1: \documentclass[12pt,preprint]{aastex}
2: \usepackage[nonamelimits]{amsmath}
3: \usepackage{graphicx}
4: \begin{document}
5:
6: \title{Three Type I X-ray bursts from Cyg X-2:
7: Application of Analytical Models for Neutron Star Mass and Radius Determination}
8:
9: \author{ Lev Titarchuk\altaffilmark{1},
10: and Nickolai Shaposhnikov\altaffilmark{2}}
11: %\affil{Center for Earth Observing and Space Research, , George Mason
12: %University, Fairfax, VA 22030-4444}
13: %\email{nshaposh@scs.gmu.edu}
14: \altaffiltext{1}{George Mason University/Center for Earth
15: Observing and Space Research, Fairfax, VA 22030; and US Naval Research
16: Laboratory, Code 7620, Washington, DC 20375-5352; lev@xip.nrl.navy.mil }
17: \altaffiltext{2}{George Mason University, School for Computational
18: Sciences; Center for Earth Observing and Space Research, Fairfax, VA 22030;
19: nshaposh@scs.gmu.edu}
20: %\altaffiltext{3}{NASA/ Goddard Space Flight Center, Greenbelt
21: %MD 20771, USA; lev@lheapop.gsfc.nasa.gov}
22: %\author{Lev Titarchuk}
23:
24: %\author{Nickolai Shaposhnikov}
25: %\affil{George Mason University}
26:
27: \begin{abstract}
28: We study the spectral and temporal properties of three Type I X-ray bursts
29: observed from Cyg X-2 with {\it the Rossi X-ray Timing Explorer}.
30: Despite the short time durations ($\sim 5$ sec) these bursts
31: show radial expansion of order of several neutron star (NS) radii.
32: We apply the analytical models
33: of spectral formation during the expansion and contraction stages to derive physical conditions
34: for the matter in the burning zone close to the surface of the NS
35: as well as The NS's mass-radius relation. Our results, combined with statistical errors,
36: show that the central object is a compact star with mass $\sim 1.4$ solar masses and radius
37: $\sim 9$ km. Our results favor
38: the softer equation of state for neutron star matter.
39: \end{abstract}
40: \keywords{accretion---radiative transfer---stars:fundamental parameters---
41: stars:individual(Cyg X-2)---stars:neutron---X-ray:bursts}
42:
43: \section{Introduction}
44: The strong energy outbursts in the atmospheres of neutron stars in low
45: mass X-ray binary (LMXB) systems are commonly referred to as X-ray
46: bursts. Soon after their discovery by \citet{gh},
47: X-ray bursts were separated into two classes: in Type I and Type II bursts.
48: Type I bursts are currently considered to be due to
49: thermonuclear
50: explosions on a neutron star (NS) surface, and Type II bursts are believed to result
51: from the instability of the gravitational
52: energy release in the accretion flow (see Lewin et al. 1993).
53: Very strong Type I bursts (hereafter bursts, unless otherwise stated) can exhibit radial
54: atmospheric expansion.
55: %luminosity in the inner shells exceeds the Eddington limit for a NS star.
56: %and radiation pressure forces move the burst atmosphere outward (expansion stage).
57: %Analytical and numerical analysis along with observational evidence show
58: %that outward radiation flux during
59: %the expansion stage of X-ray bursts does not exceed the Eddington limit
60: % for a star by more than a few percent.
61: After the burst's peak, when the luminosity falls
62: below the Eddington limit,
63: the atmosphere quickly contracts back to the NS surface.
64: %The study of X-ray bursts can provide independent information
65: %for NS mass-radius relation.
66: %Estimates of NS masses up to $2 M_{\odot}$ \citep{ZSS}
67: %have been obtained within the framework of the beat-frequency model (see e.g.
68: %Miller, Lamb \& Psaltis 1998).
69: %We address the relevance of this result to the data analysis of the burst
70: %Type I in our Discussion section below.
71:
72: \citet{t94a} (hereafter T94) and \citet{burst} (hereafter ST02) developed the analytical theory of spectral
73: formation during the expansion and contraction stages of the burst.
74: % which included the effects of
75: %Comptonization and free-free absorption and emission.
76: %Newtonian gravity, the diffusion approximation and spherical symmetry were assumed.
77: %Recently, this work has been extended by \citet{burst} (hereafter ST02) by deriving simple
78: %approximate formulae for fitting the spectra
79: %observed during the expansion stage of the burst when the atmosphere is
80: %radiatively expanded.
81: %The analytical expressions provided
82: %for color factors, and the spectral shapes are presented as functions of
83: %the input parameters: element abundances, and the NS mass and radius.
84: %Additional parameters are the burning zone temperature and the Eddington
85: %ratio $l=L_{Edd}/L$ for the expansion and contraction stage, respectively.
86: Two cases have to be considered separately when the Eddington
87: ratio $l=L/L_{\rm Edd}\sim 1$ and when $l<0.9$
88: (i.e., before and after the ``touchdown'' of the extended atmosphere).
89: In the former case, the radiation pressure has substantial influence on the
90: NS atmosphere dynamics, while in the latter one can utilize
91: the hydrostatic equilibrium approximation. The T94 theory can be used to derive
92: distances to X-ray burst sources from which only burst luminosities
93: below the Eddington limit are observed. \citet{ht} (hereafter HT95) applied the T94
94: theory to 4U 1705-44 and they found a distance $7.4\pm 1$ kpc, assuming
95: similar NS mass-radius relation for 4U 1705-44 and 4U 1820-30.
96: HT95 presented constraints on the NS mass-radius relationship
97: in X-ray burst source 4U 1820-30.
98: %Their results excluded stiffer equations of state for neutron star matter.
99: We extend this methodology by
100: deriving the temperature of the burning zone and applying the theory of
101: spectral formation during the expansion stage to the observational spectra
102: obtained at the peaks of bursts. We show that when the mass-radius
103: contours obtained by applying this theory to observations are combined with the Eddington
104: limit constrains for peak flux, we obtain separate NS mass and
105: NS radius values.
106:
107: Most X-ray bursts are observed in LMXB sources whose color-color diagrams cause them
108: to be classified as atoll sources. However, the two Z-sources GX 17+2 and Cyg X-2
109: are exceptions to this rule. The observational data of X-ray
110: missions prior to RXTE suffered from poor statistical and time
111: resolution properties and thus could not resolve the nature of burst-like events detected from these
112: sources. The first incontestable evidence of a Type I burst from Cyg X-2
113: was reported by \citet{smale} (hereafter S98). The analysis provided by
114: S98 indicates that the burst color temperature increases
115: with the transition to the contraction stage and it
116: goes down during a burst decay.
117: %which clearly indicated thermonuclear nature of the burst.
118: In this {\it Letter} we concentrate our efforts on the study of
119: the spectral properties of
120: the bursts from the source Cyg X-2.
121: We have surveyed the RXTE public data archive and find more than
122: 20 burst events where the count rate increased by
123: a factor of three on a time scale of $1\sim 2$ seconds.
124: Temporal analysis of this set of bursts for which appropriate data configurations are
125: available shows
126: that for only three bursts (including the burst reported by S98) can
127: the photospheric radius expansion be well established.
128: As long as no other burst can be qualified as a Type I,
129: we summarize the results provided by the three bursts
130: with radial expansion. In \S2 we present the results of the analytical models
131: for X-ray spectral formation
132: in the burst atmospheres along with the methodology which we implement in our data
133: analysis.
134: \S 3 describes the observational data and reduction procedure.
135: \S 4 contains a report of our results, which is followed by discussion and
136: conclusions in \S 5.
137:
138: \section{Models for a spectral formation during expansion and
139: contraction stages}
140: % analytical results and their applications}
141: T94 presented analytical approximations for the solution of the radiative transfer problem
142: during the burst, covering a wide range of values for the Eddington ratio $l$.
143: %The solution involves integrating the photon diffusion equation
144: %with the Kompaneets energy operator \citep{komp} in spherical geometry
145: %with appropriate boundary conditions.
146: In a case of the hydrostatic equilibrium
147: atmosphere ($l<0.9$) the density profile is exponential.
148: The solution for the color factor $T_h$ as a function of the luminosity $l$,
149: the helium abundance $Y_{He}$, and the gravitational acceleration $g$ is
150: obtained in the form (T94)
151: \begin{equation}
152: T_h=T/T_{eff}=1.32 K(l,Y_{He},g)\Lambda_*^{-4/5},
153: \end{equation}
154: where
155: \[K(l,Y_{He},g)=l^{3/20}(1-5Y_{He}/8)^{2/15}(1-Y_{He}/2)^{-1/60}
156: (1-l)^{-2/15}g_{14}^{1/60},\]
157: %\[\Lambda_*= 0.14 \ln (8a) + 0.1 \exp \, (-5.5a) +0.6\,,\]
158: $$
159: \Lambda_*= 0.14 \ln (8a) + 0.1 \exp \, (-5.5a) +0.6\,,~
160: a= K^{15/2}(l,Y_{He},g) ~{\rm and} ~ g_{14}=g/10^{14}{\rm cm ~s}^{-2}.
161: $$
162: %\[a= K^{15/2}(l,Y_{He},g),\]
163: %where $g_{14}=g/10^{14}$cm s$^{-2}$.
164: %The situation is quite different when $l\sim1$.
165: %The layers of the atmosphere become dynamically unstable under the effects of
166: %radiative pressure and a wind-like expanding envelope is formed.
167: %T94 formulated the hydrodynamic problem in the form of the
168: %Cauchy's problem for the first-order ODE.
169: %Numerical solutions (see T94 and
170: %ST02) indicate that atmospheric optical depth profiles obey closely
171: %the relation $\tau \sim r^{-2}$.
172: The radiative transfer equation in the $l\sim1$ case also admits a
173: semi-analytical treatment, which is presented in detail in ST02.
174: Here we present the final results, obtained for the color factor $T_h$ and
175: color temperature $kT$
176: \begin{equation}
177: \label{eq2}
178: T_h= 0.37\, m^{0.174}\, r_6^{-0.28}\, T_{b,9}^{-0.64}\,(2-Y_{He})^{-0.074},
179: \end{equation}
180: \begin{equation}
181: \label{eq3}
182: kT= 0.76\, m^{0.42}\, r_6^{-0.78}\, T_{b,9}^{-0.64}\,(2-Y_{He})^{-0.324}\,
183: {\rm keV},
184: \end{equation}
185: where $r_6$ is the NS radius in units of $10^{6}$ cm, $m=M_{NS}/M_{\odot}$ is
186: the NS mass in
187: units of the solar mass, and $T_{b,9}$ is
188: the temperature at the bottom of the burst atmosphere in units of $10^9$ K.
189: With these expressions in hand (Eqs. 1-3), we
190: apply our formalism to PCA burst data.
191: The outcome of the fitting procedure implemented to X-ray burst data
192: are the color temperature of radiation $kT_\infty$ and
193: the bolometric flux $F_b$ with corresponding statistical errors.
194: Subscripts $\infty$ and $b$ denote the fact that these values are
195: detected by a distant observer. Renormalization to the real condition at
196: the photospheric surface should account for both dilution and
197: gravitational redshift. The dilution is described by
198: the relation $L_\infty= 4\pi d^2 \xi_b F_{b},$
199: %\begin{equation}
200: %\label{eq4}
201: %L_\infty= 4\pi d^2 \xi_b F_{b},
202: %\end{equation}
203: where $\xi_b$ is the anisotropy factor determined by the emission anisotropy of the burst.
204: The local luminosity $L$ and temperature $kT$ are connected with those
205: observed at infinity by
206: \begin{equation}
207: \label{eq4}
208: L_\infty= L/(z+1)^2,\,\,\,{\rm and}\,\,\,kT_\infty= kT/(z+1),
209: \end{equation}
210: %\begin{equation}
211: %\label{eq6}
212: %kT_\infty= kT/(z+1),
213: %\end{equation}
214: where the redshift factor is $(z+1)=(1-2G M_{NS}/R_{NS}c^2)^{-1/2}
215: =(1-0.297m/r_6)^{-1/2}$.
216: %
217: %\begin{equation}
218: %z+1=\left(1-\frac{2\,G M_{NS}}{R_{NS}c^2}\right)^{-1/2}=\left(1-0.297\, \frac{m}{r_6}\right)^{-1/2}.
219: %\end{equation}
220:
221: From the definition of the effective temperature and color factor
222: we can write
223: \begin{equation}
224: \label{eq5}
225: 4\pi R_{NS}^2\sigma T_{eff}^4 =
226: 4\pi R_{NS}^2\sigma \left(T/T_h\right)^4=L.
227: \end{equation}
228: After the atmospheric ``touchdown'' occurs (or when no substantial
229: photospheric expansion occurs) we rewrite $L$ as $l\,L_{Edd}$,
230: where
231: \begin{equation}
232: \label{eq6}
233: L_{\rm Edd}=4\pi c\, G M_{NS}(z+1)/\kappa_0(2-Y_{He})=
234: L_{\rm Edd,\infty}(z+1),
235: \end{equation}
236: with $\kappa_0 = 0.2$ cm$^2$g$^{-1}$.
237: Substituting this into (\ref{eq5}),
238: after introducing the general relativistic corrections in (4), we obtain
239: \begin{equation}
240: \label{eq7}
241: r_6^2=19.5(T_h/kT_\infty)^4
242: lm/(2-Y_{He})(z+1)^3.
243: \end{equation}
244: The dimensionless luminosity $l$ measured by a local observer
245: can be expressed in terms of observed fluxes and gravitational redshift as
246: \begin{equation}
247: \label{eq8}
248: l=L/L_{\rm Edd}=F_b(z+1)/F_{b,\rm Edd},
249: \end{equation}
250: where $F_{b,{\rm Edd}}\approx L_{{\rm Edd},\infty}/(4\pi d^2\xi_b)$ is
251: the observed burst peak bolometric flux
252: (we neglect the relativistic corrections in evaluating $F_{b,{\rm Edd}}$).
253: Using the above expression for $F_{b, \rm Edd}$ and Eq. (6) for
254: $L_{\rm Edd,\infty}$, we can rewrite this equation
255: as
256: \begin{equation}
257: \label{eq9}
258: l=0.476\,\xi_b\, d_{10}^{\,2}\, F_{b,8}\, (2-Y_{He})(z+1)/m,
259: \end{equation}
260: where $d_{10}=d/10$ kpc, and
261: $F_{b,8}=F_b/10^{-8}$erg cm$^{-2}$ s$^{-1}$.
262: Because $T_h$ and $z+1$ are dependent of $r_6$
263: we can find $r_6$ as a numerical solution of equation (7)
264: for a particular set of the parameters $m$ and $Y_{He}$.
265: %Once the burst subsides and the atmosphere settles back down to the NS surface
266: %(i.e. the photospheric radius levels off) we can obtain
267: %the NS radius as indicated by this particular burst dataset.
268: % As soon as the observational points correspond to events when
269: %the burst atmosphere subsides on the NS surface , we obtain the NS radius within a statistical error bar
270: %of the analyzed subset of the data.
271: Furthermore, we can determine the NS mass $m$ using the
272: Eddington limit (6), provided that the distance to the source is known.
273: Finally, we are able to test the conditions at the burning zone during
274: the expansion stage by virtue of relation \eqref{eq3}.
275: %For the expansion stage we have $l=1$ and neglecting gravitational redshift we obtain a simple expression
276: %\begin{equation}
277: %\label{eq11}
278: %r_6^2=19.5\left(\frac{T_h}{kT_\infty}\right)^4\frac{m}{(2-Y_{He})},
279: %\end{equation}
280: %which can be solved ariphmetcally due to the simple dependence \eqref{eq2} of color factor on $r_6$. Equations \eqref{eq10} and \eqref{eq11} comprise the
281: \section{Observations}
282: The Z-source Cyg X-2 has been observed extensively
283: by the proportional array counter (PCA) onboard RXTE throughout
284: the mission. We reviewed the entire Cyg X-2 archived data set
285: in an attempt to locate thermonuclear bursts.
286: %We first searched the Cyg X-2 archive for bursts.
287: When a burst was found we determined
288: what EDS (Experimental Data System) configuration was used for that observation.
289: We then choose those bursts for which the EDS configuration was well suited to our analysis.
290: Among the more than 20 burst-like events, we choose three for our analysis:
291: Burst 1 was detected on March 27, 1996 (ObsId:10066-01-01-00, S98),
292: burst 2 was detected on August 10, 1998(ObsId: 20046-01-05), and burst 3 was
293: detected on September 12, 1998 (obsID: 30046-01-01-00).
294: The first two bursts can be classified confidently as Type I X-ray bursts.
295: The nature of burst 3 is uncertain, as we discuss below.
296: During these observations, four additional EDS modes
297: were utilized along with two standard modes: binned mode
298: (0-35 PCA channels in 16 energy bins, 2 ms time resolution),
299: event mode (36-255 PCA channels in 64 bins, 125 $\mu s$ time resolution), and
300: two single-bit modes with 125 $\mu s$ time resolution, covering the 0-13 and
301: 14-35 PCA energy channels correspondingly.
302: %The RXTE mission observational cycle
303: %consists of four different gain epochs according to PCA voltage change.
304: Burst 1 was observed during PCA gain epoch 2, while burst 2 and 3 were observed during PCA
305: gain epoch 3 resulting in a
306: different instrumental response for across the PCA channels.
307: For our temporal burst analysis we are particularly interested
308: in binned and event modes.
309: For the data reduction we use the procedure adopted
310: by \citet{smale}, in that we extract a sequence of 0.5 second spectral slices
311: during each burst.
312: Then we extract the spectra of persistent emission prior to each burst.
313: The time duration for the persistent emission spectra is determined by the
314: requirement that the persistent flux does not undergo substantial systematic
315: changes during the interval.
316: %This procedure allows us to employ the persistent
317: %flux spectra as a background spectra for each particular burst.
318: All spectra produced are dead-time corrected according to RXTE Data
319: Reduction Recipes.
320:
321: \section{Data analysis and results}
322:
323: The best fits for persistent flux background spectra were
324: obtained using the XSPEC comptt+bbody model (Titarchuk 1994b).
325: Best fit parameters for bursts
326: 1, 2, and 3 respectively are: soft photon temperature $= 1.09\pm 0.06,\, 1.04\pm 0.02,\, 1.04\pm 0.05$ keV,
327: plasma temperature $ = 3.1\pm0.1,\, 4.2^{+1.2}_{-0.6}\, ,3.0\pm 0.2$ keV, plasma optical depth
328: $ = 4.7\pm 0.3,\, 2.16^{+0.45}_{-0.6},\, 3.9\pm 0.4$,
329: black body temperature $= 0.56\pm 0.03,\, 0.47\pm 0.03,\, 0.47\pm 0.05$ keV.
330: For each burst we fit each half-second spectral slices
331: over the energy range 2.8-30 keV.
332: using the black-body model which best describes the background subtracted burst
333: emission. To obtain the burst emission we subtract the persistent
334: component as given by analysis of preburst data.
335: The radiated energy from the burst can affect the persistent
336: flux both in magnitude and its spectral properties causing systematic
337: trends in our results. Specifically, if burst results in an increase of
338: the persistent component, our analytical approach will result in slight overestimation of
339: the NS radius.\footnote{ At this point we cannot refine the method of
340: persistent emission subtraction due to the quality of data
341: and the fact that no theoretical model for interaction of burst and
342: persistent energy flux has been developed.}
343: The best fit values of the color temperatures $kT$, the reduced $\chi_{red}^2$ values of the
344: blackbody spectral fits to the data,
345: accompanied by the luminosities $l$ and the inferred NS radii,
346: are plotted on Figure 1.
347: The typical behavior of a thermonuclear runaway process is most pronounced
348: in the first burst: after a quick rise, the color temperature
349: gradually hardens from 1 keV up to 2 keV and then decreases during decay phase.
350: The situation is less obvious for burst 3, where the temperature stays rather
351: high throughout the decay. We speculate that the persistent emission
352: is affected by the burst itself.
353: %This may be observed in the events
354: %associated with the higher color temperature.
355: In fact, the photons generated
356: by the bursts can be intercepted by the accretion disk and reflected.
357: This results in apparent anisotropy of the burst radiation
358: [see \citet{lst}, \citet{ps} for details].
359:
360: The second burst has a peak flux,
361: approximately 25\% higher than that of the other two bursts. This
362: discrepancy can be again explained by considering disk reflection and/or dynamical
363: evolution of the NS - accretion disk geometry
364: \footnote{The area of emitting NS surface exposed to observer can change
365: during bursts with radial expansion.
366: A strong indication that this may occur has recently been found by the authors
367: in X-ray bursts from 4U 1728-34. We will discuss this point in a forthcoming paper.
368: }.
369: The anisotropy effect can be accounted for in our analysis
370: by the inclusion of the anisotropy coefficient in
371: equation \eqref{eq9}.
372: Unfortunately, the limited signal$-$to$-$noise of the Cyg X-2 data during the
373: burst decay part does not allow us to make any further conclusions
374: on this issue.
375: To ensure consistency with the observed peak fluxes,
376: we set the anisotropy parameter constant to 0.8
377: for the second burst, while we keep it at unity for the first and
378: the third bursts.
379:
380: Combining equations (1), (7) and (9)
381: we obtain an equation relating the NS mass and radius to quantities
382: we can obtain from the observations:
383: \begin{equation}
384: \label{eq10}
385: r_6^2=9.28\xi_b d_{10}^2F_{b,8}(T_h/kT_\infty)^4
386: (1-0.297m/r_6).
387: %r_6^2=9.28\,\xi_b\, d_{10}^{\,2}\, F_{b,8}\,\left(\frac{T_h}{kT_\infty}\right)^4
388: %\left(1-0.297\frac{m}{r_6}\right).
389: \end{equation}
390: We solve equation \eqref{eq10} for each pair of values
391: $kT_\infty$, $F_{b,8}$ and a given set of parameters $Y_{He}$ and $m$
392: to obtain the radius of the emitting surface $r_6$ and Eddington ratio $l$.
393: It should be noted that a solution exists in the narrow parameter space
394: for $r_6$, $m$ and $Y_{He}$ only.
395: In Figure 1 the values of the inferred characteristics are shown along
396: with a plot of color temperatures, obtained using blackbody fits to the data.
397: After the luminosity peak the photospheric radius drops quickly,
398: while $l$ remains close to unity. After about 4 seconds the photosphere
399: reaches ``touchdown'', the radius levels off at constant value and
400: the Eddington ratio drops rapidly.
401: Because the thickness of the atmosphere (scale height $H\approx kT/mg\sim 10^2$ cm)
402: is negligible with respect to the neutron star radius $R_{NS}$,
403: we assume the photospheric radius, $R_{ph}$ after ``touchdown'',
404: is equal to the $R_{NS}$.
405: %In applying this criteria we examine the subset
406: %of the data points which, for each burst, that should provide the neutron star radius as
407: %a solution of Eq. \eqref{eq10}. We use this resulting dataset to obtain the mass-radius
408: %relationship for particular assumed distance and helium abundance.
409: %One can avoid the necessity of specifying the distance by using equation
410: %\eqref{eq11} for the calculation of $l$. The constraints on the distance
411: %to the object can be inferred from the Eddington limit.
412: Then for a particular distance
413: we determine the NS mass assuming the Eddington luminosity at the peak of the burst.
414: Figure 2 shows two contours for distances
415: of 9 and 11 kpc and cosmic elemental abundance along with the $R$,
416: $\pi$ and $\pi'$ equations of state for neutron star matter from Baym \& Pethick (1979).
417: The points with error bars on mass and radii present
418: NS mass-radius determination.
419: For an assumed distance of 11 kpc we obtain the canonical values for the
420: neutron star parameters: mass = $1.44\pm0.06$, radius = $9.0\pm0.5$ km,
421: while assuming 9 kpc the results are less realistic:
422: mass = $0.97\pm0.04$, radius = $7.7\pm0.4$ km
423: (quoted errors on the NS mass includes only the statistical error on
424: the burst peak flux).
425: %The convergence of our method is best for the
426: %$d=11$ kpc and $Y_{He}=0.3$ values.
427: The lack of convergence is the reason why the contour for $d=9$ kpc ends
428: at $m=1.4$.
429: If we demand that our model yields $m\approx 1.4$ and $Y_{He}\approx0.3$ then
430: our results favor the soft, specifically $R$ and $\pi$,
431: equations of states for neutron stars.
432: Using formula \eqref{eq3} we estimate the burning zone temperature
433: at the maximum expansion is within the range $(4-6)10^8$ K.
434:
435: \section{Discussion and Conclusions}
436: In light of the recently discovered millisecond variability in the X-ray flux
437: from a number of LMXBs, the task of independent determination of masses and
438: radii of neutron stars in these systems becomes very important.
439: Kilohertz Quasi-periodic oscillations (QPO) are seen in
440: the power-density spectrum of many LMXBs as twin peaks with frequencies
441: within the range of 500-1200 Hz. A number of models were put forward to
442: explain these high-frequency patterns.
443: First, and most obvious, was the suggestion of an interaction between
444: Keplerian frequency at neutron star surface (or last stable orbit) and
445: the frequency of the neutron star rotation
446: [beat-frequency model (BFM): \citet{as}; \citet{mlp}].
447: Despite some discrepancies between the model predictions and the observational
448: results, [the peak separation is not constant when the twin QPO frequencies vary
449: with time (see e.g. van der Klis 2000)],
450: the beat-frequency model became fashionable and
451: has been applied to infer information on the neutron star masses \cite{ZSS}.
452: Because of general relativistic (GR) effects, no stable particle motion is allowed
453: for a circular orbit
454: radius less than $R_{isco}=6GM/c^2= 8.9\, m$ km.
455: The corresponding frequency of the orbital motion
456: (assumed close to the QPO frequency) is $\nu_{isco}\approx 2210 / m$
457: Hz. The NS masses
458: in the LMXB exhibiting kHz QPO phenomena should be equal to or even exceed
459: two solar masses if the highest observed kHz QPO is interpreted as the frequency
460: of the innermost stable orbit. For Cyg X-2 the corresponding values are 1005 Hz
461: for the maximum QPO frequency \citep{wk} and consequently $2.2 M_{\odot}$ is
462: the NS inferred mass, clearly too high.
463: {\it Our analysis of the burst spectra, which takes into
464: account all corrections due to the GR and electron scattering effects,
465: are in disagreement with the mass-radius constraints obtained using
466: the kHz QPO frequencies values evaluated within the BFM frameworks.}
467: %\footnote{\citet{KT} recently presented a detailed analysis of
468: %the predictions of the relativistic precession model (RPM)
469: %versus the timing properties of the X-ray radiation detected by RXTE from Sco X-1.
470: %They find that the best-fit of the RPM to the data, leads to large value of the NS mass,
471: %($2.7\, M_\odot$) a value inconsistent with any existed EOS.}}.
472:
473: It is worth noting that in a recent study of optical and UV lightcurves of Cyg X-2
474: by Orosc \& Kuulkers (1999) who elaborate independent constraints on the NS mass
475: in Cyg X-2, they find $M_{NS}=1.78\pm0.23\,M_{\odot}$, consistent with our mass determination
476: above
477: \footnote{There remains the problem of the too small distance estimates of
478: $d_{10}=0.72\pm 0.11$
479: obtained by Orosz \& Kuulkers (1999). But this can be explained as the uncertainty
480: in the luminosity estimate for Cyg X-2 optical counterpart, caused by an unknown
481: disk contribution.
482: We should emphasize that the X-ray radiation detected from a type I burst
483: is directly related to the physical processes (thermonuclear explosions,
484: photon transport) which take place in the NS atmosphere. Consequently
485: it carries with it direct information regarding the main NS characteristics
486: (mass, radius) and the distance to the source.}.
487:
488: Sco X-1 and Cyg X-2 have many similar features in terms of their timing and
489: spectral properties and they have almost the same upper limit for the bolometric
490: flux (see Bradshaw, Fomalont \& Geldzahler, 1999).
491: Thus we expect that the similar mass-radius values
492: inferred for Cyg-2 (mass = $1.44\pm0.06$, radius = $9.0\pm0.5$ km)
493: should also be applied to Sco X-1.
494:
495: The authors are grateful to Menas Kafatos for support of this work and
496: to the referee for fruitful suggestions. We acknowledge the thorough analysis and
497: editing of the paper by Michael Wolff.
498:
499: \begin{thebibliography}{}
500: \bibitem[Alpar \& Shaham (1985)]{as} Alpar, M. A., \& Shaham, J.
501: 1985 , Nature, 316, 239
502: \bibitem[Baym \& Pethick (1979)]{bp} Baym, G., \& Pethick, C. 1979, Ann. Rev. Astr. Ap.
503: 17, 415
504: \bibitem[Bradshaw, Fomalont \& Geldzahler (1999)]
505: {bfg} Bradshaw, C. F., Fomalont, E. B., \& Geldzahler, B. J. 1999, \apj, 512,
506: L121
507: \bibitem[Grindlay \& Heise (1975)]{gh} Grindlay, J. E., \& Heise, J. 1975,
508: \iaucirc 2879
509: \bibitem[Haberl \& Titarchuk (1995)]{ht} Haberl, F., \& Titarchuk L. 1995,
510: \aap, 299, 414 (HT95)
511: %\bibitem[Kompaneets (1957)]{komp}
512: %Kompaneets, A. S. 1957, Sov. Phys. JEPT 4, 730
513: % and van den Bergh, S. 1984, \aj, 89, 1816
514: %\bibitem[Kuznetsov \& Titarchuk (2002)]{KT}
515: %Kuznetsov, S., \& Titarchuk, L. 2002, ApJ, in press
516: \bibitem[Lapidus, Sunyaev \& Titarchuk (1985)]{lst}
517: Lapidus, I. I., Sunyaev, R. A., \& Titarchuk, L. G. 1985,
518: Astrophysics, 23, 3, 663
519: \bibitem[Lewin et al.(1993)]{lpt} Lewin, W.H. G.,van Paradijs,J.,
520: \& Taam, R. E. 1993, Space Sci. Rev., 62, 223
521: \bibitem[Miller Lamb \& Psaltis (1998)]{mlp}
522: Miller, M. C., Lamb, F. K., \& Psaltis, D. 1998,
523: \apj, 508, 791
524: \bibitem[Orosz \& Kuulkers (1999)]{OK} Orosz, J. A., Kuulkers, E. 1999, \mnras, 305,
525: 132
526: \bibitem[Popham \& Sunyaev (2001)]{ps} Popham, R., \& Sunyaev, R. 2001,
527: \apj, 547, 355
528: \bibitem[Smale (1998)]{smale} Smale, A.P. 1998, \apj,
529: 498, L41
530: \bibitem[Shaposhnikov \& Titarchuk (2002)]{burst} Shaposhnikov, N., \&
531: Titarchuk, L. 2002, \apj, 567, 1077 (ST02)
532: \bibitem[Titarchuk (1994a)]{t94a} Titarchuk, L. 1994a, \apj, 429, 330 (T94)
533: \bibitem[Titarchuk (1994b)]{t94b} Titarchuk, L. 1994b, \apj, 434, 570
534: \bibitem[Zhang et al. (1997)]{ZSS} Zhang, W., Strohmayer, T. E., \& Swank, J. H.
535: 1997, \apj, 482, L167
536: % and van den Bergh, S. 1984, \aj, 89, 1816
537: \bibitem[van der Klis(2000)]{klis} van der Klis M. 2000, \araa, 38, 717
538: \bibitem[Wijnands \& van der Klis (1998)]{wk} Wijnands R., \&
539: van der Klis M. 1998, \iaucirc 6876
540: % 11`11 q
541: \end{thebibliography}
542: \clearpage
543:
544: %\begin{figure}
545: %\includegraphics[width=6in,height=6in,angle=-90]{f1.eps}
546: %\caption{Lightcurves of three bursts integrated over the PCA energy range (upper panel) and
547: %corresponding burst fluxes after subtraction of the persistent component
548: %(lower panel).}
549: %\end{figure}
550:
551:
552: \begin{figure}
553: \includegraphics[width=7in,height=4.5in,angle=-90]{f1.eps}
554: \caption{Time dependence of the inferred photospheric radius, dimensionless luminosity $l$,
555: the color temperature and $\chi_{red}^2$ values of the spectral fits
556: from our analysis for input parameters $Y_{He}=0.3$, distance = 10 kpc,
557: and a neutron star of 1.44 solar mass.
558: The observed flux for 1998 September 12 was renormalized in the analysis by a factor
559: of 0.8 (see text).}
560: \end{figure}
561:
562: \begin{figure}
563: \includegraphics[width=5in,height=6in,angle=-90]{f2.eps}
564: \caption{Regions of the NS mass-radius parameter space admitted by
565: the burst model for the cosmic helium abundance.
566: The upper region is assuming a distance of 11 kpc and the low
567: region has a distance 9 kpc The width of each region is a function of the
568: observational errors in flux $F_{b,8}$ and $kT_\infty$.
569: The two points with error bars
570: on mass and radius represent the mass-radius determinations inferred from
571: the average peak (Eddington) luminosity and the contraction stage spectra.
572: }
573: \end{figure}
574:
575: \end{document}
576: