1: \documentclass{emulateapj}
2:
3: \usepackage{apjfonts}
4:
5: \newcommand{\msun}{M_odot}
6:
7: \newcommand{\src}{RRAT~J1819--1458}
8: \newcommand{\srcp}{RRAT~J1819--1458}
9: \newcommand{\srcx}{CXOU~J181934.1--145804}
10: \def\etal{{et~al.}}
11:
12: \catcode`\@=11
13: \newcommand{\gapprox}{\mathrel{\mathpalette\@versim>}}
14: \newcommand{\lapprox}{\mathrel{\mathpalette\@versim<}}
15: \newcommand{\propapprox}{\mathrel{\mathpalette\@versim\propto}}
16: \newcommand{\@versim}[2]
17: {\lower3.1truept\vbox{\baselineskip0pt\lineskip0.5truept
18: \ialign{$\m@th#1\hfil##\hfil$\crcr#2\crcr\sim\crcr}}}
19: \catcode`\@=12
20:
21: \shorttitle{DISCOVERY OF PULSATIONS AND A POSSIBLE SPECTRAL FEATURE IN THE X-RAY EMISSION FROM RRAT J1819--1458}
22: \shortauthors{MCLAUGHLIN ET AL.}
23:
24: \begin{document}
25:
26:
27: \title{Discovery of Pulsations and a Possible Spectral Feature in the \\ X-ray Emission from Rotating Radio Transient
28: J1819--1458}
29:
30:
31: \author{M.~A.~McLaughlin,\altaffilmark{1,2} N.~Rea,\altaffilmark{3,4}
32: B.~M.~Gaensler,\altaffilmark{4} S.~Chatterjee,\altaffilmark{4}
33: F.~Camilo,\altaffilmark{5}\\ M.~Kramer,\altaffilmark{6} D.~R.~Lorimer,\altaffilmark{1,2}
34: A.~G.~Lyne,\altaffilmark{6} G.~L.~Israel,\altaffilmark{7} \& A.~Possenti\altaffilmark{8}}
35:
36: \altaffiltext{1}{Department of Physics, West Virginia University,
37: Morgantown, WV 26501.}
38: \altaffiltext{2}{National Radio Astronomy Observatory, Green Bank, WV 24944.}
39: \altaffiltext{3}{SRON -- Netherlands Institute for Space Research,
40: Sorbonnelaan, 2, 3584 CA, Utrecht, The Netherlands.}
41: \altaffiltext{4}{School of Physics, The University of Sydney, NSW 2006, Australia }
42: \altaffiltext{5}{Columbia Astrophysics Laboratory, Columbia University,
43: New York, NY 10027.}
44: \altaffiltext{6}{Jodrell Bank Observatory, University of Manchester,
45: Macclesfield, Cheshire SK11 9DL, UK.}
46: \altaffiltext{7}{INAF -- Osservatorio Astronomico di Roma,
47: I-00040 Monteporzio Catone, Italy.}
48: \altaffiltext{8}{INAF -- Osservatorio Astronomico di Cagliari,
49: Loc. Poggio dei Pini, 09012 Capoterra, Italy.}
50:
51: \begin{abstract}
52:
53: PSR~J1819--1458 is a rotating radio transient (RRAT) source with
54: an inferred surface dipole magnetic field strength of
55: $5\times10^{13}$\,G and a 4.26-s spin period. We present
56: {\it XMM-Newton} observations of the X-ray counterpart of this source, CXOU~J181939.1--145804,
57: in which we identify pulsations and a possible spectral
58: feature.
59: The X-ray pulsations are
60: at the period predicted by the radio ephemeris, providing
61: an unambiguous identification with the radio source and confirmation
62: of its neutron star nature. The X-ray pulse has a 0.3--5~keV pulsed
63: fraction of 34\% and is aligned with the expected phase of the radio pulse. The X-ray
64: spectrum is fit well by an absorbed blackbody with $kT = 0.14$~keV
65: with the addition of an absorption feature at 1\,keV, with total
66: absorbed flux of $1.5\times10^{-13}$~ergs~cm$^{-2}$~s$^{-1}$
67: (0.3--5~keV). This absorption feature is well modeled by a Gaussian
68: or resonant cyclotron scattering model, but its significance is
69: dependent on the choice of continuum model. We find no
70: evidence for any X-ray bursts or aperiodic variability on timescales
71: of 6~ms to the duration of the observation and can place the most
72: stringent limit to date of $\le
73: 3\times10^{-9}$~ergs~cm$^{-2}$~s$^{-1}$ on the absorbed 0.3--5~keV
74: flux of any bursts.
75:
76: \end{abstract}
77:
78: \keywords{
79: pulsars: individual (J1819--1458) ---
80: radio continuum: stars ---
81: stars: flare, neutron ---
82: X-rays: stars
83: }
84:
85: \section{Introduction}
86: \label{intro}
87:
88: In 2006 February, a new class of neutron stars, the ``Rotating RAdio
89: Transients'' (RRATs) was reported (McLaughlin et al. 2006). These 11
90: objects, characterized by repeated dispersed radio bursts, have
91: periods ranging from 0.7 to 7 seconds and are located in the Galactic
92: plane at 2 -- 7 kpc distances. Their periods are longer than those of
93: most normal radio pulsars and similar to those of the populations
94: of X-ray dim isolated neutron stars (XDINS; see Haberl~2007 for a
95: review) and magnetars (see Woods \& Thompson~2006 for a review). For
96: the three sources with the highest pulse detection rates, period
97: derivatives, $\dot P$, have been determined. If the measured $\dot P$
98: values are interpreted as due to magnetic dipole spin-down, they imply
99: characteristic ages and magnetic field strengths in the general range
100: of the normal pulsar population.
101:
102: There have been several suggestions put forward on the nature of this
103: new class of neutron star. One obvious suggestion is that the RRATs
104: are related to pulsars which emit ``giant pulses'' (e.g. Knight et al. 2006).
105: However, the RRATs
106: with measured period derivatives do not appear to have high values of
107: either magnetic field strength at
108: the light cylinder or spin-down luminosity, both suggested as predictors of
109: giant-pulse activity (Cognard et al. 1996; Knight et al. 2006).
110: Zhang et al. (2006) suggest that
111: the RRATs may be neutron stars near the radio ``death line'' or may be
112: related to ``nulling'' radio pulsars. However, the period derivatives
113: measured for three RRATs do not place them near canonical pulsar
114: ``death lines'' (e.g. Chen \& Ruderman 1993) and, unlike most nulling
115: pulsars (e.g. Wang et al. 2007), we typically do not see more than one
116: pulse from the RRATs in succession. Another intriguing possibility is
117: that the sporadicity of the RRATs is due to the presence of a
118: circumstellar asteroid belt (Cordes \& Shannon 2006; Li 2006) or a
119: radiation belt such as seen in planetary magnetospheres (Luo \&
120: Melrose 2007). Or, perhaps, they are transient X-ray magnetars, a
121: particularly relevant suggestion given the recent detection by Camilo
122: et al. (2006) of transient radio pulsations from the anomalous X-ray
123: pulsar XTE~J1810--197. A final possibility is that they are similar
124: objects to PSR~B0656+14, one of three middle-aged pulsars (i.e. ``The
125: Three Musketeers''; Becker \& Truemper 1997) from which pulsed
126: high-energy emission has been detected (e.g. DeLuca et al. 2005).
127: Weltevrede et al. (2006) convincingly show that if PSR~B0656+14 were
128: more distant its emission properties would appear similar to those of
129: the RRATs. Determining the reason for the unusual emission of the
130: RRATs is important as population analyses show that their population
131: may be up to several times greater than that of the normal radio
132: pulsars (McLaughlin et al. 2006). Popov et al. (2006) show that the
133: inferred birthrate of RRATs is consistent with that of XDINS but not
134: with magnetars.
135:
136: While the radio emission properties of J1819--1458 are quite different
137: from those of ``normal'' pulsars, it appears to be a rotating neutron
138: star from which we detect radio pulses, and we henceforth give it the
139: prefix ``PSR''. PSR~J1819--1458 shows the brightest and most frequent
140: radio bursts of any of the RRAT sources. It has a 4.26-s period,
141: relatively high inferred characteristic surface dipole magnetic field
142: strength of $5\times10^{13}$~G, characteristic age
143: of 117~kyr, and spin-down luminosity of
144: $3\times10^{32}$~ergs~s$^{-1}$. The distance inferred from its
145: dispersion measure (DM) of 196$\pm3$~pc~cm$^{-3}$ is 3.6~kpc (Cordes
146: \& Lazio~2002), with considerable (at least 25\%) uncertainty. PSR~J1819--1458 is
147: characterized by radio bursts at 1.4~GHz of average duration 3~ms,
148: with bursts arriving randomly with a mean rate of one every $\sim$~3~minutes. X-ray emission was
149: detected from this source in a serendipitous 30-ks {\it Chandra}
150: ACIS-I observation toward the Galactic supernova remnant G15.9+0.2
151: (Reynolds et al. 2006)\footnote{As discussed by Reynolds et
152: al. (2006), an association between the remnant and PSR~J1819--1458
153: is extremely unlikely.}. They found that the spectrum was well
154: described by an absorbed blackbody with neutral hydrogen column
155: density $N_H = 7^{+7}_{-4} \times 10^{21}$ cm$^{-2}$ and temperature
156: $kT=0.12 \pm 0.04$\,keV, with an absorbed flux of $\sim1 \times
157: 10^{-13}$\,ergs\,cm$^{-2}$\,s$^{-1}$ between 0.3 and 5 keV. These
158: properties are consistent with emission from a cooling neutron star of
159: age $10^4-10^5$~years, broadly consistent with the characteristic age
160: of PSR~J1819--1458. No evidence for bursts or variability was found.
161: The time resolution of these {\it Chandra} observations was not
162: sufficient to allow a robust search for X-ray pulsations.
163:
164: We were awarded 43~ks of {\it XMM-Newton} time to further characterize
165: the spectrum and search for pulsations. We report here on the results
166: of these observations, in particular the detection of X-ray pulsations and a
167: possible feature in the X-ray spectrum. This is the first
168: detection of X-ray pulsations from any of the RRAT sources. In
169: Section~2 we describe the observations and both the timing and
170: spectral analyses. In Section~3 we discuss possible interpretations of
171: our results, and present our conclusions in Section~4.
172:
173: \section{Observations and Analysis}
174: \label{obs}
175:
176: The {\it XMM-Newton} observations were performed on 2006 5 April, with
177: a total observation time of 43~ks. The European Photon Imaging Camera
178: (EPIC) PN and MOS instruments were operated with medium filters and in
179: Small Window mode, providing a time resolution of 6 and 300~ms and
180: effective livetimes of 71\% and 97.5\%, respectively. The data were
181: reduced using the {\it XMM-Newton} Science Analysis System (SAS
182: version 7.0.0) and the most recent calibration files. Data from both
183: the PN and MOS instruments were used for the timing and spectral
184: analyses. We also also analyzed the Reflection Grating Spectrometer data but
185: our target was too faint to be reliably detected.
186:
187: We filtered the observation for background flares, resulting in
188: effective on-source exposure times of 32.5 and 37.8~ks (23.0 and
189: 36.9~ks including deadtime) for the PN and MOS instruments,
190: respectively. We detect a point source with J2000 coordinates: right ascension $\alpha
191: = 18^{\rm h}19^{\rm m} 34^{\rm s}$ and declination $\delta = -14^\circ 58' 03''$
192: (4$^{\prime\prime}$ error in each coordinate). This is consistent with
193: the radio-timing-derived position
194: and with the more accurate position for the X-ray counterpart
195: published in Reynolds et al. (2006). There is no evidence for extended
196: emission, with the source brightness falling off as expected given the
197: {\it XMM-Newton} point spread function.
198:
199: For both PN and MOS data, we extracted the source photons within a
200: 20$^{\prime\prime}$ circular radius centered on the source position,
201: which ensured extraction of more than 90\% of the source counts. Following
202: standard practice,
203: the size of the extraction region has not been corrected for in
204: our final quoted fluxes and luminosities.
205: The background counts were extracted from four 20$^{\prime\prime}$
206: circular regions centered on off-source positions (different for the
207: PN and MOS instruments) in the same central CCD.
208:
209: \subsection{Timing Analysis}
210:
211: For the timing analysis, we used all photons from PN and MOS1 and MOS2
212: instruments with PATTERN $\le$ 12 (i.e. allowing for single, double, triple and
213: quadruple events\footnote{http://xmm.vilspa.esa.es/external/xmm\_user\_support/documentation/uhb}).
214: This resulted in $2200\pm58$ and
215: $945\pm15$ source and background PN counts and $1380\pm41$ and $230\pm8$
216: source and background MOS1 plus MOS2 counts (normalized for a 20$^{\prime\prime}$ circular extraction
217: region).
218: Although we can measure
219: a period of 4.26~s from the arrival times of the radio bursts, the
220: periodicity is not detectable in a Fast Fourier Transform of the radio
221: time series due to the sporadic nature of the source.
222: We therefore performed a periodicity search of all PN and
223: MOS photons (i.e. with energies 0.2--15~keV) as a confirmation of the method through which the radio
224: period was derived (see McLaughlin et al. 2006). The arrival times
225: were converted to the solar system barycenter and the $Z_1^2$ test
226: (Buccheri et al. 1983) was applied. The most significant signal
227: is detected
228: with $Z_1^2 = 144.5$
229: at a frequency of $234.564\pm0.007$ mHz (all errors are quoted at the
230: 1$\sigma$ confidence level). Allowing for $4.3\times10^{6}$ independent frequencies searched
231: in the range 0.1--100~Hz, the probability of
232: chance occurrence of this signal in the absence of a real pulsation
233: would be $1.8\times10^{-25}$, showing that the periodicity from this source would be easily
234: detectable in a blind search of the X-ray photons.
235:
236: The ephemeris obtained through continued radio timing of
237: PSR~J1819--1458 with the Parkes radio telescope at 1.4~GHz using the
238: TEMPO software package predicts a barycentric frequency at the center
239: of the observation (MJD~53830.87029) of $234.566244\pm0.000001$~mHz,
240: consistent with the measured X-ray frequency. Accounting
241: for spin-down during the observation is unimportant as the change in
242: frequency due to the $-3.16\times10^{-14}$~Hz~s$^{-1}$ frequency
243: derivative over the 12-hr observation is much less than the size of an
244: individual frequency bin. The X-ray detection of
245: periodicity shows that the radio-derived period is indeed the true
246: period, and not a smaller common factor of the radio arrival times
247: (see McLaughlin et al. 2006).
248:
249: In Figure~1, we present the 0.3--5~keV background-corrected X-ray pulse profile
250: formed by folding barycentered photons from the PN and both MOS
251: instruments using the radio ephemeris. The same good time interval
252: file was used for the PN and both MOS instruments and backgrounds were
253: subtracted separately. The different exposure times for the two
254: instruments were accounted for. We chose to restrict this analysis to
255: 0.3--5~keV as below 0.3~keV, the PN and MOS detectors are not well
256: calibrated and, above 5~keV, the background dominates. In Figure~1, we
257: also present the radio profile formed from adding individual bright
258: radio pulses with the radio ephemeris. The X-ray pulse profile has a
259: 0.3--5 keV background-corrected pulsed fraction of 34$\pm$6\%, defined as
260: $(F_{max}-F_{min})/(F_{max}+F_{min})$, where $F_{max}$ and $F_{min}$ are the
261: minimum and maximum values of the X-ray pulse profile. It can be well modeled as a
262: single sinusoid with $\chi^2_\nu \sim 1.2$ (17 d.o.f.).
263: We find no significant dependence of
264: pulsed fraction on energy, with a pulsed fraction at energies
265: $0.3-1$~keV of 28$\pm7$\% and a pulsed fraction at energies from
266: 1--5~keV of 49$\pm$10\%.
267:
268: The peak of the X-ray profile, calculated by fitting a sinusoid, is at
269: phase $0.49\pm0.04$ and is aligned to the peak of the radio
270: profile at phase 0.5. The 3~pc~cm$^{-3}$ uncertainty in the
271: DM of 196~pc~cm$^{-3}$ amounts to a radio arrival time uncertainty at
272: 1.4~GHz of only 13~ms, or 0.3\% of the 4.26~s pulse period.
273:
274: Our PN data can be used to place limits on the existence of X-ray
275: bursts from the source. Binning the data on various timescales to
276: search for bursts of different widths
277: we find no evidence for
278: aperiodic variability on timescales ranging from 6~ms to the duration
279: of our observation. We used photons with energies up to 10 keV,
280: important given the hard spectra of magnetar bursts (e.g. Gavriil et al 2004).
281: Omitting frames with high particle background, we
282: find that none of the $4\times10^{6}$ frames contains more than two
283: counts within the 20$^{\prime\prime}$ region, with three frames
284: containing exactly two counts. These numbers are entirely consistent
285: with Poisson statistics. The number of events in a single frame which
286: would deviate from a steady flux at the 3-$\sigma$ level is 3
287: photons. We therefore adopt this as the upper limit on the X-ray flux
288: of any burst of width 6~ms or less. Assuming a spectrum of the same
289: shape as that fitted for the overall source (column 2 of Table~1; see Section~2.2)
290: limits the observed fluence of any burst to $\la
291: 8\times10^{-12}$~ergs~cm$^{-2}$ (0.3--5~keV), corresponding to an
292: absorbed flux limit of $\la 3\times10^{-9}$~ergs~cm$^{-2}$~s$^{-1}$
293: (0.3--5~keV) if we assume that the X-ray burst would last for 3~ms,
294: the width of the radio pulse. This is a factor of five better than the
295: limits placed by Reynolds et al. (2006). At a distance of 3.6~kpc,
296: these limits correspond to a total energy of $2\times10^{34}$~ergs and
297: luminosity of $1\times10^{37}$~ergs~s$^{-1}$. Therefore, any bursts,
298: at least during the 12-hour duration of our observation, must contain
299: much less energy than typical X-ray magnetar bursts (e.g. Woods \& Thompson
300: 2006) if their spectra is indeed similar to that
301: described in column 2 of Table~1.
302:
303: \subsection{Spectral Analysis}
304: \label{spectrum}
305:
306: For the spectral analysis, we used PN photons with PATTERN $\le$ 4 (i.e. single and double events) and
307: MOS1 and MOS2 photons with PATTERN $\le$ 12. Source and background
308: spectra were extracted from the same regions used for the timing
309: analysis and the spectral response matrices were created with the SAS
310: {\tt mkrmf} and {\tt mkarf} tools, using the bad-pixel file built for
311: our observation. The PN and MOS spectra were only used in the
312: 0.4--2\,keV energy range, a smaller range than that used for the timing analysis due to the
313: greater dependence of spectral fitting on background
314: spectra.
315: In the spectral analysis we used both rebinned and
316: unbinned spectra. Spectra were rebinned for MOS and PN
317: by a factor of two so that the energy resolution was not
318: oversampled by more than a factor of three (using the specific
319: response matrix built for this observation). Furthermore, we rebinned
320: in order to have at least 30 counts per bin so that we could
321: use the
322: $\chi^2$ statistic\footnote{http://heasarc.nasa.gov/xanadu/xspec/manual/manual.html}.
323:
324: We first modeled the PN spectrum. We tried several different models
325: and found that a single component fit was not
326: possible. Fitting with a single absorbed\footnote{If not otherwise
327: specified, abundances were assumed to be solar and fixed at the
328: values in Lodders (2003).} blackbody (as in Reynolds et al.~2006),
329: we found $N_H\sim 4.0\times 10^{21}$\,cm$^{-2}$ and $kT\sim0.14$\,keV
330: with a $\chi^2_\nu > 2.0$ (see Figure~2), while for an absorbed
331: powerlaw, we obtained $N_H\sim 1.3\times 10^{22}$\,cm$^{-2}$ and
332: $\Gamma\sim8.2$ with a $\chi^2_\nu >$2.1 (both fits have 53
333: d.o.f).
334: These values are consistent with the Reynolds et al. (2006)
335: best-fit parameters (0.5--8~keV) of $N_H = 7^{+7}_{-4} \times 10^{21}$
336: cm$^{-2}$ and $kT=0.12 \pm 0.04$\,keV (for the absorbed blackbody
337: model) and $\Gamma\sim9.5$ (for the power-law model). An inspection
338: of the residuals revealed that our high $\chi^2_\nu$ values were due
339: to the presence of strong spectral features around 1\,keV and a
340: weaker one around 0.5\,keV.
341:
342: We carefully checked whether these features might be due to
343: calibration issues, to our source and background extraction regions or
344: to residual particle flares and/or particles hitting the detector.
345: We reliably excluded all of these issues by studying in detail the
346: {\it XMM-Newton} calibration lines (see Figure~3), extracting
347: the source and background photons from several different
348: regions and investigating the spectrum using only PATTERN =
349: 0 counts (i.e. only isolated events).
350: However, we note that the 0.5\,keV feature is very
351: close to the Oxygen edge energy, and an overabundance of Oxygen in
352: the direction of the source
353: could be responsible.
354: To further investigate this
355: possibility, we fit our spectrum with an absorbed blackbody model using
356: three different photoelectric cross-sections, those of
357: Ba{\l}uci\'nska-Church \& McCammon (1992, 1998) and of Verner
358: et. al. (1996). This was aimed at studying the dependence of the
359: significance of our 0.5\,keV line on the chosen photoelectric
360: cross-section, which drives the shape of the edge. We found that the
361: residuals around 0.5\,keV remain in all cases, although less
362: significantly if we use the Verner et. al. (1996) cross-section. For
363: that reason, we assumed this cross-section for all our spectral
364: fitting, and we omitted all photons in the 0.50--0.53~keV energy range (two
365: bins in our rebinned PN spectrum) from our modeling. We tentatively
366: conclude that this line is due to the Oxygen edge, but only future
367: deeper observations will unambiguously confirm this.
368:
369: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% TABLE 1 %%%%%%%%%%%%%%%%%%
370: \begin{table*}
371: \begin{center}
372: \caption{Spectral fits for PSR~J1819--1458 with EPIC-PN}
373: \vspace{0.3cm}
374:
375: \begin{tabular}{lclclclc}
376: \hline
377: \hline
378: \multicolumn{2}{c}{Blackbody (BB) plus Neon} & \multicolumn{2}{c}{BB plus Gaussian}
379: & \multicolumn{2}{c}{BB plus edge} & \multicolumn{2}{c}{BB plus cyclotron}\\
380: \hline
381: & & & & & & & \\
382: $N_{H}$ & 0.59$^{+0.06}_{-0.04}$ & & 0.75$^{+0.12}_{-0.09}$ &
383: & 0
384: .57$\pm0.06$ & & 0.81$^{+0.09}_{-0.08}$ \\
385: $N_{Ne}$ & 6$\pm1$ & E$_{G}$ &
386: 1.11$^{+0.04}_{-0.03}$
387: & E$_{e}$ & 0.92$^{+0.03}_{-0.01}$ & E$_{cy}$ & 0.99$^{+0.03}_{-0.02}$
388: \\
389: \\
390: & & $\sigma_{G}$ & 0.21$^{+0.03}_{-0.06}$ & &
391: & w$_{c
392: y}$ & 0.37$^{+0.03}_{-0.06}$ \\
393: & & $\tau_{G}$ & 150$\pm{60}$ &$\tau_{e}$ &
394: 0.67$\pm0.14$
395: & d$_{cy}$ & 1.2$\pm0.2$ \\
396:
397: & & & & & & & \\
398: $kT$ & 0.144$^{+0.008}_{-0.006}$ & & 0.136$^{+0.012}_{-0.008}$ & &
399: 0.150$^{+0.
400: 005}_{-0.006}$ & & 0.144$^{+0.008}_{-0.006}$ \\
401: Abs. Flux& 1.5$^{+0.3}_{-0.5}$ & & 1.5$^{+0.6}_{-0.8}$ & &
402: 1.5$^{+0.3}_{-0.6}$ & &
403: 1.5$^{+0.3}_{-0.4}$ \\
404: Unab. Flux & 15$\pm2$ & & 26$^{+2}_{-8}$ & &
405: 12$^{+2}_{-2}$ & & 34$^{+4}_{-9}$ \\
406: $\chi^2_\nu$ (d.o.f.) & 1.20 (50) & & 1.19 (48) & & 1.17 (49) & &
407: 1.13 (48) \\
408:
409: \hline
410: \end{tabular}
411: \tablecomments{Fluxes are calculated in the 0.3--5\,keV energy range,
412: and reported in units of
413: $10^{-13}$~ergs\,s$^{-1}$\,cm$^{-2}$. $N_{H}$ is in units of
414: $10^{22}$cm$^{-2}$ and $N_{Ne}$ is in solar units (always assuming
415: solar abundances from Lodders 2003). The photoelectric cross-section
416: of Verner et al. (1998) has been used for all fits. The values of $kT$
417: (blackbody temperature), E$_{G}$ (Gaussian line energy),
418: $\sigma_{G}$ (Gaussian line width), $E_e$ (edge threshold energy),
419: E$_{cy}$ (cyclotron line energy) and w$_{cy}$ (cyclotron line width)
420: are in units of keV. The Gaussian line depth $\tau_G$, edge depth
421: $\tau_e$ and fundamental cyclotron line depth $d_{cy}$ are
422: dimensionless.
423: Errors are at the 1$\sigma$
424: confidence level. XSPEC models used are (from left to right): {\tt
425: vphabs*bbody, phabs*gabs*bbody, phabs*edge*bbody} and {\tt
426: phabs*cyclabs*bbody}.}
427: \end{center}
428: \label{spec}
429: \end{table*}
430: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
431:
432: Keeping the number of components as low as possible, we tried to model
433: our rebinned PN spectrum with either an absorbed blackbody (XSPEC model {\tt bbody}) or
434: power-law ({\tt powerlaw}), in addition to modeling the $\sim$1\,keV feature in several
435: ways: leaving free the abundances of the most abundant elements in the
436: interstellar medium (ISM) with lines around 1\,keV (e.g. Ne, N, Mg,
437: Fe), adding a Gaussian function (XSPEC model {\tt gabs}), a Lorentzian
438: function ({\tt lorentz}), an absorption edge ({\tt edge}) or a cyclotron
439: resonant scattering model ({\tt cyclabs}).
440: Among our trials, only the models reported in
441: Table\,1 gave satisfactory values of $\chi^2_\nu\sim1$. We show the
442: blackbody plus Gaussian fit and residuals in Figure~4. Note that there
443: is evidence in the residuals for narrow lines within our broad 1 keV
444: line, implying that the line may be a blending of narrower
445: lines. However, the addition of several single narrow lines is not
446: statistically significant, and future data with better statistics are
447: necessary for investigating this issue.
448:
449: We ran Monte Carlo simulations of 2$\times10^4$ spectra (see Rea et
450: al.~(2005, 2007) for details) to estimate the significance
451: of the spectral feature (as suggested by Protassov et
452: al.~2002)\footnote{Even though the F-test is still widely used to
453: assess the significance of features in X-ray spectra, it has been
454: shown that this is not statistically correct (Protassov et
455: al.~2002).}. The spectra were simulated using models corresponding to columns 2 and 4 from
456: Table 1 with the absorbed blackbody parameters varying within their
457: 3$\sigma$ errors (see Table\,1), absorption line parameters
458: completely free to vary, and fixing the number of counts of each
459: spectra at the PN count value for PSR\,J1819--1458. After having generated
460: 10$^4$ spectra for each of the two models, we counted how many of
461: these simulated spectra showed an absorption feature at any energy,
462: width or depth, only due to statistical fluctuations. None of the
463: simulated spectra presented a Gaussian absorption line at any energy
464: or width with a depth $\tau_G > 30$, and no
465: cyclotron component was detected with a depth $>0.3$ for any energy or
466: width. We therefore infer the significance of the line at 1\,keV to
467: be $> 1/10^4$, corresponding to a $>$ 99.99\% probability (i.e. at least 4$\sigma$) that this
468: line is not due to statistical fluctuations.
469:
470: To check whether the line was also present in the MOS data, we
471: simultaneously fit the PN and both MOS1 and MOS2 rebinned spectra (see
472: Figure~5). All models in Table~1 fit the PN, MOS1 and MOS2
473: data well. However, because of the lower number of MOS counts, our 1~keV
474: line was not significant in the MOS spectra, especially because the
475: rebinning we did in order to use the $\chi^2$ statistic left only a
476: handful of bins within the line. As a further check, we then used the
477: unbinned PN and MOS spectra and the C-statistic (Cash 1979) to assess the
478: goodness of an absorbed blackbody fit, without the inclusion of any
479: line. We found that for $N_H=0.5\times10^{22}$\,cm$^{-2}$ and
480: $kT=0.14$\,keV, the C-statistic is 1009.776 using 969 PHA bins. A
481: Monte Carlo simulation of the fit showed that the probability of
482: having a C-statistic $<$ 1010 is 98.0\%. Note that only if the C-statistic
483: probability is smaller then 50\% can the model be accepted. This
484: shows that an absorbed blackbody model alone can not reproduce the
485: data, and hence that a line is also present in the MOS spectra. We show
486: in Figure\,5 the PN, MOS1 and MOS2 spectra modeled with an absorbed
487: blackbody plus a Gaussian (column 2 in Table\,1), with a more severe PN
488: rebinning (with at least 40 counts per bin), to show
489: the agreement between the three instruments. Note that all the models in
490: Table~1 satisfactorily fit the previous {\it Chandra} data (Reynolds
491: et al. 2006), although the absorption line is not statistically
492: significant in those data because of the reduced number of counts.
493:
494: Furthermore, we checked whether the spectral feature parameters and
495: significance were dependent on the spectral binning or on the
496: continuum model. We found that the spectral feature does not vary
497: significantly when we vary the amount of spectral binning (see also Figure\,5), but
498: that it is rather dependent on the chosen continuum (as is generally
499: the case for every spectral feature). In fact, if instead of a blackbody
500: model, we assume a power--law continuum,
501: the feature is still detectable with similar
502: spectral properties but at a lower significance of 98.9\% (roughly 2.5$\sigma$).
503:
504: Hence, if we assume a blackbody continuum model
505: (reasonable for an isolated neutron star) we can show with high
506: confidence that a spectral feature is present in our data.
507: However, our statistics do not allow us to statistically prefer one
508: model to the others and reliably ascertain its nature.
509: Furthermore, if future, more sensitive observations reveal
510: a different source continuum, the significance of the
511: feature may need to be revised.
512:
513: We also performed phase-resolved spectroscopy over two intervals of
514: 0.5 in phase centered on the peak and the minimum of the pulse
515: profile, respectively. We did not find any significant spectral
516: variability between these two phase intervals at the 2$\sigma$ level.
517: However the low number
518: of counts in each spectrum does not allow us to reach any reliable
519: conclusions about the pulse-phase dependence of the spectral
520: parameters.
521:
522: \section{Discussion}
523:
524: \subsection{Interpretation of the spectral feature}
525:
526: If we accept that the spectral feature is significant, the two
527: main possible interpretations are as an atomic line or as a cyclotron
528: resonant scattering line.
529:
530: If the line is atomic, then it could be due either to the neutron star
531: atmosphere or, but less probably, to a peculiar abundances in the ISM
532: in that direction (or perhaps even a combination of the two). At
533: 1\,keV, it might then be due to the N edge (although improbable given
534: its low abundance), Ne edge (around 0.9~keV), Ne XI K-edge (at
535: 1.19\,keV), or some Fe-L lines, including Fe XX,
536: Fe XXI and Fe XXII or Fe XXIII (all around 0.9\,keV). Note that Fe lines are common in other
537: kinds of neutron stars, though mostly coming from accretion disks in
538: binary systems rather than from the neutron stars themselves (see
539: e.g. Cottam et al. 2002). The structure
540: present in the residuals (see Figure~4 middle panel), which in our data
541: is not significant, might be due to
542: a blending of narrow lines which we could only model with a broad
543: line due to our limited number of counts.
544:
545: If the feature is
546: due to proton cyclotron resonant scattering, the magnetic field inferred would be
547: $B_{cy}=1.6E_{cy}({\rm keV})/y_G~10^{14}$\,G, where $y_G =
548: (1-2GM/c^2R)^{1/2}$ is the gravitational redshift factor ($M$ and $R$
549: are the neutron star mass and radius, respectively). Assuming
550: canonical values for $M$ and $R$ of 1.4~$M_\odot$ and 10~km, we find
551: $B_{cy} = 2\times10^{14}$~G. This is slightly higher than the dipolar
552: surface magnetic field inferred through radio timing through the standard formula $B=3.2\times10^{19}{P\dot{P}^{1/2}}=5\times10^{13}$~G. However,
553: this expression can be considered an order-of-magnitude estimate as it
554: assumes a purely dipolar field, a neutron star radius of 10~km, a
555: moment of inertia of $10^{45}$~g~cm$^{2}$ and an angle between the
556: spin and magnetic axes $\alpha = 90$~degrees.
557: Accounting for
558: this and assuming $\alpha = 30$~degrees would make the timing- and
559: cyclotron-inferred fields consistent.
560: In addition, the width and
561: depth of the line are consistent with the predictions of Zane et
562: al. (2001) for proton-cyclotron absorption in highly magnetized
563: neutron stars. We cannot detect any harmonics to the fundamental
564: cyclotron line because our spectrum is
565: background dominated above 2\,keV.
566: However, as observed in some accreting
567: sources (Heindl et al. 2004) and investigated for isolated neutron stars
568: by Liu et al. (2006), in some cases the first harmonic is
569: as deep as the fundamental cyclotron frequency. Therefore, it is possible, although
570: unlikely, that
571: the 1\,keV feature is the first
572: harmonic, with the 0.5~keV fundamental
573: coincident with the depression in the spectrum that we have
574: interpreted as due to an overabundance of Oxygen.
575: We note that the observed lines are unlikely to be
576: due to electron-cyclotron absorption as the
577: inferred magnetic field would be a factor of 2000 lower
578: (i.e. the ratio of the proton to electron mass)
579: and incompatible with that measured through radio timing.
580:
581: \subsection{Relationship to other classes of neutron stars}
582:
583: Our detection of periodicity at the radio period shows that the X-ray
584: source reported by Reynolds et al. (2006) is undoubtedly the
585: counterpart to PSR~J1819--1458. The pulsed fraction and sinusoidal
586: pulse shape are similar to what is observed for other middle-aged
587: X-ray-detected radio pulsars such as B0656+14 (e.g. De Luca et
588: al. 2005), which has been observed to have similar radio
589: properties to the RRATs (Weltevrede et al.~2006).
590:
591: The thermal emission from PSR~J1819--1458 is consistent with a cooling
592: neutron star. However, the temperature from our blackbody fit appears
593: slightly higher than temperatures derived from blackbody fits for
594: other neutron stars of similar ages (see discussion by Reynolds et
595: al.~2006).
596: Note that it is possible that PSR~J1819--1458 was born
597: spinning at a sizable fraction of its present period of 4.26~s. In
598: this case, as discussed by Reynolds et al. (2006), the characteristic age could be a
599: considerable overestimate and the inferred temperature could be
600: completely consistent with its age. Note that characteristic ages
601: have been shown to be misleading for several other pulsars
602: (e.g. Gaensler \& Frail 2000; Kramer et al.~2003).
603:
604: Including PSR~J1819--1458, eight high-magnetic field radio pulsars
605: (i.e. $B > 1\times10^{13}$~G)
606: have now been observed at X-ray energies.
607: Two, J1846--0258 and B1509--58, are bright non-thermal sources (Mereghetti et al. 2002; Cusumano et al. 2001),
608: as expected given their young ages (less than 2000
609: yr). PSR~J1119--6127 is a bright thermal X-ray emitter with
610: unusual properties including a large pulsed fraction and narrow
611: pulse (Gonzalez et al. 2005).
612: PSR~J1718--3718, with magnetic field of $7\times10^{13}$~G,
613: has been detected at X-ray energies, but the faintness of the
614: counterpart does not allow detailed spectral modeling or a
615: constraining limit on pulsed fraction (Kaspi \& McLaughlin 2005). No
616: X-ray emission has been detected from PSRs J1814--1744, B0154+61 or
617: J1847--0130, which has the highest inferred surface dipole magnetic
618: field ($9\times10^{13}$~G) measured to date for any radio pulsar
619: (Pivovaroff et al. 2000; Gonzalez et al. 2004; McLaughlin et al. 2003). Radio pulsar X-ray emission
620: properties clearly vary widely, even for objects with very similar spin-down
621: properties.
622:
623: While the spectrum and luminosity of PSR~J1819--1458
624: argue against a relationship with
625: magnetars, it is possible that
626: PSR~J1819--1458 could be a transition object between
627: the pulsar and magnetar source classes. The soft X-ray spectrum does
628: have a comparable temperature to the quiescent state of XTE~J1810--197
629: ($kT \sim 0.15-0.18$ keV; Ibrahim et al.~2004; Gotthelf et al.~2004).
630: However, the radio emission characteristics of these two neutron stars
631: are quite different.
632:
633: While resonant cyclotron features are regularly observed from X-ray binary
634: systems (e.g. Truemper et al. 1978; Nakajima et al. 2006), the
635: detection of such features from isolated neutron stars is quite
636: unusual. Sanwal et al. (2002) \& Mereghetti et al. (2002)
637: discovered harmonically spaced
638: absorption lines from 1E~1207.4--5209, a radio-quiet X-ray pulsar with
639: a 424-ms spin period and timing-derived characteristic age and
640: inferred surface dipole magnetic field strength of $3\times10^{5}$~yr
641: and $3\times10^{12}$~G, respectively. Analyzing a deeper observation of the source,
642: Bignami et al. (2003) attributed these lines to
643: electron-cyclotron absorption. However,
644: the significance of two of the lines present in its spectrum has been strongly
645: questioned (Mori et al.~2005).
646:
647: Broad absorption lines, similar to those seen for
648: PSR\,J1819--1458, have been observed
649: for six out of seven XDINS
650: (see reviews by van Kerkwijk \& Kaplan 2007 and Haberl 2007). For most of these neutron stars, the
651: lines can be interpreted as due to neutral hydrogen transitions in
652: highly magnetized atmospheres. Ho et al. (2003) and Van Kerkwijk \& Kaplan (2007) argue that
653: the transition energy is similar to the proton
654: cyclotron energy for magnetic fields of the order of PSR~J1819--1458's.
655: The X--ray
656: spectrum of PSR\,J1819--1458 is very similar (although with a slightly
657: hotter blackbody temperature) to the XDINSs, although so far no
658: convincing evidence for radio bursts has been detected for any of
659: those thermally emitting neutron stars (Kondratiev et al., in
660: preparation).
661:
662: One outstanding question is
663: why absorption lines of this kind, whether due to the atmosphere or
664: to cyclotron resonant scattering, have been observed from only a
665: handful of X-ray emitting isolated neutron stars. As suggested by Mereghetti et al.
666: (2002), the age of the
667: neutron star could be one key factor. Young objects are dominated by
668: non-thermal emission, but older ones may be too faint for X-rays to be
669: detectable, making X-ray bright, middle-aged pulsars the best
670: candidates (as is the case of the XDINSs and of
671: PSR\,1819--1458). Note, however, that no such absorption lines have
672: been found for the X-ray bright, middle-aged PSR\,B0656+14, despite
673: deep searches both with {\it Chandra} (Marshall \& Schultz 2002) and
674: {\it XMM-Newton} (De Luca et al. 2005) and the theoretical predictions
675: for cyclotron and/or atmospheric features expected in its
676: emission. The explanation could well depend on the
677: viewing geometry.
678:
679: More sensitive observations are necessary to confirm the presence of
680: absorption features in the spectrum of PSR~J1819--1458. Longer X--ray
681: observations are needed to understand whether the broad 1~keV line could be a
682: blending of narrower
683: features. This would then argue for an atomic origin for the
684: line. Longer observations will also allow
685: phase-resolved
686: spectroscopy. Simulations show that a 100-ks observation would allow us to
687: achieve both of these goals.
688: Because the strength of the cyclotron line depends on
689: the angle between the observer and the magnetic field, we expect phase variation of
690: cyclotron features.
691: If the feature we detect is indeed due to
692: proton-cyclotron absorption, it provides an invaluable means of testing the assumptions
693: implicit in the characteristic magnetic fields derived through radio timing
694: measurements and a unique independent measurement of the magnetic field of an isolated neutron star.
695:
696: \section{Conclusions}
697:
698: We have discovered X-ray pulsations at the 4.26-s period inferred from
699: radio timing of PSR~J1819--1458.
700: The properties of these pulsations are similar to those
701: observed for other middle-aged radio pulsars detected at X-ray
702: energies. While the RRATs are characterized by sporadic radio emission,
703: we do not detect any X-ray bursts or aperiodic variations
704: throughout the observation and can place the most stringent limit to
705: date of $\le 3\times10^{-9}$~ergs~cm$^{-2}$~s$^{-1}$ on the absorbed
706: 0.3--5~keV flux of any X-ray bursts. We have characterized the
707: spectrum of this source and find that it is well-described by an
708: absorbed blackbody with $kT = 0.14$~keV in addition to an absorption
709: line around 1\,keV, with total absorbed flux of
710: $1.5\times10^{-13}$~ergs~cm$^{-2}$~s$^{-1}$ (0.3--5~keV). We note,
711: however, that the presence of this absorption feature is highly
712: dependent on the choice of continuum model and needs further X-ray
713: observations to be confirmed. This object is the only RRAT so far to be
714: detected at X-ray energies. X-ray observations of the other objects in this
715: source class are essential for a complete picture of how they relate to other
716: neutron star populations.
717:
718:
719: \acknowledgments
720:
721: MAM thanks the Parkes Multibeam Survey team for radio observing assistance and
722: Pete Woods for helpful discussions on XMM data analysis.
723: This work was supported by NASA through {\it XMM-Newton} Observer Program
724: grant NNX06AG20G. N.R. is supported by an NWO Postdoctoral Fellowship
725: and a Short Term Visiting Fellowship awarded by the University of
726: Sydney, and kindly thanks M. M\'endez for useful suggestions.
727:
728: \begin{thebibliography}{}
729:
730: \bibitem[Balucinska-Church \& McCammon(1992)]{1992ApJ...400..699B}
731: Ba{\l}uci{'n}ska-Church, M., \& McCammon, D.\ 1992, \apj, 400, 699
732:
733: \bibitem[Balucinska-Church \& McCammon(1998)]{1998ApJ...400..699B}
734: Ba{\l}uci{'n}ska-Church, M., \& McCammon, D.\ 1998, \apj, 496, 1044
735:
736: \bibitem[bt97]{bt97} Becker, W. \& Truemper, J. 1997, \aap, 326, 682
737:
738: \bibitem[Bhat et al.(2007)]{2007A&A...462..257B} Bhat, N.~D.~R., Gupta, Y.,
739: Kramer, M., Karastergiou, A., Lyne, A.~G., \& Johnston, S.\ 2007, \aap,
740: 462, 257
741:
742: \bibitem[Bignami et al.(2003)]{2003Natur.423..725B} Bignami, G.~F.,
743: Caraveo, P.~A., De Luca, A., \& Mereghetti, S.\ 2003, \nat, 423, 725
744:
745: \bibitem[Buccheri et al.(1983)]{buc1983} Buccheri, R., et al.\
746: 1983, \aap, 128, 245
747:
748: \bibitem[Camilo et al. (2006)]{camilo06} Camilo, F., Ransom,
749: S.~M., Halpern, J.~P., Reynolds, J., Helfand, D.~J., Zimmerman, N., \&
750: Sarkissian, J.\ 2006, \nat, 442, 892
751:
752: \bibitem[Cash(1979)]{cash79} Cash, W.\ 1979, \apj, 228, 939
753:
754: \bibitem[Chen \& Ruderman(1993)]{1993ApJ...402..264C} Chen, K., \&
755: Ruderman, M.\ 1993, \apj, 402, 264
756:
757: \bibitem[Cognard et al.(1996)]{cognard9} Cognard, I., Shrauner,
758: J.~A., Taylor, J.~H., \& Thorsett, S.~E.\ 1996, \apjl, 457, L81
759:
760: \bibitem[Cordes \& Lazio (2002)]{cordes02}
761: Cordes, J.~M., \& Lazio, T.\ 2002, astro-ph/0207156
762:
763: \bibitem[Cordes \& Shannon(2006)]{2006astro.ph..5145C} Cordes, J.~M., \&
764: Shannon, R.~M.\ 2006, submitted, astro-ph/0605145
765:
766: \bibitem[Cottam et al.2002]{cottam02} Cottam, J., Paerels, F.,
767: \& Mendez, M.\ 2002, \nat, 420, 51
768:
769: \bibitem[Cusumano et al.(2001)]{2001A&A...375..397C} Cusumano, G., Mineo,
770: T., Massaro, E., Nicastro, L., Trussoni, E., Massaglia, S., Hermsen, W., \&
771: Kuiper, L.\ 2001, \aap, 375, 397
772:
773: \bibitem[De Luca et al.(2005)]{deluca2005} De Luca, A., Caraveo,
774: P.~A., Mereghetti, S., Negroni, M., \& Bignami, G.~F.\ 2005, \apj, 623,
775: 1051
776:
777: \bibitem[Gaensler \& Frail(2000)]{ga20} Gaensler, B.~M., \&
778: Frail, D.~A.\ 2000, \nat, 406, 158
779:
780: \bibitem[Gavriil et al.(2004)]{gav04} Gavriil, F.~P., Kaspi,
781: V.~M., \& Woods, P.~M.\ 2004, \apj, 607, 959
782:
783: \bibitem[Gonzalez et al.(2005)]{2005ApJ...630..489G} Gonzalez, M.~E.,
784: Kaspi, V.~M., Camilo, F., Gaensler, B.~M., \& Pivovaroff, M.~J.\ 2005,
785: \apj, 630, 489
786:
787: \bibitem[Gonzalez et al.(2004)]{2004ApJ...610L..37G} Gonzalez, M.~E.,
788: Kaspi, V.~M., Lyne, A.~G., \& Pivovaroff, M.~J.\ 2004, \apjl, 610, L37
789:
790: \bibitem[Gotthelf et al.(2004)]{gotthelf04}
791: Gotthelf, E.~V.,
792: Halpern, J.~P., Buxton, M., \& Bailyn, C.\ 2004, \apj, 605, 368
793:
794: \bibitem[Haberl(2007)]{2007Ap&SS.tmp...73H} Haberl, F.\ 2007, \apss, 308, 73
795:
796: \bibitem[haberl]{hab} Haberl, F., Sembay, S., Altieri, B. \& Brinkmann, W. 2006, ESASP, 604, 353
797:
798: \bibitem[Haberl et al.(2004)]{2004A&A...419.1077H} Haberl, F., Zavlin,
799: V.~E., Tr{\"u}mper, J., \& Burwitz, V.\ 2004, \aap, 419, 1077
800:
801: \bibitem[heindl]{heindl} Heindl, W.~A. et al., 2004, Proceedings of "X-Ray Timing 2003: Rossi and
802: Beyond", eds. P. Kaaret, F.K. Lamb, \& J.H. Swank (Melville, NY: AIP)
803:
804: \bibitem[ho]{ho} Ho, W~ C~ G., Lai,~D., Potekhin, A.~Y., \& Chabrier, G. 2003, \apj, 599, 1293
805:
806: \bibitem[Ibrahim et al.(2004)]{ibrahim04}
807: Ibrahim, A.~I., et al.\ 2004, \apjl, 609, L21
808:
809: \bibitem[Kaspi \& McLaughlin (2005)]{kaspi05}
810: Kaspi, V.~M., \& McLaughlin, M.~A.\ 2005, \apjl, 618, L41
811:
812: \bibitem[Knight(2006)]{2006ChJAS...6b..41K} Knight, H.~S.\ 2006, Chinese
813: Journal of Astronomy and Astrophysics Supplement, 6, 41
814:
815: \bibitem[Kramer et al. (2003)]{kramer03} Kramer, M., Lyne, A.~G.,
816: Hobbs, G., Lohmer, O., Carr, P., Jordan, C. \& Wolszczan, A.\ 2003,
817: \apjl, 593, L31
818:
819: \bibitem[Li(2006)]{2006ApJ...646L.139L} Li, X.-D.\ 2006, \apjl, 646, L139
820:
821: \bibitem[liu]{liu06} Liu, D.~B. et al. 2006, \apj, 644, 439
822:
823: \bibitem[Lodders 2003]{lod03} Lodders, K. 2003, ApJ, 591, 1220
824:
825: \bibitem[luo]{luo} Luo, Q. \& Melrose, D.~B. 2007, MNRAS, 0704.2906
826:
827: \bibitem[McLaughlin et al.(2003)]{mclaughlin03} McLaughlin, M.~A.,
828: et al.\ 2003, \apjl, 591, L135
829:
830: \bibitem[McLaughlin et al.(2006)]{mclaughlin06}
831: McLaughlin, M.~A., et al.~2006, Nature, 439, 817
832:
833: \bibitem[Mereghetti et al.(2002)]{2002ApJ...574..873M} Mereghetti, S.,
834: Bandiera, R., Bocchino, F., \& Israel, G.~L.\ 2002, \apj, 574, 873
835:
836: \bibitem[Mereghetti et al.(2002)]{2002ApJ...581.1280M} Mereghetti, S., De
837: Luca, A., Caraveo, P.~A., Becker, W., Mignani, R., \& Bignami, G.~F.\ 2002,
838: \apj, 581, 1280
839:
840: \bibitem[Mori et al.(2005)]{2005ApJ...631.1082M} Mori, K., Chonko, J.~C.,
841: \& Hailey, C.~J.\ 2005, \apj, 631, 1082
842:
843: \bibitem[Nakajima et al.(2006)]{2006ApJ...646.1125N} Nakajima, M., Mihara,
844: T., Makishima, K., \& Niko, H.\ 2006, \apj, 646, 1125
845:
846: \bibitem[Pivovaroff et al.(2000)]{2000ApJ...535..379P} Pivovaroff, M.~J.,
847: Kaspi, V.~M., \& Camilo, F.\ 2000, \apj, 535, 379
848:
849: \bibitem[Popov et al.(2006)]{2006MNRAS.369L..23P} Popov, S.~B., Turolla,
850: R., \& Possenti, A.\ 2006, \mnras, 369, L23
851:
852: \bibitem[Protassov et al.(2002)]{2002ApJ...571..545P} Protassov, R., van
853: Dyk, D.~A., Connors, A., Kashyap, V.~L., \& Siemiginowska, A.\ 2002, \apj,
854: 571, 545
855:
856: \bibitem[Rea et al.~2005]{rea05}
857: Rea, N., et al. \ 2005, MNRAS, 361, 710
858:
859: \bibitem[Rea et al.~2007]{rea07}
860: Rea, N., et al. \ 2007, Ap\&SS, 308, 505
861:
862: \bibitem[Reynolds]{rey06} Reynolds, S.~.P. et al. 2006, \apjl, 639, L71
863:
864: \bibitem[Sanwal et al.(2002)]{2002ApJ...574L..61S} Sanwal, D., Pavlov,
865: G.~G., Zavlin, V.~E., \& Teter, M.~A.\ 2002, \apjl, 574, L61
866:
867: \bibitem[Schwope et al.(2007)]{2007Ap&SS.tmp..106S} Schwope, A.~D.,
868: Hambaryan, V., Haberl, F., \& Motch, C.\ 2007, \apss, 106
869:
870: \bibitem[Truemper et al.(1978)]{1978ApJ...219L.105T} Truemper, J., Pietsch,
871: W., Reppin, C., Voges, W., Staubert, R., \& Kendziorra, E.\ 1978, \apjl,
872: 219, L105
873:
874: \bibitem[van Kerkwijk \& Kaplan(2007)]{2007Ap&SS.tmp...74V} van Kerkwijk,
875: M.~H., \& Kaplan, D.~L.\ 2007, \apss, 308, 74
876:
877: \bibitem[Verner et al.(1996)]{1996ApJ...465..487V} Verner, D.~A., Ferland,
878: G.~J., Korista, K.~T., \& Yakovlev, D.~G.\ 1996, \apj, 465, 487
879:
880: \bibitem[Wang et al.(2007)]{2007MNRAS.377.1383W} Wang, N., Manchester,
881: R.~N., \& Johnston, S.\ 2007, \mnras, 377, 1383
882:
883: \bibitem[Weltevrede et al.(2006)]{2006ApJ...645L.149W} Weltevrede, P.,
884: Stappers, B.~W., Rankin, J.~M., \& Wright, G.~A.~E.\ 2006, \apjl, 645, L149
885:
886: \bibitem[Woods et al.(2005)]{2005ApJ...629..985W} Woods, P.~M., et al.\
887: 2005, \apj, 629, 985
888:
889: \bibitem[Woods \& Thompson(2006)]{woods06} Woods, P.~M. \& Thompson,
890: C. \ 2006, "Compact Stellar X-ray Sources",
891: eds. W.~H.~G. Lewin \& M. van~der~Klis, 547
892:
893: \bibitem[Zane et al.(2001)]{2001ApJ...560..384Z} Zane, S., Turolla, R.,
894: Stella, L., \& Treves, A.\ 2001, \apj, 560, 384
895:
896: \bibitem[Zhang et al.(2007)]{2007MNRAS.374.1103Z} Zhang, B., Gil, J., \&
897: Dyks, J.\ 2007, \mnras, 374, 1103
898:
899:
900: \end{thebibliography}
901:
902: \begin{figure}
903: \epsscale{.60}
904: \plotone{f1.eps}
905: \caption{Top: X-ray
906: pulse profile (0.3--5 keV) for 23-ks of EPIC-PN data and 37-ks of MOS1 and MOS2
907: data
908: on PSR~J1819--1458. The dotted line shows the best-fit sinusoid and the vertical dashed line indicates the phase of the peak of the radio pulse.
909: Bottom: radio profile formed from 114 pulses detected in 6 hr of observation
910: at 1.4~GHz with the Parkes telescope in Australia. The non-uniform baseline is likely due to
911: instrumental digitization effects. Both profiles have been folded
912: using the radio ephemeris. In both plots, two cycles of pulse phase are shown.
913: \label{profile}}
914: \end{figure}
915:
916: \begin{figure}
917: \epsscale{0.80}
918: \plotone{f2.eps}
919: \caption{
920: Top: {\it XMM-Newton} EPIC-PN spectrum of PSR~J1819--1458, modeled as an
921: absorbed blackbody. The points
922: indicate the data, while the solid line shows the corresponding
923: best-fit model.
924: Bottom: residuals of the
925: model.}
926: \end{figure}
927:
928: \begin{figure}
929: \epsscale{0.80}
930: \plotone{f3.eps}
931: \caption{The effective area, or total response efficiency, for EPIC-PN
932: with the medium filter versus incident photon energy. Note that there is no instrumental feature in
933: the effective area curve at
934: 1~keV. The calibration feature near 0.5~keV is too weak to explain the feature in our spectrum
935: close to this energy (see Haberl et al. 2006 for more details on {\it XMM-Newton} calibration.)}
936: \end{figure}
937:
938: \begin{figure}
939: \epsscale{0.80}
940: \plotone{f4.eps}
941: \caption{
942: Top: {\it XMM-Newton} EPIC-PN $\nu F_\nu$ spectrum of PSR~J1819--1458, modeled as an
943: absorbed blackbody and Gaussian line (column~2 of Table~1). The points indicate the data, while the solid line shows the corresponding
944: best fit model. Top line shows the absorbed
945: blackbody model component of the fit alone. We have omitted all photons with energies from
946: 0.5--0.53~keV as we believe that the feature in this range is due to the Oxygen edge (see Section~2.2).
947: Middle:
948: residuals of the absorbed blackbody and Gaussian line model. Bottom: residuals of the model without
949: the inclusion of the Gaussian line.}
950: \end{figure}
951:
952:
953: \begin{figure}
954: \epsscale{0.80}
955: \plotone{f5.eps}
956: \caption{Top: {\it XMM-Newton} EPIC-PN (filled circles), MOS1 (black
957: cross), and MOS2 (triangles) spectra for
958: PSR~J1819--1458, modeled as an absorbed blackbody and Gaussian line
959: (column~2 of Table~1). Grey lines show the fitted model. The PN data
960: have been binned with 40 counts per bin and the MOS1 and MOS2 data with 30 counts per bin to show that the best-fit model is not dependent on the spectral binning parameters. We have omitted all photons with energies from
961: 0.5--0.53~keV as we believe that the feature in this range is due to the Oxygen edge
962: (see Section~2.2). Bottom: residuals of
963: the above fits.}
964: \end{figure}
965:
966:
967: \end{document}
968: