only be achieved for
328: $R/R_S>1.7$ (or $R>7$ km for $M=1.4$ M$_{\odot}$) since for a more
329: compact star the entire stellar surface is always visible. The
330: visibility of only a single hot spot causes large uncertainties
331: regarding the true geometry and especially the compactness. This
332: occurs due to the fact that that visible hot spot is always found at a
333: small angle with respect to the line of sight where light bending
334: effects are less pronounced, resulting in weak dependence on $M/R$. As
335: a result a given class I pulse profile can be reproduced by a fairly
336: large set of combinations of $\alpha$, $\zeta$, and $M/R$. We find
337: that this problem cannot be overcome by deeper observations. Indeed,
338: for simulated deep exposures ($\sim$1 Ms) with
339: \textit{Constellation-X} the $90\%$ confidence limits alone encompass
340: the entire range of plausible NS radii ($7-16$ km for an assumed 1.4
341: $M_{\odot}$). Therefore, this class of pulse profiles are not
342: suitable for tight constraints on $M/R$.
343:
344: Pulse profiles that occupy region II in Figure 3 exhibit a single
345: broad pulse per rotation. In this configuration, the primary hot spot
346: is observable at all times, while the secondary is only visible for a
347: portion of the spin period. The thermal X-ray pulse profile of PSR
348: J0437--4715 \citep{Zavlin02,Zavlin06} appears to be in this class.
349: Figure 4 shows simulated observations of PSR J0437--4715, assuming
350: $R=12$ km and $M=1.4$ M$_{\odot}$, with \textit{Constellation-X} XMS
351: and the corresponding best fit confidence intervals for $R$. It is
352: apparent that with a substantial improvement in photon statistics, $R$
353: could, in principle, be constrained to better than $10$\%, and
354: potentially down to $\sim$2--3\% (at 90\% confidence) given sufficient
355: exposure time. Similar constraints are obtained for $R=10$ km and
356: $M=1.4$ M$_{\odot}$ as well. Combined with an independent mass
357: measurement from radio timing observations, this method could lead to
358: unprecendented constraints on the NS EOS.
359:
360: Thermal X-ray lightcurves that are found in region III (Fig.~2) are
361: characterized by two pulse peaks per rotation period. The nearby PSR
362: J0030+0451 is a good example of this class \citep[][see also
363: \S3.1]{Beck00,Beck02}. An advantage of these pulse profiles is that
364: each of the hot spots provides a dominant contribution to the flux of
365: one of the two pulses. This feature has several practical
366: consequences. First, it may, in principle, allow one to determine
367: whether the two polar caps are identical in terms of size and
368: temperature. Any significant differences would point to deviations
369: from a centered dipole, such as displacements of the dipole along the
370: axial (magnetic north-south) direction or small-scale multipole
371: contributions. Moreover, the ``sharpness'' of the two pulses is
372: strongly affected by $M/R$ \citep[see Fig.~3 of][]{Bog07} resulting in
373: increased sensitivity to the stellar compactness. Thus, class III
374: appears to be favorable for tight constraints on $M/R$. We find that a
375: 1 Ms simulated \textit{Constellation-X} observation of PSR J0030+0451
376: permits constraints on $M/R$ down to a $\sim$5\% level (at 90\%
377: confidence).
378:
379: Finally, class IV profiles are formed when both hot spots are
380: observable at all times. As evident in Figure 1, these pulse profiles
381: tend to have significantly lower fluxes relative to classes I-III
382: since the hot spots are observed close to edge-on for the entire spin
383: period. These pulsars may not be observable at radio wavelengths due
384: to the large impact angle $|\alpha-\zeta|$, which may exceed the
385: opening half-angle of the radio emission cone. Such object can only be
386: detected in X-rays. At present, identifying such objects as MSPs
387: through X-ray timing is difficult without a known radio counterpart,
388: although this may be possible with future X-ray observatories (see
389: \S4.1). These factors make class IV MSPs the most difficult to study
390: observationally. Most importantly, as seen in Figure 2, these
391: lightcurves are much more sensitive to the size and shape of the hot
392: spots. This introduces much larger uncertainties (roughly an order of
393: magnitude greater) than for classes I--III into the spectral and
394: lightcurve fits, resulting in weaker constraints on $M/R$.
395:
396: Thus, class II and III pulse profile seem to be most favorable for
397: tight constraints on $M/R$. Fortunately, these two classes occupy a
398: major portion of the $\alpha-\zeta$ plane\footnote{The same holds true
399: in the $\cos\alpha-\cos\zeta$ space.} for the plausible range of
400: $M/R$, implying that most MSPs should fall in these classes. In
401: \citet{Bog07} we conducted a detailed study of PSR J0437--4715, a
402: class II MSP. In \S3, we focus our analysis on two MSPs that exhibit
403: class III pulse profiles.
404:
405:
406:
407:
408:
409:
410:
411:
412:
413: \section{Application}
414:
415: As seen in \S2, for favorable geometries stringent constraints on MSP
416: parameters are possible. This suggests that even with fairly limited
417: photon statistics some useful insight into the MSP properties can be
418: obtained. Below, we illustrate this through application of our model
419: to X-ray observations of two nearby MSPs, PSRs J0030+0451 and
420: J2124--3358. Along with J0437--4715, these are the only thermal MSPs
421: for which the X-ray pulse profiles are of sufficient quality to allow
422: meaningful constraints on NS parameters, especially $M/R$. The
423: spectral analysis for each MSP was performed in
424: XSPEC\footnote{http://heasarc.gsfc.nasa.gov/docs/xanadu/xspec/} 12.3.0
425: using the model described above. As shown in
426: \citet{Pavlov97},\citet{Zavlin98}, and \citet{Bog07}, even the
427: phase-integrated spectra of MSP are significantly affected by the
428: choice of $\alpha$ and $\zeta$ due to the energy-dependent
429: limb-darkening of the atmosphere so it is necessary to consider them
430: in the spectral fits as well. In order to apply the model to the X-ray
431: pulse profiles, it was first convolved with the appropriate instrument
432: response, the encircled energy fraction was taken into account, and
433: the sky and detector background were added. The fit was performed by
434: searching the $\chi^2$ hyperspace for the minimum. We fit the pulse
435: profiles by considering the following parameters: the two temperatures
436: and effective radii of each hot spot ($T_1$, $T_2$, $R_1$, and $R_2$)
437: the two angles $\alpha$ and $\zeta$, the stellar radius $R$, the
438: offsets of the secondary hot spot from the antipodal position ($\Delta
439: \alpha$ and $\Delta \phi$), and the hydrogen column density along the
440: line of sight ($N_{\rm H}$). Unless noted otherwise, in our analysis
441: we will assume a fixed mass of $M=1.4$ M$_{\odot}$ and allow $R$ to
442: vary within the range of plausible NS radii for this mass
443: \citep[$7-15$ km; see, e.g.,][]{Latt01}.
444:
445: %
446: % FIGURE 5
447: %
448: \begin{figure}[t]
449: \begin{center}
450: \includegraphics[width=0.49\textwidth]{f5.ps}
451: \caption{(\textit{top}) \textit{XMM-Newton} EPIC-MOS1/2 phase
452: integrated spectra of J0030+0451 fitted with a two temperature
453: hydrogen atmosphere thermal model. The bottom panel shows the best
454: fit residuals. (\textit{bottom}) \textit{Chandra} ACIS-S and
455: \textit{XMM-Newton} EPIC-MOS1/2 phase integrated spectra of PSR
456: J2124--3358 fitted with a two temperature hydrogen atmosphere
457: model.}
458: \end{center}
459: \end{figure}
460:
461:
462: %
463: % FIGURE 6
464: %
465: \begin{figure}[t]
466: \begin{center}
467: \includegraphics[width=0.49\textwidth]{f6.ps}
468: \caption{(\textit{a}) \textit{XMM-Newton} EPIC-pn pulse profile of
469: PSR J0030+0451 for $0.3-2$ keV. The solid line shows the best fit
470: model. (\textit{b}) \textit{XMM-Newton} EPIC-pn pulse profile
471: of PSR J2124--3358 for $0.3-2$ keV. The solid line shows the best
472: fit model, while the dashed line shows the best fit for a centered
473: dipole (see text for best fit parameters).}
474: \end{center}
475: \end{figure}
476:
477:
478: \subsection{PSR J0030+0451}
479:
480: This 4.86-ms solitary pulsar was discovered at radio wavelengths in
481: the Arecibo drift scan survey \citep{Lom00}. It was subsequently
482: detected with \textit{ROSAT} during the final days of this mission.
483: Due to the unstable behavior of the failing PSPC detector, this data
484: provided no reliable spectral information but revealed a double peaked
485: pulse profile with pulsed fraction $\sim$50\% \citep{Beck00}. PSR
486: J0030+0451 was revisited in 2001 June 19 by \textit{XMM-Newton} for 31
487: ks \citep{Beck02}. This observation revealed a 0.3--2 keV spectrum
488: qualitatively similar to that of J0437--4715. In addition, the EPIC-pn
489: X-ray pulse profile shows a relatively high pulsed fraction
490: ($\sim$50\% for 0.3--2 keV) with two prominent pulses. \citet{Lom00}
491: have estimated from radio polarization measurements two possible
492: pulsar geometries: $\alpha=8^{\circ}$ and $\beta=1^{\circ}$ or
493: $\alpha=62^{\circ}$ and $\beta=10^{\circ}$, where $\beta$ is the
494: angular separation of the line of sight with respect to the magnetic
495: axis at closest approach. As the former combination of angles cannot
496: produce the observed double-peaked X-ray profile for all plausible
497: choices of $M/R$ (see Figs. 1 and 3), even if we allow for substantial
498: offsets of the secondary hot spots in $\alpha$ and $\phi$,
499: $\alpha=8^{\circ}$, $\beta=1^{\circ}$ can be definitively ruled
500: out. This result implies that the observed radio interpulse
501: \citep{Lom00} most likely originates from the antipodal radio emission
502: cone, not from the same emission cone as the main pulse.
503:
504: Figure 5 shows the \textit{XMM-Newton} EPIC-MOS1/2 spectra of PSR
505: J0030+0451. They are well described by two thermal components. Fitting
506: a hydrogen atmosphere model to the spectrum yields
507: $T_{1}=(1.3-2.1)\times 10^6$ K, $R_1=0.01-0.54$ km,
508: $T_{2}=(0.3-0.7)\times 10^6$ K, $R_2=0.8-3.5$ km, and $F_X=1.5 \times
509: 10^{-13}$ ergs s$^{-1}$ (0.3-2 keV), for assumed $N_{\rm
510: H}=(1-3)\times10^{20}$ cm$^{-2}$, $M=1.4$ M$_{\odot}$, $R=7-15$ km,
511: and all combinations of $\alpha$ and $\zeta$ in the range
512: $0^{\circ}-90^{\circ}$. The ranges quoted are 1$\sigma$ limits. As
513: expected, the spectral continuum is fit equaly well with a blackbody
514: model. \citet{Beck02} have suggested a broken power-law as a possible
515: alternative interpretation of the spectrum of J0030+0451. However, as
516: pointed out by \citet{Zavlin07}, this model results in unrealistic
517: values of $N{_H}$ and is thus unlikely.
518:
519: The fit to the pulse profile (Fig.~6a) was carried out by fixing
520: $N_{\rm H}=2\times10^{20}$ cm$^{-2}$ and taking a distance of $D=300$
521: pc \citep{Lom06}. For $M=1.4$ M$_{\odot}$, the radius is constrained
522: to be $R>9.4$ km or, more generally, $R/R_S>2.3$ (68\% confidence) for
523: all combinations of the other free parameters, with a best fit
524: $\chi_{\nu}^2=0.81$ (for 3 degrees of freedom). Acceptable fits were
525: obtained for radii up to $\sim$20 km. In the case of a blackbody
526: model, we find that good fits to the X-ray pulse profile require
527: implausably large stellar radii ($\gtrsim$20 km). As no NS EOS models
528: predict radii exceeding $\sim$15 km for $M=1.4$ M$_{\odot}$
529: \citep[see][]{Latt01}, the validity of the blackbody interpretation is
530: very doubtful. The lower limit on the allowed values of $\alpha$ and
531: $\zeta$ is found to be $\gtrsim$44$^{circ}$ (68\% confidence),
532: obtained when the other angle is equal to 90$^{\circ}$. Namely if
533: $\alpha=90^{\circ}$ then $\zeta=44^{\circ}$ or if $\alpha=44$ then
534: $zeta=90^{\circ}$. Finally, the currently available data is consistent
535: with both a centered and a displaced dipole field.
536:
537:
538:
539: \subsection{PSR J2124--3358}
540:
541: PSR J2124--3358 is a nearby ($D\approx 250$ pc), isolated pulsar with
542: $P=4.93$ ms \cite[][]{Bail97} first observed in X-rays with the
543: \textit{ROSAT} HRI \citep{Beck99}. As the HRI provided no useful
544: spectral information only a pulse profile was obtained with pulsed
545: fraction $\sim$33\%. J2124--3358 was later observed with
546: \textit{Chandra} ACIS-S for 30.2 ks and with \textit{XMM-Newton}
547: EPIC-MOS1/2 and EPIC-pn for 68.9 and 66.8 ks, respectively
548: \citep{Zavlin06,Hui06}. As shown by \citet{Zavlin06}, the X-ray
549: emission from J2124--3358 is also well described by a two-temperature
550: thermal spectrum. Fitting our hydrogen atmosphere model to the
551: phase-integrated spectral continuum yields $T_{1}=(1.3-2.4) \times
552: 10^6$ K, $R_1=0.03-0.5$ km, $T_{2}=(0.3-0.8) \times 10^6$ K,
553: $R_2=0.9-3.1$ km, and $L_X=1.8 \times 10^{30}$ ergs s$^{-1}$ (0.3--2
554: keV), for assumed $D=250$ pc, $N_H=1\times10^{20}$ cm$^{-2}$, $M=1.4$
555: M$_{\odot}$, $\alpha=0^{\circ}-90^{\circ}$,
556: $\zeta=0^{\circ}-90^{\circ}$, and $R=7-15$ km. The uncertainties
557: quoted represent $\pm$1$\sigma$ ranges. The derived values are
558: consistent with the results by \citet{Zavlin06}. Note that the diffuse
559: X-ray emission detected around this MSP \citep{Hui06} does not
560: contribute appreciably to the point source MSP emission as its total
561: luminosity $L_X\sim 1\times 10^{29}$ ergs s$^{-1}$ (0.1--2.4 keV) is
562: negligible.
563:
564: The X-ray pulse profile of this MSP (Fig.~6b) exhibits marginal
565: evidence for a faint secondary peak. Given that this feature is
566: evident in both the \textit{ROSAT} PSPC \citep{Beck99} and
567: \textit{XMM-Newton} EPIC-pn \citep{Zavlin06} pulse profile, it is very
568: likely genuine. The fit to the pulse profile (lower panel of Fig.~6b)
569: was carried out by fixing $N_{\rm H}=1\times10^{20}$ cm$^{-2}$
570: \citep{Zavlin06} and assuming the dispersion measure derived distance
571: of $D=250$ pc. Assuming $M=1.4$ M$_{\odot}$, the radius is constrained
572: to be $R>7.8$ km (68\% confidence), with best fit $\chi_{\nu}^2=$ (for
573: 3 degrees of freedom). The angles $\alpha$ and $\zeta$ are constrained
574: to be $\ge$12$^{\circ}$ (68\% confidence) for the other angle equal to
575: $90^{\circ}$ (i.e.~if $\alpha=90^{\circ}$ then $\zeta=12^{\circ}$ or if
576: $\alpha=12$ then $zeta=90^{\circ}$). Although the suggestive asymmetry
577: of the pulse profile hints at the presence of an offset dipole, the
578: poor photon statistics permit a centered dipole configuration. As with
579: PSR J0030+0451, a blackbody model requires unrealistically large
580: stellar radii ($\gtrsim$20 km for M=1.4 M$_{\odot}$) for
581: J2124--3358. Thus, we conclude that a blackbody model does not provide
582: a valid description of the surface emission properties of PSR
583: J2124--3358 as well.
584:
585:
586: \section{DISCUSSION}
587:
588:
589: %
590: % FIGURE 7
591: %
592: \begin{figure}[t]
593: \begin{center}
594: \includegraphics[width=0.45\textwidth]{f7.eps}
595: \caption{Contours of constant pulsed fraction(\textit{solid lines})
596: for X-ray H atmosphere emission from two antipodal hot spots in the
597: 0.3--2 keV band as a function of the pulsar obliquity ($\alpha$) and
598: viewing angle ($\zeta$) for a 1.4 M$_{\odot}$ MSP with radius 10,
599: 12, and 14 km (from \textit{top} to \textit{bottom},
600: respectively). The X-ray emission spectrum is taken to be that of
601: PSR J0437--4715. The diagonal \textit{dot-dashed}, \textit{dashed}
602: and \textit{dotted} lines show $10^{\circ}$, $30^{\circ}$, and
603: $60^{\circ}$ opening half-angles of the radio emission cone,
604: respectively. All MSPs found between each pair of lines are
605: detectable at radio frequencies (see text for details).}
606: \end{center}
607: \end{figure}
608:
609:
610:
611: \subsection{Searches for Radio Quiet MSPs}
612:
613: The effect of light bending combined with the (nearly) antipodal
614: configuration of the two MSP hot spots ensure that the thermal
615: radiation is observable at Earth for any combination of $\alpha$ and
616: $\zeta$. In contrast, at radio frequencies a pulsar is not observable
617: if $|\alpha-\zeta|$ exceeds the opening half-angle $\rho$ of the radio
618: emission cone. This brings forth the intriguing prospect of detecting
619: and identifying such radio quiet MSPs in X-rays using blind pulsation
620: searches. With the currently abailable X-ray observatories
621: (\textit{Chandra} and \textit{XMM-Newton}) this endeavor is rather
622: difficult due to the intrinsic faintness of MSPs \citep[see,
623: e.g.,][]{Cam07} and their relatively low X-ray pulsed fractions
624: ($\lesssim$50\%). On the other hand, for the next generation of large
625: X-ray observatories (\textit{Constellation-X} and \textit{XEUS}), the
626: proposed the $\gtrsim$10-fold increase in sensitivity makes such a
627: survey quite feasible.
628:
629: Figure 7 shows the X-ray pulsed fraction of a 1.4 M$_{\odot}$ MSP with
630: a 10, 12, and 14 km radius as a function of $\alpha$ and $\zeta$,
631: assuming a thermal emission spectrum like that of PSR J0437--4715
632: \citep{Zavlin06,Bog07}. Also shown are lines delineating the region
633: of the $\alpha-\zeta$ plane for which a pulsar with a given radio
634: emission cone width is observable at radio frequencies. Note that the
635: true opening angle of the radio cone for a given MSP is not known and
636: difficult to measure reliably but could be as high as
637: $\sim$$60^{\circ}$ \citep[see][]{Kra98}. If we assume a uniform
638: distribution of pulsar obliquities ($\alpha$) and viewing angles
639: ($\zeta$), for $\rho\lesssim 30^{\circ}$ a substantial portion
640: ($\sim$45\%) of the MSP population is invisible to us in the radio.
641: On the other hand, if we consider an X-ray timing survey with a
642: limiting pulsed fraction sensitivity of $\sim$10\%, only $\sim$5--20\%
643: (depending on $M/R$) of the MSPs will go undetected as pulsed sources
644: though they will still be detected as X-ray sources (provided they are
645: not heavily absorbed). The Galactic population of MSPs may in fact be
646: preferentially clustered in a certain range of $\alpha$ due to the
647: poorly understood effects of the accretion and magnetic field
648: reduction processes during the LMXB phase on the NS \citep[see,
649: e.g.,][for a review]{Bhatt91}. A deep X-ray timing survey of nearby
650: ($\lesssim$1--2 kpc) MSPs may, in principle, reveal whether this is
651: indeed the case.
652:
653: The detectability of the thermal polar cap emission from MSPs for all
654: combinations of $\alpha$ and $\zeta$ further suggests that the
655: \textit{entire} MSP population of certain Galactic globular clusters
656: \citep[e.g. 47 Tucanae, see][]{Grind02,Bog06a,Cam07} could be detected
657: using high angular and temporal resolution X-ray imaging and timing
658: (e.g.~with the aid of the proposed \textit{Generation-X}
659: observatory). Identifying the whole population of MSPs in a cluster
660: has important implication for studies of globular cluster evolution
661: and internal dynamics \citep[see, e.g.,][and references
662: therein]{Camilo05}.
663:
664:
665:
666: \subsection{Constraining Magnetic Field Evolution Models}
667:
668: As shown in \citet{Bog07} and in this present paper, the morphology of
669: the thermal X-ray pulse profiles of MSPs offer useful insight into the
670: magnetic field geometry of the pulsar (Fig.~1). This has important
671: implications for pulsar magnetic field evolution models. For instance,
672: Ruderman (1991; see also Chen \& Ruderman 1993 and Chen et al.~1998)
673: has postulated the presence of crustal ``plates'' on the NS surface
674: formed by shear stresses on the crust caused by neutron superfluid
675: vortex lines pinned to lattice nuclei (see Fig.~3 of Chen et
676: al.~1998). The motion of these plates would cause the magnetic fields
677: of MSPs to migrate across the stellar surface, resulting in either an
678: aligned magnetic field or one ``pinched'' at the spin pole. However,
679: these configurations would result in very little ($\lesssim$few
680: percent) modulation of the thermal X-ray flux due to the close
681: alignment of the polar caps with the spin axis, regardless of the
682: viewing geometry. This is at odds with the observed thermal X-ray
683: pulsed fractions of PSRs J0437--4715, J0030+0451, and J2124--3358,
684: which are in the range 30--50\%. Thus, although the model of Ruderman
685: (1991) can reproduce the observed radio properties of MSPs, it is
686: inconsistent with the observed thermal X-ray pulse profiles of
687: MSPs. In particular, the X-ray pulse profiles of PSRs J0437--4715,
688: J0030+0451 and J2124--3358 indicate that the magnetic axes of these
689: MSPs are significantly misaligned from the spin axis. This, in turn,
690: implies that the dipole fields of these (and likely all) MSPs do not
691: have a tendency to align with the spin axis nor migrate towards one of
692: the spin poles.
693:
694:
695:
696:
697:
698: \section{CONCLUSIONS}
699:
700: We have examined the properties of our model of thermal emission from
701: hot spots on the surface of a neutron star covered by a hydrogen
702: atmosphere, relevant for MSPs. Our investigation has demonstrated
703: that energy-resolved modeling of the thermal X-ray pulse profiles and
704: phase-resolved spectroscopy of the continuum emission can, in
705: principle, be used to determine $M/R$ and the pulsar geometry to high
706: accuracy, given observational data of sufficient quality and favorable
707: values of $\alpha$ and $\zeta$. As shown in \S2.1 the thermal pulse
708: profiles are surprisingly insensitive to the details of the polar cap
709: size and shape, the distance to the pulsar, and the uncertainty in the
710: instrument effective area. This method represents the only feasible
711: approach of studying MSP magnetic field, surface properties, and
712: compactness, especially for isolated MSPs. Barring any deleterious
713: effect such as additional hidden spectral components \citep[see,
714: e.g,][and references therein]{Bog06b} this approach can, in principle,
715: lead to unprecedented insight into NS properties.
716:
717:
718: Our model is found to be in agreement with the observed emission from
719: the nearby solitary MSPs J0030+0451, and J2124--3358. As with PSR
720: J0437--4715 \citep{Bog07}, the relatively large pulsed fractions
721: observed in PSRs J0030+0451 and J2124--3358 require the existence of a
722: light-element atmosphere on the stellar surface and cannot be
723: reproduced by a blackbody model for realistic NS radii. For
724: J0030+0451 and J2124--3358 we are able to place interesting limits on
725: the allowed compactness of $R>9.4$ km and $R>7.8$ km (68\% confidence)
726: assuming $M=1.4$ M$_{\odot}$. Based on our findings in \S2, we expect
727: deeper observations of this MSP to lead to much tighter constraints on
728: $M/R$, which in turn may firmly rule out certain families of NS EOS. The
729: available thermal X-ray pulse profiles also provide useful constraints
730: on magnetic field models of MSPs. Specifically, the positions of the
731: magnetic polar caps on the NS surface implied by the X-ray data
732: indicate that the magnetic field closely resembles the conventional
733: oblique dipole model of pulsars rather than more exotic field
734: configurations.
735:
736: The success of this approach towards elucidating crucial NS properties
737: motivates further studies of the nearby sample of MSPs with both
738: \textit{Chandra} and \textit{XMM-Newton}. Moreover, this makes MSPs
739: particularly important targets for upcoming X-ray mission such as
740: \textit{Constellation-X} and \textit{XEUS}. The great increase in
741: throughput of these facilities will allow searches for new MSPs,
742: detailed observations of a larger sample of known radio MSPs, and
743: unprecedented constraints on key NS properties, especially the NS EOS.
744:
745: \acknowledgements We would like to thank Ramesh Narayan, Bryan
746: Gaensler, Deepto Chakrabarty, Pat Slane, and Alice Harding for
747: numerous useful suggestions. This work was funded in part by NASA
748: \textit{Chandra} grant G07-8033A. The research presented has made use
749: of the NASA Astrophysics Data System (ADS).
750:
751: %\facilities{CXO,XMM-Newton}
752:
753: \begin{thebibliography}{}
754:
755: \bibitem[Bailes et al.(1997)]{Bail97} Bailes, M., Johnston, S., Bell,
756: J. F., Lorimer, D. R., Stappers, B. W., Manchester, R. N., Lyne, A. G.,
757: Nicastro, L., D'Amico, N., \& Gaensler, B. M. 1997, \apj, 481, 386
758:
759: \bibitem[Becker \& Tr\"umper(1999)]{Beck99} Becker, W. \& Tr\"umper, J. 1999, A\&A, 341, 803
760:
761: \bibitem[Becker \& Aschenbach(2002)]{Beck02} Becker, W. \& Aschenbach,
762: B. 2002, Proceedings of the 270. WE-Heraeus Seminar on Neutron Stars,
763: Pulsars, and Supernova Remnants, Eds. W. Becker, H. Lech,
764: J. Tr\"umper, p. 64
765:
766: \bibitem[Becker et al.(2000)]{Beck00} Becker, W., Tr\"umper, J., Lommen, A. N., \& Backer, D. C. 2000, \apj, 545, 1015
767:
768: \bibitem[Beloborodov(2002)]{Belo02} Beloborodov, A. M. 2002, \apj, 566, L85
769:
770: \bibitem[Bhattacharya \& van den Heuvel(1991)]{Bhatt91} Bhattacharya, D. \& van den Heuvel, E. P. J. 1991, Phys. Rep. 203, 1
771:
772: \bibitem[Biggs(1990)]{Biggs90} Biggs, J. D. 1990, 245, 514
773:
774: \bibitem[Bogdanov et al.(2006a)]{Bog06a} Bogdanov, S., Grindlay,
775: J. E., Heinke, C. O., Camilo, F., Freire, P. C. C, \& Becker, W. 2006,
776: \apj, 646, 1104
777:
778: \bibitem[Bogdanov et al.(2006b)]{Bog06b} Bogdanov, S., Grindlay,
779: J. E., \& Rybicki, G. B. 2006, \apj, 648, L55
780:
781: \bibitem[Bogdanov et al.(2007)]{Bog07} Bogdanov, S.,
782: Rybicki, G. B., \& Grindlay, J. E. 2007, \apj, 670, 668
783:
784: \bibitem[Cadeau et al.(2007)]{Cad07} Cadeau, C., Morsink., S. M.,
785: Leahy, D., \& Campbell, S. S. 2007, ApJ, 654, 458
786:
787: \bibitem[Cameron et al.(2007)]{Cam07} Cameron, P. B., Rutledge, R. E.,
788: Camilo, F ., Bildsten, L., Ransom, S. M., \& Kulkarni, S. R. 2007,
789: \apj, 660, 587
790:
791: \bibitem[Camilo \& Rasio(2005)]{Camilo05} Camilo, F. \& Rasio,
792: F.~A.~2005, ASP Conf. Ser. Vol. 328: Binary Radio Pulsars,
793: ed. F. A. Rasio \& I. H. Stairs (San Francisco: ASP), p. 147
794:
795: \bibitem[Chen \& Ruderman(1993)]{Chen93} Chen, K. \& Ruderman, M. S. 1993, \apj, 408, 179
796:
797: \bibitem[Chen et al.(1998)]{Chen98} Chen, K., Ruderman, M., \& Zhu, T. 1998, \apj, 493, 397
798:
799: \bibitem[Grindlay et al.(2002)]{Grind02} Grindlay, J. E., Camilo, F.,
800: Heinke, C. O., Edmonds, P. D., Cohn, H., \& Lugger, P. 2002, \apj,
801: 581, 470
802:
803: \bibitem[Harding \& Muslimov(2002)]{Hard02} Harding, A. K. \&
804: Muslimov, A. G. 2002, \apj, 568, 862
805:
806: \bibitem[Hui \& Becker(2006)]{Hui06} Hui, C. Y. \& Becker, W. 2006,
807: A\&A, 448, L13
808:
809: \bibitem[Kramer et al.(1998)]{Kra98} Kramer, A., Xilouris, K. M., Lorimer, D. R., Doroshenko, O., Jessner, A., Wielebinski, R., Wolszczan, A., \& Camilo, F. 1998, \apj, 501, 270
810:
811: \bibitem[Lattimer \& Prakash(2001)]{Latt01} Lattimer, J. M. \&
812: Prakash, M. 2001, \apj, 550, 426
813:
814: \bibitem[Lommen et al.(2000)]{Lom00} Lommen, A. N., Zepka, A., Backer,
815: D. C., McLaughlin, M., Cordes, J. M., Arzoumanian, Z., Xilouris,
816: K. 2000, \apj, 545, 1007
817:
818: \bibitem[Lommen et al.(2006)]{Lom06} Lommen, A. N., Kipphorn, R. A.,
819: Nice, D. J., Splaver, E. M., Stairs, I. H., \& Backer, D. C. 2006,
820: \apj, 642, 1012
821:
822:
823: \bibitem[McClintock et al.(2004)]{McC04} McClintock, J. E., Narayan,
824: R., \& Rybicki, G. B. 2004, \apj, 615, 402
825:
826: \bibitem[Narayan \& Vivekanand(1983)]{Nar83} Narayan, R. \& Vivekanand, M. 1983, A\&A, 122, 45
827:
828: \bibitem[Pavlov \& Zavlin(1997)]{Pavlov97} Pavlov, G. G. \& Zavlin,
829: V. E. 1997, \apj, 490, L91
830:
831:
832: \bibitem[Ruderman(1991)]{Rud91} Ruderman, M. 1991, \apj, 366, 261
833:
834: \bibitem[Zavlin et al.(1996)Zavlin, Pavlov, \& Shibanov]{Zavlin96}
835: Zavlin, V. E., Pavlov, G. G., \& Shibanov, Yu. A. 1996, A\&A, 315, 141
836:
837: \bibitem[Zavlin \& Pavlov(1998)]{Zavlin98} Zavlin, V. E. \& Pavlov,
838: G. G. 1998, A\&A, 329, 583
839:
840: \bibitem[Zavlin et al.(2002)]{Zavlin02} Zavlin, V. E., Pavlov, G. G.,
841: Sanwal, D. , Manchester, R. N., Tr\"umper, J., Halpern, J. P., \&
842: Becker, W. 2002, \apj, 56 9, 894
843:
844: \bibitem[Zavlin(2006)]{Zavlin06} Zavlin, V. E. 2006, \apj, 638, 951
845:
846: \bibitem[Zavlin(2007)]{Zavlin07} Zavlin, V.~E. 2007, Ap\&SS, 308, 297
847:
848: \bibitem[Zhang \& Cheng(2003)]{Zhang03} Zhang, L. \& Cheng, K. S. 2003, A\&A, 398, 639
849:
850: \end{thebibliography}
851:
852:
853:
854: \clearpage
855:
856:
857:
858:
859:
860:
861:
862:
863:
864:
865:
866:
867:
868:
869:
870:
871:
872:
873:
874:
875:
876:
877:
878:
879:
880:
881:
882:
883:
884:
885:
886: \end{document}
887:
888:
889:
890:
891:
892:
893:
894:
895:
896:
897:
898:
899:
900:
901:
902:
903:
904:
905:
906:
907: