astro-ph0702454/ms.tex
1: %%
2: %% This is file `esapub.tex', 
3: %% generated with the docstrip utility. 
4: %% 
5: %% The original source files were: 
6: %% 
7: %% esapub.dtx  (with options: `manual') 
8: %% ============================================ 
9: %% This is the manual describing the usage of 
10: %%      esapub.cls 
11: %% ============================================ 
12: %% Copyright 1999 Patrick W Daly 
13: %% Max-Planck-Institut f\"ur Aeronomie 
14: %% Max-Planck-Str. 2 
15: %% D-37191 Katlenburg-Lindau 
16: %% Germany
17: %% E-mail: daly@linmpi.mpg.de 
18: %%
19: %% ------------------------------------------------- 
20: \ProvidesFile{esapub.tex} 
21:           [2001/04/25 1.1 (PWD)] 
22: \documentclass[a4paper,twocolumn]{esapub2005} % European paper 
23: \pagestyle{empty} 
24:  
25: % introduce this option for the ESA publications style 
26: \bibliographystyle{alpha} 
27: 
28: 
29: \def\pl{{\sc pl}} 
30: \def\compps{{\sc compps}} 
31: \def\kte{kT_{\rm e}} 
32: \def\taut{\tau_{\rm T}} 
33: \def\chiq{$\chi^2$} 
34: \def\ktin{kT_{\rm in}} 
35: \def\bb{{\sc bb}} 
36: \def\ktbb{kT_{\rm bb}} 
37: \def\rbb{R_{\rm bb}} 
38: \def\gs{ {\sc Gaussian}} 
39: \def\nh{ {$N_{\rm H}$}} 
40: \def\dbb{ {\sc diskbb}} 
41: \def\be{\begin{equation}} 
42: \def\ee{\end{equation}} 
43: \def\beq{\begin{eqnarray}} 
44: \def\eeq{\end{eqnarray}} 
45: \def\gax {\ifmmode{_>\atop^{\sim}}\else{${_>\atop^{\sim}}$}\fi}  
46: \def\IGR{IGR~J00291+5934}
47: \def\SAX{SAX~J1808.4-3658}
48: \def\J1751{XTE~J1751-305}
49: \def\hete{HETE~J1900.1-2455}
50: \def\1807{XTE~J1807-294}
51: 
52: 
53: 
54: 
55: 
56: 
57:  
58: \usepackage{times} 
59: \usepackage{natbib} 
60: \usepackage{graphicx}
61:  
62: \title{Accreting X-ray millisecond pulsars observed with INTEGRAL} 
63: \author{Maurizio Falanga} 
64: \affil{CEA Saclay, DSM/DAPNIA/Service d'Astrophysique, F-91191, Gif
65: sur Yvette, France; e-mail: mfalanga@cea.fr} 
66: \affil{Unit\'e mixte de recherche Astroparticule et  
67: Cosmologie, 11 place Berthelot, 75005 Paris, France}
68:  
69: \newcommand{\btx}{\textsc{Bib}\TeX} 
70: \newcommand{\filename}{esapub} 
71:  
72: \begin{document} 
73:  
74: \keywords{pulsars: individual (XTE J1807-294, IGR
75: J00291+5934, HETE J1900.1-2455) -- starts: neutron -- X-ray: binaries} 
76:  
77: \maketitle 
78:  
79: \begin{abstract} 
80: I review the properties of three X-ray accreting millisecond pulsars
81: observed with INTEGRAL.  Out of
82: seven recently discovered accretion-powered pulsars (one discovered by
83: INTEGRAL), three were observed with the INTEGRAL satellite up to 300 keV. 
84: Detailed timing and spectral results will be
85: presented, including data obtained during the most recent outburst of
86: the pulsar HETE J1900.1-2455. Accreting X-ray millisecond pulsars are key
87: systems to understand the 
88: spin and accretion history of neutron stars. They are also a good
89: laboratory in which to study the source spectra, pulse profile, and
90: phase shift between 
91: X-ray pulses in different energy ranges which give additional information
92: of the X-ray production processes and emission environment.
93: \end{abstract} 
94:  
95: \section{Introduction} 
96: 
97: Many low-mass X-ray binaries (LMXB) consist of a neutron star
98: accreting from a low mass companion star ($<$ 1 $M_{\odot}$).  
99: Accreting matter may spin up the neutron
100: star (NS); therefore, one of the possible endpoints of
101: the evolution of a low-mass X-ray binary is expected to be a
102: millisecond pulsar, i.e. a
103: rapidly spinning NS with a rather weak, $\sim 10^{8}$ Gauss, surface
104: magnetic field. Although evidence for  rapidly spinning NS in LMXBs
105: was inferred from the burst oscillations that were seen during type I
106: X-ray bursts in several systems \citep[see][for a review]{stroh01},
107: the detection of millisecond pulsations in persistent emission remained
108: elusive for many years until the discovery of the first accreting
109: millisecond pulsar by \citet{wvdk98}. Since that
110: time, a total of seven accreting MSPs have been detected.
111: 
112: All of the accreting MSPs are X-ray transients; they spend most of
113: the time in a quiescent phase, with X-ray luminosities of order
114: of $10^{31}-10^{33}$ erg s$^{-1}$. They sometimes show X-ray outbursts
115: reaching X-ray luminosities of $10^{36}-10^{37}$ erg s$^{-1}$,
116: during which coherent pulsations are observed with
117: frequencies in the range between 180 and 435 Hz  \citep[see reviews
118: by][]{w05,p06}. This frequency is interpreted as the NS rotation
119: frequency given by a hot spot (or spots) in an atmospheric layer of
120: the rotating NS \citep{Chakrabarty03}. MSPs represent a new
121: class of objects connecting  accretion powered X-ray pulsars with
122:  rotation powered millisecond radio pulsars.  
123: 
124: MSP  energy spectra are successfully fitted   by a two-component model 
125: consisting of a multicolor blackbody soft X-ray emission and a 
126: Comptonized spectrum for the hard X-ray emission. 
127: The soft thermal component could be associated with the 
128: radiation from the accretion disc and/or the heated NS surface around the 
129: shock \citep[see e.g.][]{gp05,p06}. The hard emission is likely to be
130: produced by thermal Comptonization in the hot accretion shock on the
131: NS surface \citep{gdb02,pg03} with the seed photons coming from the
132: stellar surface. The observed hard spectra are similar to the spectra
133: observed from atoll sources in their hard, low-luminosity state \citep{b00}.
134: A representative spectrum of the three observed MSP (XTE~ J1807-294
135: \citep{mfa05}, IGR~J00291+5934  \citep{mfb05}, and HETE J1900.1-2455
136: \citep{mf06}) with INTEGRAL is shown in Figure 1, 2 and 3.  
137: 
138: Recent years have shown INTEGRAL gave additional contributions to the
139: study of MSPs. Our timing analysis of
140: IGR~J00291+5934 showed for the first time that INTEGRAL/ISGRI is
141: capable of detecting  the pulse profile of a 1.67 ms pulsar up to 150
142: keV. This allowed us to study the pulsed fraction up to this high energy. We
143: confirmed for the first time the increase of the pulsed fraction with
144: energy in an accretion powered MSP, explained by the action of the
145: Doppler effect on the exponentially cutoff Comptonization spectrum
146: from the hot spot or from a componization model where we account the
147: pulsed fraction to be produced in a corona cloud. We also measured
148: soft time lags with a complex energy dependence. Similarly to SAX
149: J1808.4-3658 and XTE~J1751-350, the time lags increase rapidly with
150: energy until ~10 keV. However, in IGR~J00291+5934, the
151: time lags first increase and then decrease slightly, saturating above
152: 15 keV, and possibly reaching zero around 50 keV. This is a intriguing
153: result, and if confirmed or observed in another object could be a
154: serious challenge to any time-lag model. Additionally, with RXTE we
155: measured an increase of the NS spin period during accretion for the
156: first time. This provided a strong
157: confirmation of the theory of 'recycled' pulsars in which the old
158: neutron stars in LMXBs become millisecond radio pulsars through
159: spin-up by transfer of angular momentum by the accreting material
160: \citep{mfb05}. 
161: 
162: Analysis of the recent ToO observation of HETE~J1900.1-2455 is also
163: discussed. We observe in this object Type I X-ray bursts showing,
164: at high energy, evidence of photospheric radius expansion in the burst
165: profile. Using JEM-X/ISGRI data at high energy, and assuming the bolometric
166: burst peak luminosity during photospheric radius expansion to be
167: saturated at the Eddington limit, we measure
168: the source distance to be around 5 kpc. A very interesting result is
169: that the pulsations at the  spin frequency are not always observed in
170: this source. 
171: 
172: \section{X-ray spectra} 
173: 
174: 
175: \begin{figure}[ht]
176: \centering 
177: \includegraphics[width=0.65\linewidth,angle=-90]{fig_5.ps} 
178: %\vspace{4cm} 
179: \caption{Simultaneous INTEGRAL, XMM-Newton and RXTE 
180:         spectra of XTE~J1807--294 fitted with an absorbed disk black body, {\sc
181: 	diskbb}, plus {\sc compps} model. The EPIC-pn and MOS2 spectra in the 
182:         0.5--10 keV range and PCA/HEXTE in the 3--200 keV and
183: 	IBIS/ISGRI in the 20--200 keV range are shown. The {\sc
184: 	diskbb} model is shown by a dotted curve, the dashed curve
185: 	gives the {\sc compps} model, while the total spectrum is
186: 	shown by a solid curve. The lower panel presents the residuals.} 
187: \label{fig:XTEJ1807} 
188: \end{figure} 
189:  
190: \begin{figure}[ht] 
191: \centering 
192: \includegraphics[width=0.65\linewidth,angle=-90]{3472fig3.ps} 
193: %\vspace{4cm} 
194: \caption{The unfolded spectrum of \IGR\ fitted with an absorbed
195: 	{\sc compps} model. The data points correspond to the JEM-X
196: 	(5--20 keV)  and ISGRI (20--200 keV) spectra,
197: 	respectively. The total spectrum  of the model is shown by a
198: 	solid curve. The lower panel presents the residuals between
199: 	the data and the model.}  
200: \label{fig:IGRJ00201}
201: \end{figure} 
202:  
203: \begin{figure}[ht] 
204: \centering 
205: \includegraphics[width=0.65\linewidth,angle=-90]{f2.ps} 
206: %\vspace{4cm} 
207: \caption{The unfolded spectrum of HETE~J1900.1--2455 fitted with a
208: 	{\sc compps} model plus a {\sc bb} and a gaussian line. The
209: 	data points correspond to the PCA (3--22 keV), JEM-X (3--22
210: 	keV), HEXTE (16--90 keV) and the ISGRI (20--300 keV) spectra,
211: 	respectively. The blackbody model is shown by a  dot-dashed
212: 	curve, the dotted curve gives the {\sc compps} model, the
213: 	dashed curve is the line, the total spectrum is shown by the
214: 	solid curve. The lower panel presents the residuals between
215: 	the data and the model. 
216: }
217: \label{fig:HeteJ1900} 
218: \end{figure} 
219:  
220: For the INTEGRAL observed MSP sources the high energy spectrum could be
221: described with a photon index of $\sim$ 1.8--2.0 and a cut-off energy at $\sim$
222: 80--100 keV. However, the model does not describe the spectrum below
223: 15 keV well, which requires a more complex description. 
224: MSP  energy broad-band spectra are successfully fitted by a
225: two-component model consisting of a multicolor blackbody soft X-ray
226: emission and a  
227: Comptonized spectrum, for the hard X-ray emission. The soft thermal
228: component could be associated with the radiation from the accretion
229: disc and/or the heated NS surface around the shock. The hard emission
230: is likely to be produced by thermal Comptonization in the hot
231: accretion shock on the NS surface with the seed photons coming from the
232: stellar surface.  The spectra are best fitted with the thermal
233: Comptonization model {\sc compps} in the slab geometry 
234: \citep{ps96}. The main model parameters are the Thomson optical depth 
235: $\tau_{\rm T}$ across the slab, the electron temperature 
236: $kT_{\rm  e}$, the soft seed photon temperature $kT_{\rm seed}$, and  
237: the inclination angle $\theta$ between the slab normal and the line of sight. 
238: The seed photons are supposed to be injected from the bottom of the
239: slab. The soft thermal emission is fitted by a simple blackbody, {\sc bb}, 
240: or a multi-temperature disc blackbody, {\sc dbb}, model \citep{mitsuda84}.  
241: The best fit parameters for the three observed MSP with INTEGRAL are
242: reported in Table \ref{table:spec}. In Fig. \ref{fig:XTEJ1807},
243: \ref{fig:IGRJ00201},  and \ref{fig:HeteJ1900}, we show the unfolded spectrum 
244: and the residuals of the data to the {\sc bb} or {\sc dbb} plus 
245: {\sc compps} model.  
246: 
247: The spectra of \1807, \IGR, and \hete\ are well described by 
248: a combination of thermal Comptonization  and a disk black body or
249: simple  black body. 
250: The hard spectral component contributes most of the observed flux (70--80
251: per cent), even though a soft component (disk black body) is required
252: by the data. If the accretion disk is truncated at the 
253: magnetosphere radius where the material can accrete along the magnetic 
254: field towards the pole of a NS, the matter over the 
255: pole is heated by a shock to a temperature of $\sim$ 30--50 keV. The hot spot 
256: at the surface with temperature $\sim$ 0.8--1.5 keV gives rise to the seed 
257: photons for Comptonization in the hot plasma. Since the hard X-ray 
258: emission is pulsed (see Sec. 3.1), a fraction of it must 
259: originate from the regions confined by the magnetic field. The most 
260: obvious source of hard X-rays is the place where material collimated by 
261: the magnetic fields impacts  the surface. 
262:  
263: For \1807, if we consider the inclination of the system to be
264: $60^{\circ}<i<83^{\circ}$, this allows us to determine the inner disk radius
265: which lies  in the range 20--40 km (for the distance of 8 kpc). 
266: The apparent area of the seed
267: photons is $A_{\rm seed} \sim 26 (D/8 \ {\rm kpc})^{2} {\rm km}^{2}$,
268: which is also what expected from a hot spot on a NS surface. 
269: Many characteristics of the source are similar to those observed in
270: other MSPs, SAX~J1808.4--3658 \citep{gdb02,pg03} and XTE~J1751--305
271: \citep{gp05}. For \IGR\ no additional thermal-like
272: component (either blackbody or disk blackbody) was required by the
273: fit, very likely because the bulk of its emission occurs below 3 keV,
274: outside the covered energy range. However, the apparent area of the seed 
275:   photons, which turns out to be 
276: $A_{\rm seed}\sim 21 (D/5\ {\rm kpc})^{2}$ km$^2$, could correspond to
277: a  hot spot radius of $\sim2.5$ km during the INTEGRAL ToO outburst phase.  
278: 
279: Only \hete's spectrum differs from other MSP in requiring thermal 
280: soft X-ray emission with nearly double the temperature. 
281: We could not distinguish between the multi-temperature blackbody and 
282: single blackbody models, as both gave  comparable parameters and $\chi^2$.   
283: However, for a distance of 5 kpc,  the disk blackbody gives 
284: an  inner disk radius, $R_{\rm in} \sqrt{\cos\,i} = 2.6$ km, 
285: smaller than the expected NS radius. For the blackbody emission, the
286: fit implies the apparent area of the emission region  
287: $A_{\rm seed} \sim 14$ km$^2$, which could be consistent  with the  
288: heated NS surface around the accretion shock \citep{pg03,gp05}.
289: From our spectral fits we infer that this emission is not
290: likely produced in a multi-temperature accretion 
291: disk but  more likely arises from thermal emission at the NS
292: surface.
293:  
294: 
295: We note that the product $\taut\times\kte$ is very close 
296: in the three sources. The spectral shape and the product $\taut\times\kte$ 
297: are very stable during the outburst as is observed 
298: in other sources too \citep[e.g.][]{gp05}. The constancy of the
299: spectral slopes  during the outbursts and their extreme similarity 
300: in different MSPs can be used as an argument that the 
301: emission region geometry does not depend on the accretion rate. 
302: If the  energy dissipation takes place in a hot shock, while the cooling 
303: of the electrons (that emit X/$\gamma$-rays via thermal Comptonization) 
304: is determined by the reprocessing of the hard 
305: X-ray radiation at the neutron star surface (so called two-phase model, 
306: the spectral slope is determined by the energy balance in the hot phase 
307: and  is a function of the geometry. At constant geometry (e.g. slab), 
308: the temperature depends on the optical depth, but $\taut \times \kte$ 
309: is approximately constant. 
310:  
311: \begin{table*}[htb] 
312: \begin{center} 
313: \caption{\label{table:spec}Best-fit spectral parameters with  
314:  {\sc compps} + {\sc bb} (or {\sc diskbb}) model.}   
315: \begin{tabular}{llll} 
316: \hline 
317: \noalign{\smallskip}  
318:  & XTE~J1708-294 &  IGR~J00291+5934& HETE~J1900.1-2455 \\ 
319: \hline 
320: \noalign{\smallskip}  
321: $N_{\rm H} (10^{22} {\rm cm}^{-2})$ & $0.56$ (f) & $0.28$ (f) & $0.16$ (f)\\ 
322: $kT_{\rm in}$ or $kT_{\rm BB}$(keV) & -- & 0.43$^{+0.04}_{-0.04}$ & 
323: 0.8$^{+0.02}_{-0.02}$\\ 
324: $R_{\rm in} \sqrt{\cos i}^{\ a}$ (km) & $13.4^{+2.2}_{-1.3}$ & -- & --\\ 
325: $R_{\rm bb}^{\ a}$ (km)      & -- & -- & $4.8^{+0.7}_{-0.6}$\\ 
326: %$E_{\rm Fe}$ (keV) & -- & -- & 6.42$_{-0.15}^{+0.1}$ \\ 
327: %$\sigma_{\rm Fe}$ (keV)  & -- & -- & 0.5 (f) \\ 
328: $kT_{\rm e}$ (keV) & $37.2^{+28}_{-10}$ & $49^{+2}_{-6}$ & $27.9^{+1.8}_{-1.4}$\\ 
329: $kT_{\rm seed}$ (keV)    & $0.75^{+0.04}_{-0.04}$ &
330: $1.49^{+0.16}_{-0.32}$ & $1.4^{+0.16}_{-0.32}$\\  
331: $\tau_{\rm T}$ & $1.7^{+0.5}_{-0.8}$ & $1.12^{+0.04}_{-0.07}$ & $2.0^{+0.06}_{-0.1}$\\ 
332: $A_{\rm seed}^{a}$ (km$^2$)  & $26^{+23}_{-12}$  & 
333: $20.7^{+12.6}_{-4.5}$ & $14.2^{0.3}_{0.3}$\\ 
334: cos $\theta $  & $0.79^{+0.07}_{-0.06}$  & $0.6^{+0.06}_{-0.09}$  &
335: 0.59$^{+0.05}_{-0.07}$ \\ 
336: $L_{\rm bol}^{a}$ ($10^{36}$ erg s$^{-1}$) & 3.6 & 3.7 & 4.9\\ 
337: \noalign{\smallskip}  
338: \hline  
339: \noalign{\smallskip}  
340: %\multicolumn{4}{l}{$^{a}$  Assuming a distance of 5 kpc.} 
341: \end{tabular}  
342: \end{center} 
343: \end{table*} 
344: 
345: 
346: \section{Timing}
347: 
348: 
349: The timing analysis of
350: IGR~J00291+5934 showed for the first time that INTEGRAL/ISGRI is
351: capable of detecting  the pulse profile of a 1.67 ms pulsar up to 150
352: keV. This allowed us to study the pulsed fraction up to this high energy. We
353: confirmed for the first time the increase of the pulsed fraction with
354: energy in an accretion powered MSP. This can be explained by two
355: models:  the action of the
356: Doppler effect on the exponentially cutoff Comptonization spectrum
357: from the hot spot, or  a componization model where  the
358: pulsed fraction is  produced in a corona cloud. We also measured
359: soft time lags with a complex energy dependence. Similarly to SAX
360: J1808.4-3658 and XTE~ J1751-350, the time lags increase rapidly with
361: energy until ~10 keV. However, in IGR~~J00291+5934, the
362: time lags first increase and then decrease slightly, saturating above
363: 15 keV, and possibly reaching zero around 50 keV, see Fig. 6.
364: 
365: \subsection{Pulsed fraction of X-rays}
366: 
367: \begin{figure}[ht] 
368: \centering 
369: \includegraphics[width=0.53\linewidth,angle=90]{f1.eps} 
370: %\vspace{4cm} 
371: \caption{This cartoon illustrates the different emission patterns 
372:   responsible for the time lags of the pulsed emission.  
373:   $\theta_{\rm e}^{\rm hot}$ and $\theta_{\rm e}^{\rm ref}$  are the
374:   dimensionless temperature of the accretion column and reflector,
375:   respectively, and $C_{\rm ill}\sim 0.1$ is the disk illumination fraction. 
376: Soft time lag of the pulsed emission is the result of downscattering
377:   of hard X-ray photons in the relative cold plasma of the disk. 
378: A fraction of hard X-ray photons $1-C_{\rm ill}$  directly seen by the
379:   observer are  upscattered soft  photons coming from NS and  the
380:   disk. We account for the pulsed flux (pulsed fraction) as being
381:   produced in the  corona cloud.}
382: \label{fig:geom} 
383: \end{figure} 
384: 
385: We found for \IGR\ using the RXTE and INTEGRAL/ISGRI data that the
386: pulsed fraction gradually increases with energy  
387: from $\sim 6\%$ at 6 keV to $\sim 12-20 \%$ near 100 keV. This is the 
388: first time such behaviour has been measured for any of the
389: accretion-powered MSPs.  
390:  
391: Since MSPs are rapidly rotating, a first interpretation is given 
392: through a Doppler boosting model which affects the fast rotating spot
393: emitting  pattern as a black body.  
394: When the spot moves towards the observer, the emission 
395: increases, while for a spot moving away, the flux drops. 
396: The Doppler factor reaches the maximum a quarter of a period before the 
397: peak of the projected area,  shifting the emission peak towards an
398: earlier phase. The observed flux due to the Doppler effect varies 
399: as the Doppler factor to the power $(3+\Gamma)$ 
400:  \citep{pg03,vp04}, where the photon index $\Gamma$ could be a function 
401: of energy. If the Doppler factor varies around 1 with 2\% amplitude, 
402: we get 10\% variability. 
403:  
404: The Comptonized spectrum can be approximated as 
405: \be 
406: F_E \propto E^{-(\Gamma_0-1)} \exp\left( -[E/E_{\rm c}]^{\beta}\right) , 
407: \ee 
408: where  $E_{\rm c}\sim \kte$ is the energy of the cutoff  and parameter 
409: $\beta\sim2$ describes its sharpness. 
410: The local photon index is then 
411: \be 
412: \Gamma(E)\equiv 1 - \frac{\mbox{d} \ln F_E }{\mbox{d} \ln E} = 
413: \Gamma_0 + \beta (E/E_{\rm c})^{\beta} . 
414: \ee 
415: At low energies, $\Gamma\approx\Gamma_0$, and 
416: rms (or pulsed fraction) is a very weak function of energy. 
417: Close to the cutoff, the spectral index rapidly increases and 
418: the pulsed fraction should grows with energy, as observed (see 
419: Fig.~\ref{fig:PF}a). 
420: The Comptonization models predict  softening of the total spectrum 
421: with simultaneous hardening of the pulsed spectrum at higher energies. 
422:  
423: 
424: \begin{figure}[ht] 
425: \centering 
426: \includegraphics[width=0.9\linewidth]{f3.eps} 
427: %\vspace{4cm} 
428: \caption{The observed  energy dependent pulsed fraction of
429: (a) \IGR\, (b) XTE J1751-305, and (c) SAX J1808.4-3658  along with the
430:   best-fit comtonization model. 
431: }
432: \label{fig:PF} 
433: \end{figure} 
434: 
435: 
436: On the other hand, the pulsed fraction of \IGR\ vs
437: energy can be also explained by the energy 
438: dependent  electron cross-section $\sigma_e(E) =\sigma_{\rm  T}(1-2z)$ 
439: and consequently by Compton cloud  optical depth as a function energy   
440: \begin{equation}
441: \tau_{\rm cl}(E)=\tau_{{\rm T, cl}}(1-2z),
442: \label{electron_tau}
443: \end{equation}
444: where $z=E/m_ec^2$ is a dimensionless photon energy.
445: In Figure \ref{fig:PF} we show the energy dependent pulsed fraction of
446: (a) \IGR, (b) \J1751\ and (c) \SAX.
447: In fact, we assume that the accretion column is embedded in
448: Compton cloud  as shown in Figure \ref{fig:geom}  \citep[see also][for
449:  the geometry details]{TCW}.   
450: % \citep[see][for the details of  geometry]{tok}. 
451: Because $\tau_{\rm cl}(E)$ decreases with energy,  at higher
452: energies a larger fraction of the pulsed direct hard X-ray radiation   
453: \begin{equation}
454: A_{\rm rms,es}(E)=A_{\rm rms,es}(0)\exp[-\tau_{\rm cl}(E)]
455: \label{A-fraction}
456: \end{equation}
457: originating  in the accretion column can escape to the observer. 
458: 
459: A different scenario of the energy dependent amplitude  formation
460: would be if 
461: one can assume that there is no electron (Compton) cloud between the
462: accretion column (where the Comptonization spectrum is formed) and the
463: observer, but the energy dependence of the amplitude is formed as a
464: result of   upscattering.    
465: The soft photons upscattered  off of hot electrons (of the accretion
466: column)   increase their energy  with a number of scatterings. 
467: On the other hand, the amplitude  of the pulsed radiation exponentially
468: decreases 
469: with a number of scatterings $A_{\rm rms}(u)=A_{\rm rms}(0)\exp(-\beta u)$,
470: where $\beta$ is the inverse of an average number of scatterings
471: \citep[see e.g.][]{ST80} and  consequently with a energy of the
472: upscattered photon 
473: \begin{equation}
474: A_{\rm rms,up}[u(E)]=A_{\rm rms,up}(E_s)(E/E_s)^{-\alpha_{\rm cl}}
475: \label{A-upscatter_fraction}
476: \end{equation}
477:  (where $\alpha_{\rm cl}=\beta/4\theta$ is approximately the energy  index
478:  of the Comptonization spectrum).  A characteristic seed (disk and NS)
479:  photon energy  is $E_s<  5$ keV  and consequently,  all photons at
480:  higher energies ($>5$ keV)  are produced by upscattering.  The
481:  amplitude  of upscattered photons  (in which the energy is higher 5 keV)
482:  should decrease with energy.     
483: 
484: In the general case,  some fraction of the pulsed  soft photons,
485: $A_{\rm rms,up}$,  upscatter  off hot electrons in the accretion column on the
486: way out as another fraction of the pulsed photons, $A_{\rm rms,es}$, forming
487: the hard X-ray tail escape to the observer passing through the Compton cloud.  
488: %  one should take into account both of these effects 
489: %combined effect of photon upscattering and escape 
490: %on the amplitude  of the pulsed radiation.  
491: In this  case a following combination of Eqs. (\ref{A-fraction}
492: and \ref{A-upscatter_fraction}) 
493: \begin{equation}
494: A_{\rm rms}(E)= A_{\rm rms,up}(E/E_s)^{-\alpha_{\rm cl}} + A_{\rm rms,es}e^{-\tau_{\rm cl}(E)} 
495: \label{A-general_fraction}
496: \end{equation}
497: leads us to the formula of the emergent pulsed amplitude.
498: 
499: 
500: In Figure \ref{fig:PF}  we present  the best-fit model (see
501: Eq. \ref{A-general_fraction})  along with data points.  We  fit only
502: the data points which correspond to  energies higher than the seed
503: photon energy.   We found for (a) \IGR\ the Compton cloud 
504: optical depth to be $\tau_{\rm T, cl} = 3\pm0.2$ and $A_{\rm
505:   rms,es} = 98^{+2}_{-6}$ \%,  $A_{\rm rms,up} = 5.2\pm0.9$ \%. For (b)
506:   $\tau_{\rm T, cl} = 0.25\pm0.09$ and $A_{\rm 
507:   rms,es} = 3.5\pm0.8$ \%,  $A_{\rm rms,up} = 1.8\pm1.1$ \% and (c)
508:     $\tau_{\rm T, cl} = 0.4\pm0.2$ and $A_{\rm rms,es} = 5.9\pm6$ \%,
509:       $A_{\rm rms,up} = 2.5\pm1.2$ \%, respectively. For all of the fits
510:       $\alpha_{\rm cl}$ was $\sim0.8$.
511: 
512: 
513:   
514: \subsection{Time lag}
515: 
516: \begin{figure}[ht] 
517: \centering 
518: \includegraphics[width=0.9\linewidth]{f2.eps} 
519: %\vspace{4cm} 
520: \caption{The measured soft time lag of the pulse profile versus energy
521:  (crosses) with respect to the first energy channel. The
522:  best-fit curve using the Comptonization model (see Eq. 7)
523:  is shown with the solid line.   The dot-dashed lines in panel (a)  correspond to the upper and lower limits of the electron number densities of the
524:  Comptonization emission area, $n_{\rm e}^{\rm hot}$ and  of  the
525:  reflector, $n_{\rm e}^{\rm ref}$   in \IGR.
526:    The panels (b) and (c)  are related  to \J1751\ and (c) for \SAX.
527:  The dot-dashed lines correspond to the upper and lower
528:  limits of  $n_{\rm e}^{\rm ref}$.
529: }
530: \label{fig:TL} 
531: \end{figure} 
532: 
533: Additional information for the X-ray production processes and
534: emission environment can be obtained by studying the pulse profile
535: and phase shift between X-ray pulses at different energy ranges. For
536: \SAX\ and later for \J1751\ it was found that the low-energy pulses
537: lag behind the high-energy pulses (soft phase/time lags) monotonically
538: increasing with energy  and saturating at about 10--20 keV
539: \citep{CMT98,Ford00,gp05}.
540: 
541: First, this phase/time lag effect was interpreted  as a result
542: of the photon delay due to downscattering of  hard X-ray photons in the
543: relatively cold plasma of the disk or NS surface \citep{CMT98,TCW}.   
544: They argued that the photon time lags were an intrinsic signature of
545: interaction of the Comptonized radiation with the NS and accretion disk
546: plasma. Moreover, the absolute values of time lags (about hundreds
547: $\mu$s) are consistent with the electron scattering time scale $t_{\rm
548:   C}=\tau_{\rm T}(L/c)$.  The effective Thomson  optical depth of the cold
549: reflector $\tau_{\rm T}=n_e\sigma_{\rm T}L$ is about a few and
550: typical sizes of the NS photosphere and half-width of the disk $L$ are
551: of the 
552: order of $10^{6}$ cm. On the other hand, \citet{pg03} suggested that
553: the lags may be produced by a combination of different angular
554: distribution of the radiation components  and relativistic effects.
555: The observed spectrum of \IGR\ consists of a black body 
556: from the neutron star surface and a component produced by 
557: Comptonization of these seed photons in the hot electron  region, 
558: presumably a shock which can be represented as a plane-parallel slab. 
559: The angular distributions of the black body and Comptonized 
560: photons emitted by the slab are significantly different. 
561: The difference in the emission patterns causes the two components 
562: to show a different variability pattern as a function of the pulsar phase, 
563: with the hard Comptonized component leading the soft black body 
564: component. However, the  soft lags found  up to $\sim100$ keV in the 1.67 ms
565: accreting MSP \IGR, show a more complex energy dependence \citep{mfa05}. 
566: 
567: Recently, \citet{MFLT06} derived a  Comptonization model for the
568: observed properties of the energy dependent soft/hard time lags. This
569: accounts for  the soft lag by downscattering of hard
570: X-ray photons in the relatively cold plasma of the disk or neutron
571: star surface. A fraction of soft X-ray photons 
572: coming from  the disk or neutron star surface are upscattered off hot
573: electrons of accretion column  and this effect leads  to the hard lags
574: as a result of thermal Comptonization of the  soft photons. The model
575: allows  the observed soft and hard 
576: lags  to be reproduced by the downscattered and upscattered radiation
577: as a function of 
578: the electron number densities of the reflector, $n_{e}^{\rm ref}$, and
579: the accretion column,  $n_{ e}^{\rm hot}$. In the case of the
580: accretion-powered millisecond pulsars \IGR, \J1751,  and \SAX\ the
581: observed time lags agree very strongly with the model (see
582: Fig. \ref{fig:TL}).  The resulting time lags are seen as a linear
583: combination of the 
584: positive (hard) time lags formed in the Comptonization emission area and
585: the negative (soft) ones formed in accretion disk and NS photosphere as a
586: result of reflection of the hard radiation (see Fig. \ref{fig:geom}): 
587: $$
588: \Delta t  =  -\frac{C_{\rm ill}}{\sigma_{\rm T}n_{\rm e}^{\rm ref} c} \times
589: $$
590: \begin{equation}
591: %\nonumber \\ 
592: %& &\!\!\!
593: \biggl[\frac{1}{4\theta_{\rm ref}}\ln
594: %\Biggl|
595: \frac{1-4\theta_{\rm
596:       ref}/z}{1-4\theta_{\rm ref}/z_{*}}
597:  %     \Biggr|\!\!
598:       -\frac{n_{\rm
599:   e}^{\rm ref}}{n_{\rm e}^{\rm hot}}\frac{1-C_{\rm ill}}{C_{\rm ill}}
600: \frac{\ln(z/z_{*})}{\ln[1+(3+\alpha)\theta_{\rm hot}]}\Biggr]
601: \label{res_timelag}
602: \end{equation}
603: where $n_{\rm e}^{\rm ref}$ is the electron number density of the reflector,
604: $n_{ e}^{\rm hot}$ is the electron number density of the
605: Comptonization emission area 
606: (accretion column) and $\theta_{\rm ref}=kT_e^{\rm ref}/(m_{e} c^2)$ is a
607: dimensionless temperature of the reflector.  
608: We assume a typical value of $\theta_{\rm ref} < 0.7$ keV/511 keV.   
609:  $kT_{\rm e}^{\rm hot}$ and $\alpha$  are the best-fit parameters for the hot
610: plasma temperature and spectral index of the  Comptonization spectrum,
611: and $\theta_{\rm  hot}=kT_{ e}^{\rm hot}/(m_{ e} c^2)$. 
612: 
613: We also assume that the seed photon energy is near 
614: the lowest energy of the downscattered photons, i.e. $z_s\approx z_{\ast}$.   
615: An observational value of $E_{\ast}$  is  about  3 keV, and we
616: assume that the  value of  the illumination factor $C_{\rm ill}$ is
617: about 0.1 \citep[see for details][]{MFLT06}. To examine the predicted
618: time lag using the Comptonization model 
619: Eq. (\ref{res_timelag}), we use the observed pulse phase lag data
620: and best-fit spectral parameters  $kT_{\rm e}^{\rm hot}$  and
621: $\alpha$ of the accreting MSPs \IGR, \J1751, and \SAX. The \SAX\ data
622: were taken from \citet{CMT98,Ford00} and represented in
623: Figure \ref{fig:TL} by open circles and filled squares, respectively. The
624: data for \J1751 and  \IGR\ were taken by  \citet{gp05,mfa05}. For the reflector
625: temperature we used  $kT_{\rm e}^{\rm ref} = 0.4$ keV for all  fits.
626: The  model to fit the time lags in \J1751\ and \SAX\ 
627: has only one free parameter (see \citep{MFLT06}), the number
628: density of the ``cold'' reflector $n_{\rm e}^{\rm ref}$. Presumably, in \J1751\
629: and \SAX\ the density of the Comptonization region is much higher than
630: that of the ``cold'' reflector, i.e. $n_{\rm e}^{\rm hot}\gg n_{\rm
631:   e}^{\rm ref}$. It is not surprising that the seed photons are
632: Comptonized (upscattered) in the very dense plasma of the accretion column
633: \citep[see e.g.][]{BS76,BW06}. The best-fit values are $n_{\rm e}^{\rm
634:   ref}= 6.3\times10^{19}$ cm$^{-3}$ and $3.2\times10^{19}$ cm$^{-3}$
635:   for \J1751\ and \SAX, respectively. The fits of the time lag
636: data for \IGR\ provide us the best-fit values of the ``cold'' and hot
637: plasma  densities. Both the positive (upscattering) and negative
638: (downscattering) time lags  contribute to the apparent time lags
639: because $n_{\rm e}^{\rm ref}= 6.9\times10^{18}$ cm$^{-3}$ and $n_{\rm
640:   e}^{\rm hot}=  2.1\times10^{18}$ cm$^{-3}$ are of the same order of
641: magnitude.     
642: 
643: The time lag  data of  \IGR, \J1751, and \SAX\ are collected in
644: different time intervals lasting from hours to days
645: \citep{CMT98,Ford00,gp05,mfa05}. However, a hydrodynamical 
646: (density perturbation) time,  $t_{hydro}$, in the
647: innermost part of X-ray NS source is on the order of the ratio of the
648: NS radius to the sound speed, namely  $t_{hydro}\sim R_{\rm NS}/c_{\rm
649: sound}\gax 0.1$ s. 
650: Thus, during the data collection periods the densities of the
651: surrounding plasma can vary. As a result, the time lags also
652: change because they are very sensitive to density variations (see
653: Eq. \ref{res_timelag}). In fact,  these density variations can be inferred
654: from the time lag data. One can make a band between two curves of time
655: lags vs energy which contain all of the time lag data points (see
656: Fig. \ref{fig:TL}). This allows  constraints to be put on $n_{\rm
657:   e}^{\rm ref}$ and $n_{\rm e}^{\rm hot}$.  
658: The lower and upper curves in Figure \ref{fig:TL}
659: correspond (a) to $n_{\rm e}^{\rm ref}=  (6.1-8.0)\times10^{18}$ cm$^{-3}$
660: and  $n_{\rm e}^{\rm hot} = (1.6-2.6)\times10^{18}$ cm$^{-3}$, respectively.
661: For the sources (b) \J1751\ and (c) \SAX\, the density variations are 
662: $n_{\rm e}^{\rm ref}= (6.0-6.6)\times10^{19}$ cm$^{-3}$ and  $n_{\rm
663: e}^{\rm ref}= (2.9-3.6)\times10^{19}$ cm$^{-3}$.   
664: Thus, the plasma density of the ``cold'' reflector can change as much as
665: 10 \% during the entire data collection.  
666: 
667: 
668: Soft lags are observed only if $n_e^{ref}\ll n_e^{hot}$ and thus the
669: relative fraction of the dowscattering time lags in the total time lag
670: sum is $\sim$100\%.   
671:  However, Comptonization in a non-uniform accretion medium might
672:  account for the observed time lag as a non-monotonic function of energy. 
673: Using this model,  upper and lower limits of the
674: atmosphere density (density variation) in the region of phase/time lag
675: measurements were determined. Using the observed energy dependent
676: pulsed amplitude one can find the variation of the Thomson optical depth of
677: Compton cloud in which the accretion column is embedded.    
678: 
679:  
680: \section{Bursts}
681: 
682: MSP are also known to exhibit Type-I X-ray bursts. These bursts were
683: observed for \SAX\ and XTE~J1814-338 \citep{gmh06,gc06}. Several X-ray bursts
684: have also been observed  for \hete\ by various observatories
685: \citep{gmm06}. One of these bursts was also observed  during the
686: INTEGRAL ToO observation \citep{mf06}. In Figure \ref{fig:lc_burst} we 
687: show the JEM-X and ISGRI burst light curve (28 October 2005, 10:25:12
688: UTC) in different energy bands. The burst rise time was $0.23\pm0.05$ 
689: s. The double peak profile is clearly evident at 
690: high energy (lower panel) within the first 12 s, while  
691: during this time the intensity at lower energy (upper panel) remains
692: constant.  This can be interpreted as a consequence of a photospheric
693: radius  expansion (PRE) episode during the  first part of the outburst
694: \citep[see  e.g.,][]{vh95}. When a burst undergoes a PRE episode, 
695: the luminosity remains nearly constant at 
696: the Eddington value, the atmosphere expands, and its temperature  
697: decreases resulting in  a double-peak profile  observed at high
698: energies. The tail of the burst at high energy can be seen for about 5
699: s after the PRE episode.  
700:  
701: A time-resolved analysis of the net burst spectrum based on the 
702: JEM-X/ISGRI  3-50 keV energy band data was well fit by a 
703: photoelectrically absorbed blackbody. During the first 12 s, the unabsorbed 
704: bolometric flux was almost constant at $F_{\rm peak}=9.5(2)\times10^{-8}$ erg 
705: cm$^{-2}$ s$^{-1}$, while the blackbody temperature dropped in the middle, 
706: simultaneously with an increase by a factor of $\sim1.5$ in blackbody 
707: radius. The observed temperature reached a peak at $\sim$ 2.5 keV, and then 
708: gradually decreased. The softening of the emission towards the end of the 
709: decay phase is also indicated by the e-folding decay times 
710: of $12.5\pm0.5$ s in the 3--6 keV to $4.3\pm0.7$ s in the 12--20 keV
711: energy band.  This behavior is typically observed during PRE X-ray bursts.
712:  
713: The burst fluence was $f_{\rm b}=1.67(6)\times10^{-8}$erg cm$^{-2}$,
714: calculated by integrating the measured $F_{\rm bol,bb}$ over the burst
715: duration of $\sim 50$ s. The effective burst duration was $\tau=f_{\rm
716: b}/F_{\rm peak}=18.2(8)$ s, and the ratio of the observed persistent flux to
717: the net peak flux was $\gamma=F_{\rm pers}/F_{\rm  peak}=0.021(1)$.  
718: The burst has the same spectral parameters as previous 
719: bursts for this source observed with HETE-2 or RXTE \citep{gmh06}. 
720: Assuming a helium burst at the Eddington limit  and canonical 
721: NS parameters (1.4 solar mass and radius of 10 km),  the source
722: distance is estimated to be $\sim5$ kpc.  
723: 
724: From the observed {\it INTEGRAL} burst properties and the  mass
725: accretion rates  inferred  from  
726: the persistent luminosity, the present theory predicts that this
727: burst is pure helium burning. For helium 
728: flashes, the fuel burns rapidly, since there are no slow weak 
729: interactions, and the local Eddington limit is often exceeded. These 
730: conditions lead to PRE bursts with a duration, set mostly by the time 
731: it takes the heat to escape, of the order of 5--10 s, as 
732: observed. In the framework of the 
733: thermonuclear-flash models \citep[e.g.,][]{vh95} the burst 
734: duration, $\tau < 20$ s, and the ratio of 
735: the observed persistent flux to the net peak flux $\gamma\approx0.02$ indicate
736: a hydrogen-poor burst which is in agreement with the companion star being a 
737: helium-rich brown dwarf \citep{kaaret06}. 
738:  
739: 
740: Because there were no other bursts observed during the {\em INTEGRAL}
741: observation, the burst recurrence time, $\Delta t_{rec}$,  must be at
742: least one day. We can compute the ratio of the total energy 
743: emitted in the persistent flux to that emitted in  the burst 
744: $$
745: \alpha= \frac{F_{pers}}{f_b} \Delta t_{rec} = \frac{\gamma}{\tau}
746: \Delta t_{rec}  =   \frac{0.021}{18.2} \Delta t_{rec}  > 100, 
747: $$
748: which is consistent with  pure helium bursts \citep[see e.g.][]{vh95}. 
749: Taking  the burst total energy release $E_{b}=5\times 10^{39}$ erg   
750: (derived from the fluence $f_b$)  
751: and  He burning efficiency of $\epsilon_{\rm He}\approx 1.7$ MeV/nucleon $\approx 1.6\times10^{18}$
752: erg g$^{-1}$, we estimated the amount of fuel burned during the burst 
753: $E_{b}/\epsilon_{\rm He}\sim3.1\times10^{21}$ g. 
754: For the mean mass accretion rate of 2 per cent of the Eddington,
755: a burst recurrence time of 2.2 days is expected.
756: 
757: 
758: \begin{figure}[ht] 
759: \centering 
760: \includegraphics[width=1.3\linewidth,angle=-90]{f3.ps} 
761: %\vspace{4cm} 
762: \caption{A bright X-ray burst detected from HETE~J1900.1--2455. The JEM-X 
763:   (3--20 keV; upper panel) and IBIS/ISGRI (18--40 keV; lower panel) 
764:   net light curves are shown (background subtracted). The time bin is 
765:   0.5 s for both IBIS/ISGRI and JEM-X light curve. At high energy the 
766:   burst shows strong evidence of photospheric radius expansion.}
767: \label{fig:lc_burst}
768: \end{figure} 
769: 
770: 
771:   
772: \section{Conclusions} 
773: \label{sec:conclusions} 
774: 
775: 
776: 
777: The science of accreting millisecond pulsars has been carried forward
778: by the high-energy capabilities of INTEGRAL in concert with the 
779: timing resolution capabilities of RXTE and the  spectral
780: resolution of XMM at low energies. INTEGRAL observations have 
781: contributed to our ability to characterize the high energy spectrum of these
782: sources, and to study their interesting time-lag and pulse fraction
783: behavior over a wide energy range, up to $\sim$150 keV. 
784: 
785: The reason for the lack of coherent pulsations in the persistent emission
786: from LMXBs has been a longstanding open question in X-ray
787: astronomy. For the first time, we are able to observe a system (\hete) known
788: to contain a millisecond period pulsar, which has continued in
789: outburst after its pulsations have disappeared. 
790: The  transition of \hete\ from an X-ray
791: millisecond pulsar to a persistent LMXB could indicate that
792: there is a population of suppressed  X-ray millisecond pulsars among the
793: non-pulsating LMXBs. Detailed observations of this source at the epoch
794: of pulsation supression can help to solving the long-standing issue of
795: missing pulsations in persistent LMXB emission. 
796: 
797: 
798: \section*{Acknowledgments} 
799: 
800: I am grateful to the INTEGRAL MSP paper collaborators J. Poutanen,
801: E. W. Bonning, L. Kuiper, J. M. Bonnet-Bidaud, A. Goldwurm,
802: W. Hermsen, L. Stella, and L. Titarchuk on the time/phase lag paper.    
803: 
804: \begin{thebibliography}{} 
805: 
806: \bibitem[Barret et~al. (2000)]{b00}
807: Barret, D.,  Olive, J.~F.,  Boirin, L., Done, C., Skinner,
808: G.~K., \& Grindlay, J.~E.  2000, ApJ, 533, 329
809: 
810: \bibitem[Basko \& Sunyaev (1976)]{BS76} 
811: Basko, M. M.,  \& Sunyaev, R.A.  1976, MNRAS, 175, 395
812: 
813: \bibitem[Becker \& Wolff (2006)]{BW06} 
814: Becker, P. \& Wolff, M.  2006, ApJ, accepted, astro-ph/0609035 
815: 
816: \bibitem[Chakrabarty et~al. (20003]{Chakrabarty03}
817: Chakrabarty, D. et al., 2003, Nature, 424, 42
818: 
819: \bibitem[Cui et al. (1998)]{CMT98} 
820: Cui, W., Morgan, E., \& Titarchuk L.  1998, ApJ, 504, 27
821: 
822: \bibitem[Falanga et al. (2005a)]{mfa05}  
823: Falanga,  M., Bonnet-Bidaud, J. M., Poutanen, J., et al. 2005a, A\&A,
824: 436, 647   
825:   
826: \bibitem[Falanga et al. (2005b)]{mfb05}  
827: Falanga,  M., Kuiper, L., Poutanen, J., et al. 2005b, A\&A, 444, 15 
828: 
829: \bibitem[Falanga et al. (2006)]{mf06}  
830: Falanga, M., L., Poutanen, J., Bonning, E., W., et al. 2007, A\&A,
831: 464, 1069
832: 
833: \bibitem[Falanga \& Titarchuck (2007)]{MFLT06}  
834: Falanga, M., Titarchuck, L., 2007, ApJ, 661, 1084
835: 
836: \bibitem[Ford (2000)]{Ford00} Ford,  E., C., 2000, ApJ, 535, L119 
837: 
838: \bibitem[Galloway et al. (2005)]{gmm05} 
839:  Galloway, D. K.,  Markwardt, C. B., Morgan, E. H.,  Chakrabarty, D., \& 
840:   Strohmayer, T. E. 2005, ApJ, 622, L45 
841: 
842: \bibitem[Galloway et al. (2006)]{gmm06}  
843: Galloway, D. K., Morgan, E., H., Krauss H,. M., Kaaret P.,
844: Chakrabarty, D., 2006, ApJL,  [arXiv:astro-ph/0609693]
845: 
846: \bibitem[Galloway et al. (2006)]{gmh06}  
847: Galloway, D., K., Muno, M., P., Hartman, J., M., Savov, P., Psaltis,
848: D., Chakrabarty, D. , 2006, ApJS, [arXiv:astro-ph/0608259]
849: 
850: \bibitem[Galloway et al. (2006)]{gc06}  
851: Galloway D., K., Cumming, A., 2006, ApJL, in press [arXiv:astro-ph/0607213]
852: 
853: \bibitem[Gierli\'nski et al. (2002)]{gdb02}  
854: Gierli\'nski, M.,  Done, C., \& Barret, D. 2002, MNRAS, 331, 141  
855:   
856: \bibitem[Gierli\'nski \& Poutanen (2005)]{gp05}  
857:     Gierli\'nski, M., \& Poutanen, J.  2005,  MNRAS, 359, 1261  
858: 
859: \bibitem[Kaaret et al. (2006)]{kaaret06} 
860: Kaaret, P., Morgan, E. H., Vanderspek, R., \& Tomsick, J. A. 2006,
861: ApJ, 638, 963  
862: 
863: \bibitem[Mitsuda et al. (1984)]{mitsuda84}  
864: Mitsuda, K., Inoue, H.,  Koyama, K., et al. 1984, PASJ, 36, 741  
865:  
866: \bibitem[Poutanen \& Svensson (1996)]{ps96}  
867: Poutanen, J., \& Svensson, R. 1996, ApJ, 470, 249  
868: 
869: \bibitem[Poutanen \& Gierli\'nski (2003)]{pg03}  
870: Poutanen, J., \& Gierli\'nski, M. 2003, MNRAS, 343, 1301  
871:    
872: \bibitem[Poutanen (2006)]{p06} 
873: Poutanen, J. 2006, Adv. Space Res., in press [arXiv:astro-ph/0510038] 
874: 
875: \bibitem[Strohmayer (2001)]{stroh01}
876: Strohmayer, T. E., 2001, Adv. Space Res., 28, 511
877: 
878: \bibitem[Sunyaev \& Titarchuk (1980)]{ST80}
879: Sunyaev, R.A. \& Titarchuk, L.G.  1980,  A\&A, 86, 121  
880: 
881: \bibitem[Titarchuk et al. (2002)]{TCW}
882: Titarchuk, L.G., Cui, W., \& Wood, K.S., 2002, ApJ, 576, L49 
883: 
884: \bibitem[Verbunt \& van den Heuvel (1995)]{vh95}  
885: Verbunt, F., \& van  den Heuvel, E. P. J. 1995,  
886: in X-ray binaries, eds. W. H. G. Lewin, J. van Paradijs,  \& E. P. J. van  
887: den Heuvel (Cambridge: Cambridge University Press), p. 457
888: 
889: \bibitem[Viironen \& Poutanen (2004)]{vp04} 
890: Viironen, K., \& Poutanen, J. 2004, A\&A, 426, 985 
891:  
892: \bibitem[Wijnands \& van der Klis (1998)]{wvdk98}  
893: Wijnands, R., \& van der Klis, M. 1998, Nature,  394, 344 
894:  
895: \bibitem[Wijnands (2005)]{w05}  
896: Wijnands, R. 2005, in Pulsars New Research (NY: Nova Science  
897: Publishers), in press [astro-ph/0501264]  
898:    
899: 
900: \end{thebibliography} 
901: \end{document} 
902: %% 
903: %% <<<<< End of generated file <<<<<< 
904: %% 
905: %% End of file `esapub.tex'. 
906: