1: %the Version 8, 7/28/01
2: %\documentclass{aastex}
3:
4: %% preprint produces a one-column, single-spaced document:
5:
6: \documentclass[12pt,preprint]{aastex}
7:
8: %% preprint2 produces a double-column, single-spaced document:
9:
10: % \documentclass[preprint2]{aastex}
11: % reduces spacing between lines
12: \tighten
13: %\eqsecnum
14:
15: %\received{ January 1999}
16: %\accepted{}
17: %\journalid{337}{15 January 1995}
18: %\articleid{11}{14}
19:
20: %\slugcomment{In press, ApJ )}
21: \def\lsim{\lower.5ex\hbox{$\; \buildrel < \over \sim \;$}}
22: \def\gsim{\lower.5ex\hbox{$\; \buildrel > \over \sim \;$}}
23: \def\lax {\ifmmode{_<\atop^{\sim}}\else{${_<\atop^{\sim}}$}\fi}
24: \def\gax {\ifmmode{_>\atop^{\sim}}\else{${_>\atop^{\sim}}$}\fi}
25: \def\etal{{\it et al.\/} }
26: \def\gtorder{\mathrel{\raise.3ex\hbox{$>$}\mkern-14mu
27: \lower0.6ex\hbox{$\sim$}}}
28: \def\ltorder{\mathrel{\raise.3ex\hbox{$<$}\mkern-14mu
29: \lower0.6ex\hbox{$\sim$}}}
30:
31:
32: \def\pmb#1{\setbox0=\hbox{#1}%
33: \kern-0.015em\copy0\kern-\wd0
34: \kern0.03em\copy0\kern-\wd0
35: \kern-0.015em\raise0.0433em\box0 }
36:
37: \begin{document}
38:
39: \title{B-field Determination from Magnetoacoustic Oscillations in kHz QPO Neutron Star
40: Binaries: Theory and Observations}
41:
42:
43: \author{L. G. Titarchuk}
44: \affil{George Mason University/CEOSR, Fairfax VA}
45: \email{lev@xip.nrl.navy.mil}
46:
47: \author{C. F. Bradshaw}
48: \affil{School of Computational Sciences, George Mason University,
49: Fairfax VA} \email{cbradshaw@tstag.com}
50:
51: \author{ K.S. Wood}
52: \affil{Naval Research Laboratory, Washington DC}
53: \email{wood@ssd0.nrl.navy.mil}
54: \vskip 0.5 truecm
55:
56: %c\font\rom=cmr10
57: %c\centerline{\rom Accepted, Astrophys. J. }
58:
59:
60: \begin{abstract}
61: We present a method for determining the B-field around neutron stars
62: based on observed kHz and viscous QPO frequencies used in combination with the best-fit
63: optical depth and temperature of a Comptonization model.
64: In the framework of the transition layer QPO model,
65: we analyze magnetoacoustic wave formation in the layer between a
66: neutron star surface and the inner edge of a Keplerian disk.
67: We derive formulas for the magnetoacoustic wave frequencies for
68: different regimes of radial transition layer oscillations. We
69: demonstrate that our model can use the QPO as a new kind of probe to determine
70: the magnetic field strengths for 4U 1728-42, GX 340+0, and Sco X-1 in the zone where the QPOs
71: occur. Observations indicate that the dependence of the viscous frequency
72: on the Keplerian frequency is closely related to the inferred
73: dependence of the magnetoacoustic wave frequency on the Keplerian
74: frequency for a dipole magnetic field. The magnetoacoustic wave dependence
75: is based on a single parameter, the magnetic moment of the star as estimated from the field
76: strength in the transition layer. The best-fit magnetic moment parameter is about $(0.5-1)\times10^{25}$
77: G cm$^3$ for all studied sources. From observational data, the
78: magnetic fields within distances less 20 km from neutron star for all three sources
79: are strongly constrained to be
80: dipole fields with the strengths $10^{7-8}$ G on the neutron star surface.
81: \end{abstract}
82:
83: %\keywords{X-rays: bursts-stars --- accretion disks, LMXB, X-ray
84: %binary --stars: neutron ---stars:
85: %individual: 4U 1728-34, GX 340+0, Sco X-1}
86:
87: \keywords{accretion disks --- stars: neutron --- stars: individual
88: (4U 1728-34, Scorpius X-1; GX 340+0)--- stars: magnetic fields--- X-rays: bursts}
89:
90: \section{Introduction}
91:
92: %The most reliable estimates of magnetic field strengths for neutron
93: %stars (NS) are from radio pulsars, with field strengths derived from spin down
94: %typically of order $\sim 10^{12}$ G
95: %(Bhattacharya, \& Srinivasan 1995). Millisecond
96: %pulsars are an exception to this clustering, with magnetic fields
97: %less than $10^9$G.
98: Strong magnetic fields of neutron stars, typically of order $\sim 10^{12}$ G (Bhattacharya, \& Srinivasan 1995)
99: are considered
100: crucial to the radio pulsar activity but equilibrium arguments concerning
101: NS magnetic field pressure and ram pressure balance lead to magnetic
102: field strength estimates
103: for heavily accreting NS sources (for example X-ray bursters) of $\ltorder10^9$ G.
104: In these sources no persistent pulsations have been found despite careful searches
105: (Wood et al. 1991; Vaughan et al. 1994).
106: Different scenarios of magnetic field evolution lead to different
107: estimates of magnetic field strength and decay. Thus, it is desirable
108: to find a direct way of measuring or estimating the neutron star's magnetic
109: field in these binary systems. Any model used to make such estimates should have minimum assumptions about
110: observed data and model relationships to produce accurate
111: results. {\it Our model, described herein, has only one parameter that can be adjusted
112: for agreement, the magnetic moment.
113: All other parameters are strongly constrained by observations.}
114: It develops a new probe of the magnetic field strength and geometry, based on radial
115: magnetoacoustic oscillations.
116:
117: Titarchuk, Lapidus \& Muslimov (1998), hereafter TLM98,
118: considered the possibility of dynamical adjustments of a Keplerian
119: disk to the innermost sub-Keplerian boundary conditions to explain
120: most observed QPOs in bright low mass X-ray binaries (LMXBs).
121: They concluded that an isothermal transition sub-Keplerian layer
122: between the NS surface and its last Keplerian orbit forms as a result
123: of this adjustment. The TLM98 model is general treatment applicable to both NS and
124: black hole (BH) systems. The primary problem in both NS and BH systems is
125: understanding how the flow changes from pure Keplerian to sub-Keplerian as a radius
126: decreases to small values. The TLM98 authors suggested
127: that the discontinuities and abrupt transitions in their solution result
128: from derivatives of quantities such as angular velocities (weak shocks).
129: Subsequently, Titarchuk \& Osherovich (1999 hereafter, TO99) identified
130: the transition layer (TL) model's corresponding TL
131: viscous frequency $\nu_V$ and diffusion or break frequency
132: $\nu_{b}$ for 4U 1728-34, and predicted values for
133: $\nu_b$ related to diffusion in the boundary layer.
134: %Then, Titarchuk, Osherovich \& Kuznetsov (1999 hereafter TOK99)
135: %confirmed, with observations of a number of QPO sources (Wijnands et
136: %al. 1999), a relationship, valid for both NS and BH candidate sources, of
137: %$\nu_b\propto \nu_V^{1.6}$.
138: TLM98 argued that a shock should occur where the Keplerian disk adjusts
139: to sub-Keplerian flow. They interpreted the low frequencies as acoustic oscillations
140: and expected them to provide reliable estimates of radial
141: velocities that would be very close to the acoustic velocities.
142:
143: Sco X-1's derived acoustic velocities (Titarchuk et al. 2001 hereafter
144: TBGF) translated to a TL plasma temperature, assuming only thermal motion,
145: of about 19 keV. However photon spectral fitting by TBGF yielded a TL temperature of 2.8 keV.
146: TBGF attributed this discrepancy to a dominant contribution to
147: the plasma matter energy density from the magnetic field.
148: %TBGF estimated
149: %the magnetic field strength to be $B < 0.7\times10^6$G in the TL
150: %of Sco X-1 at about 1-1.5 NS radii from the surface.
151: {\it The viscous QPOs provide information identifiable with a
152: magnetoacoustic (MA) oscillation effect that can be modeled}.
153:
154: In this {\it Letter}, we develop a general
155: formalism for MA radial oscillations in the
156: TL. In \S 2, we describe the data sources used to test our
157: MA oscillation model. The formulation of the problem and derivation of
158: MA frequencies are described in \S 3. We show the results of the
159: MA oscillation interpretation for the TO99 QPO viscous frequencies
160: and the best-fit normalization parameters of magnetic field strengths
161: for a number of QPO sources are present in \S 4.
162: Our summary and conclusions follow in \S 5.
163:
164: \section{Observations}
165: The model can only be applied to QPOs in which the full TLM has been worked out and in which
166: the QPOs identified with oscillations of a bounded medium have been seen.
167: All observations\footnote{This research has made use of data obtained
168: through the High Energy Astrophysics Science Archive Research Center
169: Online Service, provided by the NASA/Goddard Space Flight Center.}
170: referenced herein were made with the Rossi X-ray Timing Explorer's (RXTE).
171: Sco X-1 data were obtained during August 3 and 22, 1997, February 27
172: and 28, 1998, and June 10-13, 1999.
173: Data were extracted from research archives for GX 340+0 and 4U 1728-34 for
174: observations on November 14, 1998 and February 28, 1999, respectively.
175: In addition, we use the results from observations of 4U 1728-32 between
176: February 15-March 1, 1996 by Ford \& van der Klis (1998)
177:
178: \section{Magnetoacoustic Oscillations in the Transition Layer}
179:
180: In the references cited above, the treatment of a magnetic field in a fluid disk
181: has specified neither the multipole order that characterizes the field
182: nor the strength of the field.
183: %Yet, the
184: %invariance of $\delta$ (Osherovich \& Titarchuk 1999) in NS systems
185: %can be viewed as evidence that the central object is
186: %magnetized because it behaves like an object of finite
187: %radius with a symmetry-breaking feature.
188: %The feature, which is the
189: %magnetic field, channels plasma flow nearly
190: %perpendicular to the disk along an angle that departs from being
191: %perpendicular by a small invariant amount.
192: %Osherovich, Tzur \& Gliner (1984) first showed that the quadrupole term
193: %introduces north-south asymmetry in the makeup of the magnetic atmosphere
194: %surrounding a gravitating object. Recently, this asymmetry phenomenon has been fully
195: %confirmed for the solar corona (e.g., Gibson et al. 1999).
196:
197: We derive the frequency of the QPO associated with MA oscillations
198: and the correlation of the MA frequency with the Keplerian
199: frequency $\nu_{\rm K}$. The MA frequency is derived as the eigenfrequency
200: of the boundary-value problem resulting from a MHD treatment of the
201: interaction of the disk with the magnetic field.
202: The problem is solved for two limiting boundary conditions encompassing
203: realistic possibilities. The solution yields a velocity
204: identified as a mixture of the sound speed and the Alfv\'en velocity,
205: becoming either at the appropriate limits. Our treatment does not specify
206: how the eigenfrequency is excited or damped. However, it makes clear that the QPO
207: is a readily stimulated resonant frequency.
208:
209: In order to derive the MA oscillation frequencies in the
210: isothermal transition layer, we consider layer perturbations such that the
211: density $\rho=\rho_0+\rho_1$, the velocity ${\bf v=v_0+v_1}$ and ${\bf
212: B=B_0+B_1}$. By combining the continuity equation, equation of motion, and an ideal
213: gas equation, we obtain an equation for MA oscillations (Alfv\'en 1942)
214: \begin{equation}
215: {{\partial^2 {\bf v_1}}\over{\partial
216: t^2}}-s^2\nabla(\nabla\cdot{\bf v_1})
217: +{\bf v_{\rm A}}\times\nabla\times[\nabla\times({\bf
218: v_1\times v_{\rm A}})]=F({\bf v_0, \partial v_0/\partial r, B_0}, \rho_0)
219: \label{eq:alven}
220: \end{equation}
221: where $s^2=(\partial P/\partial \rho)$ is the square of the sound velocity
222: (assumed constant with respect to radius)
223: and ${\bf v_{\rm A}}={\bf B_0}(4\pi\rho_0)^{-1/2}$ is the
224: Alfv\'en velocity.
225: To determine the eigenfrequencies of
226: MA oscillations in the TL, we set equation
227: (\ref{eq:alven}) to zero and introduce a cylindrical coordinate system in
228: the TL with a radial axis in the disk plane, z-axis perpendicular to
229: the plane, and azimuthal axis along the circle in the disk plane
230: [see Fig. 1 in Titarchuk, Osherovich \& Kuznetsov (1999), hereafter TOK,
231: for the transition layer geometry].
232:
233: The magnetic field acts along a direction in space that is the same throughout the TL,
234: intersecting the disk everywhere at the same angle $\delta$.
235: Practical ranges for $\delta$ are
236: from a few to $\sim 15$ degrees (Titarchuk \& Osherovich 2001).
237: The model used for all three sources (see \S 4 )
238: assumes that $\delta=0$ exactly, which reinstates
239: axial symmetry and makes the problem mathematically tractable.
240: The simplified problem is spatially one dimensional in the radial direction,
241: and oscillations represented by the eigenfunctions are radial fluid movements.
242:
243: We consider the radial oscillations in the TL
244: for which we should assume that ${\bf B_0}=B_0{\bf e_z}$, i.e., the mean
245: magnetic field strength ${\rm B_0}$ is perpendicular to the disk and
246: ${\bf v_1}= v_1{\bf e_{r}}$ is along the radial direction in the disk.
247: We wish to explore how solutions for flow in the TL are affected by boundary conditions
248: and B field geometry, for acoustic and magnetic oscillations.
249: Under these assumptions, equation (\ref{eq:alven}) can be rewritten
250: as:
251: \begin{equation}
252: {{\partial^2 v_1}\over{\partial t^2}}-s^2{{\partial}\over{\partial
253: r}}\left({1\over r}{{\partial r v_1}\over{\partial r}}\right)- v_{\rm A}
254: {{\partial}\over{\partial r}}\left({1\over r}
255: {{\partial r v_{\rm A}v_1}\over{\partial r}}\right)=0.
256: \label{eq:new_alven}
257: \end{equation}
258: Combining this equation with the two boundary conditions of the
259: TL, the neutron star surface $r=R_{in}$
260: and the outer boundary, $r=R_{out}$, defined by the Keplerian QPO
261: frequency, allows us to analyze the eigenfrequencies of the MA
262: oscillations in the TL under both the stiff
263: $v_1(R_{in})=v_1(R_{out})=0$
264: and the free $\partial v_1/\partial r(R_{in})=\partial v_1/\partial r(R_{out})=0$
265: boundary conditions.
266:
267: In the two extreme cases, either $v_{\rm A}\ll s$ or $v_{\rm A}\gg s$
268: equation (\ref{eq:new_alven}) can be examined
269: analytically under the appropriate boundary conditions. Introducing a
270: new function $y=rv_1$ in the case of $v_{\rm A}\ll s$ and, $y=rv_{\rm
271: A}v_1$, in the case of $v_{\rm A}\gg s$, we can write equation
272: (\ref{eq:new_alven}) in the form
273: \begin{equation}
274: {{\partial^2 y}\over{\partial
275: t^2}}=4Ax^{(2-\alpha)/2}{{\partial^2y}\over{\partial x^2}}
276: \label{eq:partial_alven}
277: \end{equation}
278: where $x=r^2$ is a new variable, $\alpha =0$ and $A=A_s=s^2$ for the
279: pure acoustic case, and $\alpha \geq 6$ and
280: $A=A_{m}=B_{\ast}^2r_{\ast}^{\alpha}/4\pi\rho$
281: for the magnetic case. The index $\alpha$ is related to the multipole
282: magnetic field, $v_{\rm A}^2=A_{m}/r^{\alpha}$ ($\alpha=6, 8, 10$ are for
283: the dipole, quadrupole, and octopole, respectively).
284: $B_{\ast}$ is the magnetic field strength at a radius
285: $R_{in}<r_{\ast}<R_{out}$, and $\rho$ is the TL mean density.
286:
287: First, we need to find a solution of the eigenvalue problem for
288: equation (\ref{eq:partial_alven}) under the stiff boundary
289: conditions by obtaining a particular nontrivial solution using the
290: separation variables method, such that
291: \begin{equation}
292: y(x,t)=X_{\lambda}(x)T_{\lambda}(t)
293: \end{equation}
294: where $X_{\lambda} (x)$ is an eigenfunction satisfying
295: \begin{equation}
296: X^{\prime\prime}+{{x^{(\alpha-2)/2}}\over{4A}}\lambda^2 X=0
297: \end{equation}
298: and boundary conditions $X(x_{in})=X(x_{out})=0$.
299: The solution of these equations can be expressed through the Bessel
300: $Z$ functions
301: \begin{equation}
302: X_{\lambda}=x^{1/2}
303: Z_{2/(\alpha+2)}\left[{{2}\over{(\alpha+2)}}{{\lambda}\over{A^{1/2}}}
304: x^{(\alpha+2)/4}\right].
305: \end{equation}
306: $X_{\lambda}$ is a linear superposition of the Bessel
307: $J_{2/(\alpha+2)}$ and $J_{-2/(\alpha+2)}$ functions:
308: \begin{equation}
309: X_{\lambda}=x^{1/2}
310: [J_{2/(\alpha+2)}(z_{\lambda})~+~CJ_{-2/(\alpha+2)}(z_{\lambda})]
311: \label{eq:besselj}
312: \end{equation}
313: where
314: \begin{equation}
315: z_{\lambda}={{2}\over{(\alpha+2)}}{{\lambda}\over{A^{1/2}}}x^{(\alpha+2)/4}.
316: \label{eq:besselz}
317: \end{equation}
318: Wood et al. (2001) demonstrated that an asymptotic form
319: of the Bessel function for $z_{\lambda}\gax 1$ can be used to find
320: all eigenvalues including the first one:
321: \begin{equation}
322: J_{2/(\alpha+2)}(z_{\lambda})=\left({2\over{\pi
323: z_{\lambda}}}\right)^{1/2}
324: \cos[z_{\lambda}-(\alpha+6)\pi/4(\alpha+2)]
325: \end{equation}
326: and
327: \begin{equation}
328: J_{-2/(\alpha+2)}(z_{\lambda})=\left({2\over{\pi
329: z_{\lambda}}}\right)^{1/2}
330: \cos[z_{\lambda}-(\alpha-2)\pi/4(\alpha+2)].
331: \end{equation}
332:
333: Using these asymptotic forms, the boundary conditions, and equations
334: (\ref{eq:besselj} \& \ref{eq:besselz}),
335: the eigenvalues are determined by
336: \begin{equation}
337: \cos(z_{out}-\varphi_1)\cos(z_{in}-\varphi_2)-\cos(z_{out}-\varphi_2)\cos(z_
338: {in}-\varphi_1)=0,
339: \label{eq:eigen}
340: \end{equation}
341: where $\varphi_1=(\alpha-2)\pi/4(\alpha+2)$ and
342: $\varphi_2=(\alpha+6)\pi/4(\alpha+2)$.
343: Equation (\ref{eq:eigen}) can be reduced to
344: $\sin(z_{out}-z_{in})=0$ or $\beta=z_{out}-z_{in}=n\pi;~~n=1,2,3...$,
345: with a solution
346: \begin{equation}
347: \lambda_n={{\alpha+2}\over{2}}{{n\pi
348: A^{1/2}}\over{r_{out}^{(\alpha+2)/2} -r_{in}^{(\alpha+2)/2}}}.
349: \end{equation}
350: The time dependence, $T(t)$ is found from the equation
351: $T^{\prime\prime}+\lambda_n^2T=0$.
352: Thus, $T(t)=\sin(\lambda_n t+\phi_0)=\sin(2\pi\nu_n t+\phi_0)$ where
353: $\phi_0$ is an initial phase. For the acoustic case ($A=A_s=s$ and $\alpha=0$),
354: the main eigenvalue
355: $\lambda_1=\pi /(r_{out}-r_{in})= \pi/L$ and
356: $\nu_s=\nu_1=\lambda_1/2\pi=s/2L$,
357: where $L$ is the TL radial size.
358:
359: For the magnetic case ($\alpha\geq6$)
360: \begin{equation}
361: \nu_{M}=\nu_1=
362: {{(\alpha+2)B_{\ast}}\over{4(4\pi\rho)^{1/2}[r_{out}-r_{in}(r_{in}/r_{out})^
363: {\alpha/2}]}}\left({{r_{\ast}}\over{r_{out}}}\right)^{\alpha/2}={{(\alpha+2)
364: v_{\rm A}(r_{out})}\over{4[r_{out}-r_{in}(r_{in}/r_{out})^{\alpha/2}]}}.
365: \end{equation}
366: For the free boundaries, the boundary conditions for $y$ are
367: \begin{equation}
368: {{\partial y}\over{\partial x}} +{{(\alpha-2)}\over{4}}{y\over
369: x}=0~~~~{\rm at}~~ x=x_{in}~~ {\rm and}~~x=x_{out},
370: \end{equation}
371: from which, the transcendental equation for determining the eigenvalue is
372: \begin{equation}
373: \tan\beta=-{{2(\alpha-2)\beta}\over{(\alpha+2)\{\eta\beta^2/
374: (\eta-1)^2+[(\alpha-2)/(\alpha+2)]^2\}}},
375: \label{eq:eigen_tan}
376: \end{equation}
377: where
378: $\beta=z_{out}-z_{in}$ and
379: $\eta=z_{out}/z_{in}=(r_{out}/r_{in})^{(\alpha+2)/2}$.
380:
381: Analysis of equation (\ref{eq:eigen_tan}) shows that for
382: the values most relevant to observations ($\eta\gax 2$),
383: \begin{equation}
384: \beta_{M}\approx
385: {{\pi}\over2}+{2\over{\pi}}[(\pi/2)^2\eta/(\eta-1)^2+1/4]
386: \end{equation}
387: for $\alpha=6$ and $\beta_{s}\approx \{1.5/[1+1.5\eta/(\eta-1)^2]\}^{1/2}$
388: for $\alpha=0$.
389:
390: The free-boundary conditions will therefore introduce additional factors appearing in
391: formulas for $\nu_1$ between 0.5 and 1 for the magnetic case and
392: $1/\pi$ for the acoustic case.
393: From this, we conclude that the free-boundary conditions can decrease substantially
394: the magnetic oscillation frequency and must
395: decrease the acoustic frequency by at least a factor of 3 (see Eqs. 17-18 below).
396:
397: \section{Magnetic Field Strength Determination and the Magnetoacoustic
398: Oscillation Frequency}
399:
400: We can construct an approximate formula for the MA frequency
401: $\nu_{MA}$ using the asymptotic forms for the two extreme (acoustic
402: and magnetic) cases
403: \begin{equation}
404: \nu_{MA}\approx (\nu_{s}^2+\nu_M^2)^{1/2}=
405: \{s^2/4(r_{out}-r_{in})^2+[(\alpha+2)/4]^2v_{\rm
406: A}^2(r_{out})/[r_{out}-r_{in}(r_{in}/r_{out})^{\alpha/2}]^2\}^{1/2}.
407: \label{eq:ma_freq}
408: \end{equation}
409: The approximate relation in equation (\ref{eq:ma_freq}) is exact if
410: the Alf\'ven velocity is assumed to be constant
411: through the TL. The difference, by a factor
412: $(\alpha+2)^2/4$, in
413: coefficients for $s^2$ and $v_{\rm A}^2$ is due to different behaviors
414: of $s$ and $v_{\rm A}$ as a function of $r$ ($s=const$ and $v_{\rm
415: A}\propto r^{-\alpha/2}$).
416:
417: Under the free boundary, equation (\ref{eq:ma_freq}) is modified to
418: \begin{equation}
419: \nu_{MA}\approx \{(\beta_s/\pi)^2s^2/4(r_{out}-r_{in})^2+
420: (\beta_{M}/\pi)^2[(\alpha+2)/4]^2v_{\rm
421: A}^2(r_{out})/[r_{out}-r_{in}(r_{in}/r_{out})^{\alpha/2}]^2\}^{1/2}
422: \label{eq:ma_free}
423: \end{equation}
424: and provides the dependence of $\nu_{MA}$ on $\nu_{\rm K}$. This
425: dependence is due to the explicit dependence of $r_{out}$ on the
426: Keplerian frequency
427: $\nu_{\rm K}$:
428: \begin{equation}
429: r_{out}=(GM/2\pi\nu_{\rm K}^2)^{1/3},
430: \label{eq:rout}
431: \end{equation}
432: where $G$ is the gravitational constant and $M$ is the NS mass.
433: To determine $s$, we define $s=(kT/m_p)^{1/2}$ as the sound speed for the
434: isothermal TL where $kT$ is the best-fit value of the Comptonization temperature.
435: To determine $v_{M}=B(r_{out})/(4\pi \rho_0)^{1/2}$, we express
436: $\rho=\tau m_p/(\sigma_T L)$ with $\tau$ as the best-fit optical depth
437: parameter. These are not adjustable parameters for
438: determining magnetic field strengths.
439:
440:
441: {\it Thus, we obtain $\nu_{MA}$ as a function of only one adjustable parameter,
442: the magnetic-field strength at a TL radius
443: $r_{\ast}$}. We can replace this parameter by the multipole moment
444: $B(r_{\ast})r_{\ast}^{\alpha/2}$. In the dipole case $\alpha=6$, the multipole moment is the
445: dipole moment $\mu$ (in G cm$^{3}$).
446: The procedure for determining the magnetic field strength in the
447: TL may realized if there is an observable dependence
448: between the viscous and Keplerian frequencies.
449:
450: The spectra of 4U 1728-34, GX 340+0, and Sco X-1 were fitted by
451: the Comptonization model of Sunyaev \& Titarchuk (1980) from
452: 8-25 keV, to obtain a plasma temperature
453: (kT) and Thompson optical depth ($\tau$).
454: Figure 1 shows the results from the model and the fitted data
455: for 4U 1728-34, GX 340+0, and Sco X-1 (Ford \& van der Klis 1998).
456:
457: We found that a dipole geometry $(\alpha=6)$ gives a good fit and higher
458: poles are excluded by observed data.
459: It is also evident from equations (\ref{eq:ma_freq}-\ref{eq:rout}) that
460: the MA frequencies are
461: well described by a power law dependence on $\nu_{\rm K}$ when
462: $\nu_M>\nu_s$, i.e.,
463: $\nu_{MA}\approx \nu_{M}\propto \nu_{\rm K}^{(\alpha+2)/3}$. The power law index in the
464: dipole case
465: is $(\alpha+2)/3=8/3$.
466: The MA frequency dependence on $\nu_{\rm K}$ is too steep
467: for the quadrupole and the octopole cases [where $(\alpha+2)/3=10/3$ and
468: 4, respectively] to be supported by the data.
469: The fits of our MA model to the data provide strong constraints
470: for the boundary condition types where the stiff conditions are ruled out
471: for 1728-34 and the free conditions for GX 340+0. The fit qualities are practically
472: insensitive for $M =1.2\pm 0.2 M_{\odot}$. The best-fit mass parameter is around 1.2.
473:
474: The best-fit parameters for 4U 1728-34 (Compton values $kT=8.6 \pm 2$ keV,
475: $\tau = 5.8 \pm 0.5$; $\chi^2/dof = 19.8/35$)
476: %are obtained
477: %as the best-fit parameters of the Comptonization model to
478: %the X-ray spectrum )
479: are $B_{\ast}=(0.86\pm0.03)\times10^6$ G,
480: %$m=M/M_{\odot}=1.1\pm 0.2$
481: and $r_{\ast}=1.69\times10^6$ cm,
482: corresponding to a magnetic moment $\mu=4.2\times10^{24}$ G cm$^3$.
483: We fixed the NS radius, $r_{in}$ to be (as a canonical value) three Schwarszchild radii $3~r_{\rm S}$. An
484: extrapolation of the magnetic field from $r_{\ast}$ towards the NS radius
485: $r_{in}$ gives us $B_{NS}=4.5\times10^6$ G and $B_{NS}=1.3\times10^7$ G
486: for the dipole and octopole fields, respectively.
487:
488: The best-fit parameters for GX 340+0 (Compton values $kT=2.87 \pm 0.02$
489: keV, $\tau = 11.2 \pm 0.2$;
490: $\chi^2/dof = 49.8/35$) are $B_{\ast}=(0.5\pm0.05)\times10^6$ G
491: and
492: %$m=1.2\pm 0.2$ and
493: $r_{\ast}=2.7\times10^6$ cm,
494: corresponding to a magnetic moment $\mu=1\times10^{25}$ G cm$^3$. An
495: extrapolation of the magnetic field
496: from $r_{\ast}$ towards the NS radius $r_{in}$ yields
497: $B_{NS}=8.3\times10^6$ G and $B_{NS}=5.3\times10^7$ G for the dipole
498: and octopole fields, respectively.
499:
500: %Our free-boundary condition provided the best solution to meet
501: %the data constraints for Sco X-1.
502: The best-fit parameters for Sco X-1
503: (Compton values $kT=2.87 \pm 0.03$ keV, $\tau = 11.3 \pm 0.3$; $\chi^2/dof
504: =9.9/33$) are $B_{\ast}=(1.\pm0.05)\times10^6$ G,
505: %$m=1.2\pm 0.2$
506: and
507: $r_{\ast}=2.12\times10^6$ cm,
508: corresponding to a magnetic moment $\mu=9.5\times10^{24}$ G cm$^3$ in
509: the TL. An extrapolation of the magnetic field
510: from $r_{\ast}$ towards the NS radius $r_{in}$ gives us
511: $B_{NS}=8.1\times10^6$ G and $B_{NS}=3.3\times10^7$ G for the dipole
512: and octopole fields, respectively.
513:
514: These derived magnetic-field strengths are
515: for the TL only and can be smaller than the true
516: magnetic field outside of the layer because of its finite conductivity.
517: The NS magnetic-field strengths can be increased by factors of 1.7 and 2.5
518: for the dipole and octopole extrapolation respectively if we assume that the NS radius
519: is around $2.5~r_{\rm S}$ instead of $3~r_{\rm S}$.
520: % In fact, Haberl \& Titarchuk (1995) have
521: %already shown such a NS mass-radius relation for 4U 1820-30 and recently they obtained a similar
522: %relation for 4U 1728-34 (Titarchuk \& Haberl 2001).
523:
524: %The model, in its simplest form, provides a surprisingly good fit to the variation
525: %seen in all three sources without difficulty and without further parametrization beyond
526: %fitting the magnetic dipole moment. Values obtained in all three sources are
527: %reasonable.
528: The magnetic field strength can be roughly checked using an
529: equipartition estimate for which $B^2/8\pi=nkT$.
530: For example, Sco X-1's TL density $\rho=2.7\times10^{-5}$ g cm$^{-3}$
531: is obtained using the best-fit optical depth $\tau=11.3$.
532: The TL size, $L=10^6$ cm, is obtained using the observed
533: Keplerian frequency and best-fit temperature $kT= 2.9$ keV.
534: Thus, we obtain a magnetic field, $B=1.3\times10^6$ G, at radius
535: $r=2.3\times10^6$ cm.
536:
537: \section{Summary and Conclusions}
538: Since the discovery of neuron stars in the 1960s magnetic fields have been estimated almost
539: exclusively from {\it rotational} effects, i.e. from periodic pulses with their
540: periods $p$ and derivatives $\dot p$ [e.g. Bhattacharya \& Srinivasan (1995)]. The new idea here is that, when the field configuration is nearly
541: axis-symmetric, there may arise circumstances in which the field is estimated from
542: radial oscillations (QPOs) of plasma interacting with magnetosphere.
543:
544: %We have presented a model that goes beyond TO99 to treat the radial oscillations in the
545: %transition layer surrounding a neutron star.
546: The model preconditions are:
547: (1) the MA QPO must be detected observationally, (2) a Keplerian QPO
548: frequency must be observed and (3) the temperature and the optical
549: depth of the QPO emission region must be obtained from the photon spectra.
550: From our model we derived tightly constrained dipole magnetic
551: fields of the TL for the bright LMXBs 4U 1728-34, GX 340+0,
552: and Sco X-1. The observed low frequency correlation of the kHz QPOs
553: strongly confine the theoretical dependence of
554: $\nu_{MA}$ on $\nu_{\rm K}$, which is mostly determined by the
555: magnetic frequency dependence on $\nu_{\rm K}$
556: (see dashed lines in Figure 1) and very sensitive
557: to the pole order (dipole, quadrupole, {\it etc.}).
558: The correction to acoustic oscillations [the first term in formulas (17-18)]
559: only slightly changes
560: the theoretical shape of the low frequencies of $\nu_{\rm K}$.
561:
562: %Our analysis further confirms that the low-noise QPO frequency
563: %is a radial-oscillation frequency of the TL as a whole and
564: %demonstrates that it is dictated by MA oscillations. In addition, the
565: %absolute normalizations of these frequencies are determined only by the TL
566: %magnetic field strength, temperature, density, and neutron star mass.
567:
568: We re-fit the stellar magnetic dipole moment in each source to derive a range of
569: values from $(0.5-1)\times 10^{25}$ G cm$^{3}$.
570: %This range is plausible
571: %in all sources. We confirm that three LMXBs, characterized presumably
572: %(at least 4U 1728-32) by spin rates of order 300-400 Hz, have derived
573: %dipole field strengths on the order of $10^{6-7}$ G in the
574: %TL and on the order of $10^{7-8}$ G on the NS surface $10^{7-8}$ G.
575: %These values support an evolutionary tie
576: %between LMXB sources and millisecond radio pulsars.
577: The field estimate arrived at this manner is close to that may be evolutionary end points
578: of LMXBs, namely milisecond pulsars, which have derived
579: values (from $\sqrt{p\dot p}$) of $\sim 10^8$ G (Bhattacharya 1995).
580:
581: We wish to thank Sergey Kuznetsov, Eric Ford, Peter Jonker and Michael
582: van der Klis for the Sco X-1, 4U 1728-34 and GX 340+0 data files.
583: We are truly grateful to the referee, J. Imamura for his thorough analysis of
584: the presented manuscript.
585:
586: \begin{thebibliography}{}
587:
588: \bibitem[Alfven (1942)]{alf}
589: Alfv\'en, H. 1942, Nature, 150, 405
590:
591: \bibitem[Bhattacharya(1995)]{bs95}
592: Bhattacharya, D. 1995, in {
593: X-ray Binaries}, eds. W. H. G. Lewin, J. van Paradijs, \& E. P. J. van
594: den Heuvel, (Cambridge: U. Press), 233
595:
596: \bibitem[Bhattacharya \& Srinivasan (1995)]{bs95}
597: Bhattacharya, D., \& Srinivasan, G. 1995, in {
598: X-ray Binaries}, eds. W. H. G. Lewin, J. van Paradijs, \& E. P. J. van
599: den Heuvel, (Cambridge: U. Press), 495
600:
601: \bibitem[Ford \& van der Klis (1998)]{fv}
602: Ford, E., \& van der Klis, M. 1998, ApJ, 506, L39
603:
604: %\bibitem[Gibson et al. (1999)]{gib}
605: %Gibson, S.E., et al. 1999, ApJ, 520, 871
606:
607: %\bibitem[Haberl, \& Titarchuk (1995)]{HT95}
608: %Haberl, F. \& Titarchuk, L. 1995, A\&A, 299, 414
609:
610: \bibitem[Lang (1998)]{lang98} Lang, K. R., 1998, {Astrophysical
611: Formulae}, (Berlin: Springer)
612:
613: %\bibitem[Popham \& Sunyaev (2000)]{ps00} Popham, R. \& Sunyaev, R.,
614: %2000, ApJ, 457, 355
615:
616: %\bibitem[Psaltis et al. (2000)]{ps}
617: %Psaltis, D., Belloni, T. \& van der Klis, M. 1999, ApJ, 520,262
618:
619: %\bibitem[Soria (1999)]{sor}
620: %Soria, R. 1999, Thesis, Australian National University: Canberra
621:
622: %\bibitem[Soria (1997)]{sor97}
623: %Soria, R., Li, J., \& Wickramasinghe, D. T. 1997, ApJ, 487, 769
624: %\bibitem[Osherovich, \& Titarchuk (1999)]{ot99}
625: %Osherovich, V., \& Titarchuk, L. 1999, ApJ, 522, L113
626:
627: \bibitem[Sunyaev \& Titarchuk (1980)]{sun80} Sunyaev, R. \&
628: Titarchuk, L., 1980, A \& A, 86, 121
629:
630: %\bibitem[Titarchuk \& Haberl (2001)]{TH01}
631: %Titarchuk, L. \& Haberl, F. 2001, in preparation
632:
633: \bibitem[Titarchuk et al. (2001)]{t01}
634: Titarchuk, L., Bradshaw, C. F., Geldzahler, B. J.,
635: \& Fomalont, E. B. 2001, ApJ, 555, L45 (TBGF)
636:
637: \bibitem[Titarchuk, \& Osherovich (2001)]{to01}
638: Titarchuk, L., \& Osherovich, V. 2001, ApJ, 555, L55
639:
640: %\bibitem[Titarchuk, \& Osherovich (2000)]{to00}
641: %Titarchuk, L. \& Osherovich, V. 2000, ApJ, 537, L39
642:
643: \bibitem[Titarchuk, \& Osherovich (1999)]{to99}
644: Titarchuk, L. \& Osherovich, V. 1999, ApJ, 518, L95 (TO99)
645:
646: \bibitem[Titarchuk \etal (1999)]{TOK99} Titarchuk, L., Osherovich, V., \&
647: Kuznetsov, S., 1999, ApJ, 525, L129
648:
649: \bibitem[Titarchuk, Lapidus \& Muslimov (1998)]{TLM98}
650: Titarchuk, L., Lapidus, I.,\& Muslimov, A. 1998, \apj, 499, 315
651: (TLM98)
652:
653: %\bibitem[van der Klis, (2000)]{van}
654: %van der Klis, M. 2000, ARAA, 38, 717
655:
656: %\bibitem[Wijnands \& van der Klis (1999)]{wv}
657: %Wijnands, A.D. \& van der Klis, M. 1999, ApJ, 514, 939
658: \bibitem[Vaughan et al. (1994)]{vau}
659: Vaughan, B. et al. 1994, ApJ, 435, 362
660:
661: \bibitem[Wood et al. (2001)]{woo}
662: Wood, K.S., Titarchuk, L.G., Ray, P.S., Wolff, M.T., Lovellette, M.N.,
663: Bandyopadhyay, R.M. 2001, ApJ, in press
664:
665: \bibitem[Wood et al. (1991)]{woo}
666: Wood, K.S., et al. 1991, ApJ, 379, 295
667:
668:
669: \end{thebibliography}
670:
671:
672: \clearpage
673:
674: \clearpage
675: \begin{figure}
676: \epsscale{0.7}
677: \plotone{f1.eps}
678: \caption{MA frequency vs Keplerian frequency for 4U 1728-34
679: (black squares represent QPO frequencies from Ford \& van der Klis 1998,
680: blue, from TOK and magenta, from Jonker et al. 2000). 1-$\sigma$ errors bars indicate
681: statistical fitting errors
682: ($\Delta\chi^2=1$). The value for Q ($=\nu$/FWHM) approaches 1.0 for
683: viscous QPOs corresponding to Keplerian frequencies of 320-360 Hz (Ford \& van der Klis
684: 1998), giving formal frequency errors significantly larger
685: than the plotted error bars. The solid lines are the theoretical best-fits for the MA
686: frequencies. The dashed lines are for the magnetic frequencies only (see text).}
687: \label{fig1}
688: \end{figure}
689: \clearpage
690:
691: \end{document}
692: