1: %\documentstyle[12pt,psfig,preprint]{aastex}
2:
3: %\documentstyle[emulateapj,psfig]{article}
4: \documentclass[11pt,preprint]{aastex}
5: \usepackage{psfig}
6:
7: \begin{document}
8:
9: \title{CONSTRAINING BRANS-DICKE GRAVITY WITH ACCRETING MILLISECOND PULSARS
10: IN ULTRACOMPACT BINARIES}
11:
12: \author{Dimitrios Psaltis\altaffilmark{1}}
13:
14: \affil{Physics and Astronomy Departments, University of
15: Arizona, 1118 E. 4th St., Tucson, AZ 85721\\
16: dpsaltis@physics.arizona.edu}
17:
18: \altaffiltext{1}{also, Sabanci University, Orhanli-Tuzla, 34956 Istanbul,
19: Turkey}
20:
21: \begin{abstract}
22: The five accreting, millisecond X-ray pulsars in ultracompact binaries
23: that were recently discovered with the {\em Rossi X-ray Timing
24: Explorer\/} provide excellent candidates for constraining the
25: deviations from general relativity described by the Brans-Dicke
26: parameter $\omega_{\rm BD}$. I calculate the expected rate of change
27: of the orbital periods of these binaries and discuss the prospects of
28: constraining $\omega_{\rm BD}$ to values that are an order of
29: magnitude larger than current constraints. Finally, I show how
30: measurements of the orbital period derivative in ultracompact binaries
31: can be used to place lower bounds on their orbital inclination.
32: \end{abstract}
33:
34: \keywords{gravitation--X-rays: binaries}
35:
36: \section{INTRODUCTION}
37:
38: Neutron stars in compact binaries provide some of the best physical
39: settings for testing the predictions of general relativity. The double
40: neutron-star systems (the prototypical of which contains the
41: Hulse-Taylor pulsar) have led to measurements of general relativistic
42: predictions, such as the periastron precession and the Shapiro delay,
43: and have given the first indirect evidence for the existence of
44: gravitational waves (for a recent review see Will 2001).
45:
46: When compared to solar system tests, however, double neutron star
47: systems have provided only limited constraints, on alternative
48: theories of gravity such as the Brans-Dicke theory (Will 2001). This
49: is mostly due to the fact that the difference in the orbital period
50: evolution of the binaries between general relativity and Brans-Dicke
51: gravity is related to the mass difference of the two neutron stars in
52: each system (Will 2001) and the two members of all double neutron-star
53: systems have very similar masses (see Thorsett \& Chakrabary 1999).
54: Moreover, Brans-Dicke gravity can be constructed to be only slightly
55: different from general relativity, through a single parameter
56: $\omega_{\rm BD}$, making the constraint on $\omega_{\rm BD}$ rather
57: weak.
58:
59: Neutron stars in nature appear in various types of binaries with very
60: small mass companions, which can in principle be used in placing
61: stronger constraints on Brans-Dicke gravity. Tens of millisecond radio
62: pulsars have been discovered in orbits around low mass white dwarfs,
63: but with orbital separations too large for gravitational radiation to
64: affect their orbital period evolution (Phinney \& Kulkarni
65: 1994). Neutron stars with low mass companions in close orbits often
66: appear as bright X-ray sources but their orbital period evolution is
67: dominated by mass transfer and mass loss from the companion stars in
68: the form of a magnetic wind (Verbunt 1993). Moreover, most of these
69: neutron stars show no periodic modulations of their X-ray flux at the
70: stellar spin frequency, hampering measurements of the orbital periods
71: and their evolution (Vaughan et al.\ 1994).
72:
73: The most compact X-ray binary known to date, 4U~1820$-$30, which
74: consists of a neutron star in a 11~min orbit around a $\sim 0.067
75: M_\odot$ companion (Rappaport et al.\ 1987), shows periodic X-ray
76: eclipses, which were used by Morgan, Remillard, \& Garcia (1988) to
77: place an upper limit on its orbital period derivative of
78: $\dot{P}/P<3\times 10^{-7}$~yr$^{-1}$. This limit was subsequently
79: used by Will \& Zaglauer (1989) to constrain the Brans-Dicke parameter
80: to $\omega_{\rm BD}\gtrsim 140$ or $600$, depending on the stiffness
81: of the neutron-star equation of state. However, 4U~1820$-$30 is near
82: the center of the globular cluster NGC~6624 and hence an apparent
83: change of its orbital period may be induced by gravitational
84: acceleration in the potential of the cluster. This has been given as
85: a possible reason for the orbital period {\em decrease\/} at a rate of
86: $\dot{P}/P\simeq -1.1\times 10^{-7}$~yr$^{-1}$ that was later inferred
87: for this source by Tan et al.\ (1991). Furthermore, changes in the
88: pattern of X-ray emission and the geometry of the X-ray eclipses may
89: be responsible for the apparent changes of the orbital period, as
90: suggested by van der Klis et al.\ (1993), who used a longer baseline
91: than Tan et al.\ (1991) and measured a less statistically significant
92: orbital period decrease at a rate of $\dot{P}/P\simeq -5.3\times
93: 10^{-8}$~yr$^{-1}$. Subsequent analysis of archival and more recent
94: data from {\em RXTE\/} by Chou \& Grindlay (2001) gave a marginal
95: detection of orbital period decrease at a rate of $\dot{P}/P\simeq
96: -3.47\times 10^{-8}$~yr$^{-1}$. All these measurements render the
97: constraint on the Brans-Dicke parameter $\omega_{\rm BD}$ from
98: 4U~1820$-$30 quite uncertain.
99:
100: Recently, five accreting millisecond pulsars were discovered by {\em
101: RXTE\/} in very compact binaries with orbital periods between 40~min
102: and 4.3~hr (SAX~J1808.4$-$3658: Chakrabarty \& Morgan 1998;
103: XTE~J1751$-$305: Markwardt et al.\ 2002; XTE~J0929$-$3314: Galloway et
104: al.\ 2002; XTE~J1807$-$294: Markward, Smith, \& Swank 2003;
105: XTE~J1814$-$338: Markwardt \& Swank 2003). Because the primary stars in
106: these systems are millisecond pulsars, their orbital periods can be
107: measured with high accuracy without significant systematic
108: effects. Their orbital separations are very small and hence angular
109: momentum losses via magnetic stellar winds are expected to be
110: minimal. Moreover, all these binaries are in the galactic disk and
111: thus are not subject to significant gravitational accelerations.
112: Therefore, the five accreting millisecond pulsars provide prime
113: candidates for constraining deviations from general relativity and,
114: in particular, for constraining the parameter $\omega_{\rm BD}$ of
115: Brans-Dicke gravity.
116:
117: In this article, I follow the analysis of Will \& Zaglauer (1989) to
118: calculate the constraints on the Brans-Dicke parameter $\omega_{\rm
119: BD}$ imposed by the measurement of an orbital period derivative for an
120: accreting millisecond pulsar. I argue that the expected constraints
121: can be an order of magnitude larger than the constraints from solar
122: system tests. I finally discuss how a limit on (or measurement of) the
123: orbital period derivative can be used in placing a lower bound on the
124: orbital inclinations of the binaries.
125:
126: \section{ORBITAL EVOLUTION IN BRANS-DICKE GRAVITY}
127:
128: In this section, I sketch the derivation of the rate of change of the
129: orbital period of a binary in Brans-Dicke gravity. I follow closely
130: the analysis of Will \& Zaglauer (1989) but make use of more accurate
131: relations that describe the evolution of the companion stars.
132:
133: I consider the evolution of the nearly-circular orbit of a binary
134: consisting of a neutron star, with mass $m_1$, and a low-mass
135: companion, with mass $m_2$. The rate of change of the orbital angular
136: momentum $J=\mu(Gma)$, where $m=m_1+m_2$, $\mu=m_1m_2/m$, and $a$ is
137: the semi-major axis, is
138: \begin{eqnarray}
139: \frac{\dot{J}}{J}&=&\frac{1}{J}\frac{\partial J}{\partial m_1}\dot{m}_1
140: +\frac{1}{J}\frac{\partial J}{\partial m_2}\dot{m}_2
141: +\frac{1}{J}\frac{\partial J}{\partial a}\dot{a}\nonumber\\
142: &=&\left(1-\frac{\beta}{q}-\frac{1}{2}\frac{1-\beta}{1+q}\right)
143: \frac{\dot{m}_2}{m_2}+\frac{1}{2}\left(\frac{\dot{a}}{a}\right)\;,
144: \label{jdot}
145: \end{eqnarray}
146: where $q\equiv m_1/m_2$ is the mass ratio in the binary and
147: $\beta=-\dot{m}_1 /\dot{m}_2$ is a parameter that describes the rate
148: of mass loss from the system, if the mass-transfer is non-conservative.
149:
150: For systems such as the newly discovered ultracompact binaries,
151: orbital angular momentum may be lost because of gravitational-wave
152: radiation or mass loss from the systems. Therefore, the left-hand-side
153: of equation~(\ref{jdot}) is equal to
154: \begin{equation}
155: \frac{\dot{J}}{J}=\frac{\dot{J}_{\rm rad}}{J}+j_{\rm w}(1-\beta)
156: \frac{1+q}{q}\frac{\dot{m_2}}{m_2}\;,
157: \label{jdot2}
158: \end{equation}
159: where $\dot{J}_{\rm rad}$ is the rate of loss of angular momentum
160: caused by gravitational-wave radiation and $j_{\rm w}$ is the specific
161: angular momentum carried away by the stellar wind, in units of $2\pi a^2/P$,
162: where $P$ is the orbital period.
163:
164: I am interested in comparing the predictions of different gravity
165: theories to the observed rate of change of the orbital period, which I
166: calculate using (e.g., Will \& Zaglauer 1989)
167: \begin{equation}
168: \frac{P}{2\pi}=\frac{m}{m_1^3 m_2^3}J^3 G^{-2}
169: \end{equation}
170: so that
171: \begin{equation}
172: \frac{\dot{P}}{P}=\left(\frac{1-\beta}{1+q}-
173: \frac{3}{2}\frac{1-\beta}{1+q}\right)\frac{\dot{m}_2}{m_2}
174: +\frac{3}{2}\frac{\dot{a}}{a}\;.
175: \label{Pdot1}
176: \end{equation}
177:
178: In order to evaluate the rate of change of the semi-major axis, I will
179: assume that the companion star always fills its Roche lobe, i.e., that
180: its radius $R_2$ is equal to (Eggleton 1983)
181: \begin{equation}
182: R_2=\frac{0.49 q^{-2/3}}{0.6q^{-2/3}+\ln(1+q^{-1/3})}a\;.
183: \end{equation}
184: Using the fact that
185: \begin{equation}
186: \frac{\dot{q}}{q}=-\left(\frac{\beta+q}{q}\right)\frac{\dot{m}_2}{m_2}\;,
187: \end{equation}
188: I can evaluate the rate of change of the companion's radius to be
189: \begin{equation}
190: \frac{\dot{R}_2}{R_2}=\frac{\dot{a}}{a}+\frac{2}{3}\left(\frac{\beta+q}{q}
191: \right)
192: \left[1-\frac{0.6+0.49q^{1/3}(1+q^{-1/3})^{-1}}{0.6+q^{2/3}\ln(1+q^{-1/3})}
193: \right]\frac{\dot{m}_2}{m_2}\;.
194: \end{equation}
195: I can then turn the last equation into
196: \begin{equation}
197: \frac{\dot{a}}{a}=\left\{\xi_{\rm ad}-
198: \frac{2}{3}\left(\frac{\beta+q}{q}
199: \right)
200: \left[1-\frac{0.6+0.49q^{1/3}(1+q^{-1/3})^{-1}}{0.6+q^{2/3}\ln(1+q^{-1/3})}
201: \right]\right\}\frac{\dot{m}_2}{m_2}\;,
202: \label{adot}
203: \end{equation}
204: where I have introduced the adiabatic index $\xi_{\rm ad}\equiv d\ln R_2/
205: d\ln m_2$ for the companion star.
206:
207: I now combine equations~(\ref{jdot}), (\ref{jdot2}), (\ref{Pdot1}), and
208: (\ref{adot}) to obtain
209: \begin{equation}
210: \frac{\dot{P}}{P}=3\left(\frac{n}{D}\right)\frac{\dot{J}_{\rm rad}}{J}\;,
211: \label{Pdot}
212: \end{equation}
213: where
214: \begin{equation}
215: n\equiv\frac{1}{2}\left(\xi_{\rm ad}-\frac{1}{3}\frac{1-\beta}{1+q}
216: -\frac{2}{3}\frac{\beta+q}{q}{\cal A}\right)\;,
217: \end{equation}
218: \begin{equation}
219: D\equiv 1+\frac{1}{2}\xi_{\rm ad}-\frac{1}{2}\left(\frac{1-\beta}{1+q}\right)-
220: \frac{1}{q}\left[\beta+j_w(1-\beta)(1+q)\right]-\frac{1}{3}\frac{\beta+q}{q}
221: {\cal A}\;,
222: \end{equation}
223: and
224: \begin{equation}
225: {\cal A}\equiv \left[1-\frac{0.6+0.49q^{1/3}(1+q^{-1/3})^{-1}}{0.6+q^{2/3}\ln(1+q^{-1/3})}
226: \right]\;.
227: \end{equation}
228:
229: Equation~(\ref{Pdot}) is more general than the one derived by Will \&
230: Zaglauer (1989), as it is valid for a wider range of mass ratios of
231: the binary. It provides the predicted rate of change of the orbital
232: period of a binary, given a rate of angular momentum lost by the
233: emission of gravitational waves and by mass loss from the binary. For
234: a Brans-Dicke gravity, this rate is equal to (Will \& Zaglauer 1989)
235: \begin{equation}
236: \frac{\dot{J}_{\rm rad}}{J}=-\frac{1}{3}\left[\frac{96}{5}\frac{\mu m^2}{a^4}
237: \left(\frac{k_1}{12}\right)+\frac{2\mu m}{a^3}{\cal G}\xi s_1^2\right]\;,
238: \end{equation}
239: where
240: \begin{equation}
241: {\cal G}\equiv 1-\xi s_1\;,
242: \end{equation}
243: \begin{equation}
244: k_1={\cal G}^2\left[12\left(1-\frac{1}{2}\xi\right)+\xi
245: \left(1-2\frac{m_1 s_1}{m_1+m_2}\right)\right]\;,
246: \end{equation}
247: \begin{equation}
248: s\equiv -\left.\frac{\partial \ln m_1}{\partial G}\right\vert_0
249: \end{equation}
250: is the so-called neutron-star sensitivity, and
251: \begin{equation}
252: \xi\equiv\frac{1}{2+\omega_{\rm BD}}
253: \label{xi}
254: \end{equation}
255: is a parameter that describes the deviation of Brans-Dicke gravity
256: from general relativity. Note that general relativity corresponds to
257: $\omega_{\rm BD}\rightarrow \infty$ and hence to $\xi=0$. In writing
258: the above equations, I assumed that the companion to the neutron star
259: is a non-relativistic star and hence the sensitivity of the latter is
260: negligible.
261:
262: Given a rate of change of the orbital period of the binary
263: ($\dot{P}/P$), the orbital parameters of the binary (i.e., $m_1$,
264: $m_2$, $P$, and $a$), the properties of the neutron star (i.e., the
265: value of $s$) and of the companion star (i.e., the value of $\xi_{\rm ad}$),
266: and the properties of mass loss in the binary (i.e., the values of
267: $\beta$ and $j_{\rm w}$), equations~(\ref{Pdot})-(\ref{xi}) can be
268: used to place a constraint on the Brans-Dicke parameter $\omega_{\rm
269: BD}$.
270:
271: \section{APPLICATION TO ULTRACOMPACT BINARIES}
272:
273: In section \S2, I sketched (following Will \& Zaglauer 1989) the
274: derivation of the rate of change of the orbital period of an
275: ultracompact binary, when orbital angular momentum is lost due to
276: emission of gravitational radiation, in Brans-Dicke gravity. In this
277: section, I will discuss the observed properties of the recently
278: discovered ultracompact binaries and their prospect for constraining
279: deviations of this theory from general relativity.
280:
281: \noindent {\em Properties of the binaries.---\/}Five accreting
282: millisecond pulsars have been discovered to date, in ultracompact
283: binaries with orbital periods between 40~minutes and 4.3 hours. Their
284: orbital periods $P_{\rm orb}$, projected semi-major axes $a$,
285: eccentricities $e$, and mass functions $f$ are summarized in Table~1.
286:
287: \begin{table}[t]
288: \centerline{
289: \footnotesize
290: \begin{tabular}{lcccl}
291: \multicolumn{5}{c}{Table 1: Observed Properties of Ultracompact Binaries}\\
292: \hline
293: Source & $P_{\rm orb}$ (min) & $a$ (lt-ms) & $f$~$(M_\odot)$ & Reference \\
294: \hline
295: SAX~J1808.4$-$3658 & 120.9 & 62.809
296: & $3.78\times 10^{-5}$ & Chakrabarty \& Morgan 1998\\
297: XTE~J0929$-$3314 & 43.6 & 6.290
298: & $2.7\times 10^{-7}$ & Galloway et al.\ 2002\\
299: XTE~J1751$-$305 & 42.4 & 10.1134
300: & $1.278\times10^{-6}$ & Markwardt et al.\ 2002\\
301: XTE~J1807$-$294 & 40.1 & 4.80 & $1.54\times 10^{-7}$
302: & Markwardt et al.\ 2003; Markwardt priv.\ comm.\\
303: XTE~J1814$-$338 & 256.5 & 390.3 & $2.016\times 10^{-3}$
304: & Markwardt \& Swank 2003; Markwardt priv.\ comm.\\
305: \hline
306: \end{tabular}}
307: \end{table}
308:
309: The constraint on the parameter $\omega_{\rm BD}$ depends on the
310: measured orbital period and semi-major axis of each orbit, as well as
311: on the mass of the neutron star and of the companion star. A
312: combination of observational and theoretical arguments strongly
313: constrain the mass of the former to lie in the narrow range $\simeq
314: 1.3-2.2 M_{\odot}$ (see, e.g., Lattimer \& Prakash 1999). The mass of
315: the companion star is then calculated as a function of the unknown
316: inclination $i$ of the binary using the mass function, i.e.,
317: \begin{equation}
318: f=\frac{(m_2 \sin i)^3}{(m_1+m_2)^2}\;.
319: \label{f}
320: \end{equation}
321: For reasons related to the stability of mass transfer in X-ray binary
322: systems, I will only consider cases in which the companion star is
323: less massive than the neutron star.
324:
325: For the parameters of the ultracompact binaries discussed here, the
326: minimum absolute value of the rate of change of the orbital period
327: increases with increasing mass of the neutron star or the companion
328: star. This is caused by the fact that both the semi-major axis of a
329: binary of given period and the rate of emission of gravitational waves
330: increase with increasing total mass of the system. As
331: equation~(\ref{f}) shows, the limiting absolute value for the rate of
332: change of the orbital period corresponds to the maximum inclination (i.e,
333: $\sin i=1$) and the minimum accepted value for the neutron-star mass, which
334: we take to be 1.3~$M_\odot$.
335:
336: \begin{figure}[t]
337: \centerline{
338: \psfig{file=f1.ps,angle=0,width=10truecm}}
339: \figcaption[]{\footnotesize The rate of orbital period
340: evolution ($\tau_{\rm P}^{-1}\equiv \dot{P}/P$) as a function of the
341: Brans-Dicke parameter $\omega_{\rm BD}$, for different values of the
342: neutron-star sensitivity $s$. For this plot, I used the orbital
343: parameters of XTE~J1808.4$-$3658 and assumed a neutron star mass of
344: $1.3 M_\odot$, no mass loss, an edge-on orbit, and $\xi_{\rm ad}=0$ for
345: the companion star.
346: \label{fig:NSsens}}
347: \end{figure}
348:
349: \noindent {\em Properties of the neutron stars.---\/}The
350: separations of the ultracompact binaries under study are much larger
351: than the radii of the neutron stars. Therefore, in general relativity,
352: the specifics of the neutron-star structure, and hence its equation of
353: state, do not enter in the calculation of the angular momentum lost
354: due to gravitational radiation. However, this is not true in
355: Brans-Dicke gravity, in which the self-gravitational binding energies
356: of the neutron stars (i.e., their sensitivities $s$) affect the result
357: (Will \& Zaglauer 1989).
358:
359: A number of values for the sensitivity have been calculated
360: by Will \& Zaglauer (1989) and by Zaglauer (1990) using the relation
361: \begin{equation}
362: s\equiv=-\left(\frac{\partial \ln m_1}{\partial \ln G}\right)_N=\frac{3}{2}
363: \left[1-\left(\frac{\partial \ln m_1}{\partial \ln N}\right)_G\right]\;,
364: \end{equation}
365: where $N$ is the baryonic number in the star. According to these
366: calculations, the sensitivity increases with increasing mass and with
367: the stiffness of the equation of state, as both make the stars more
368: compact and hence increase their self-gravitational binding energies.
369:
370: The minimum absolute value of the rate of change of the orbital period
371: also increases with increasing neutron star sensitivity, as shown for
372: a typical case in Figure~\ref{fig:NSsens}. The most stringent limit on
373: the Brans-Dicke parameter $\omega_{\rm BD}$ will, therefore, be obtained
374: when using the smallest sensitivity for the lightest neutron star
375: considered. For the rest of the paper, I will assume this to be equal
376: to 0.2.
377:
378: \begin{figure}[t]
379: \centerline{
380: \psfig{file=f2.ps,angle=0,width=10truecm}}
381: \figcaption[]{\footnotesize The rate of orbital period
382: evolution ($\tau_{\rm P}^{-1}\equiv \dot{P}/P$) as a function of the
383: Brans-Dicke parameter $\omega_{\rm BD}$, for different values of the
384: adiabatic index $\xi_{\rm ad}$ of the companion star. For this plot, I
385: assumed that the neutron star sensitivity is equal to 0.2 and used the
386: same parameters as in Fig.~\ref{fig:NSsens}.
387: \label{fig:ksiad}}
388: \end{figure}
389:
390: {\em Properties of the companion stars.---\/}The response of the
391: companion star to mass loss is described by the adiabatic index
392: $\xi_{\rm ad}$ and depends on the nature of the star as well as on a
393: number of factors such as the presence of external irradiation and the
394: rate of mass loss. If all external effects are neglected and the star
395: is considered to be a polytrope, the parameter $\xi_{\rm ad}$ takes the
396: canonical value of $\xi_{\rm ad}=-1/3$.
397:
398: The companions to the neutron stars in the ultracompact binaries under
399: consideration are believed to be white dwarfs (see Deloye \& Bildsten
400: 2003 and references therein), given their inferred very low masses and
401: small sizes (both are of order a few hundredths of the solar mass and
402: radius, respectively). The response of such stars to adiabatic mass
403: loss was recently calculated in detail by Deloye \& Bildsten (2003), who
404: found that the corresponding parameter $\xi_{\rm ad}$ decreases with
405: increasing mass of the white dwarf and with decreasing atomic weight
406: of the main compositional element, but depends rather weakly on its
407: temperature. For the inferred values of the masses of the white
408: dwarfs in the ultracompact binaries, they showed that $-0.3\lesssim
409: \xi_{\rm ad}\lesssim 0$.
410:
411: The dependence of the rate of orbital period evolution on the
412: parameter $\xi_{\rm ad}$ is shown in Figure~\ref{fig:ksiad}. Clearly, for a
413: weaker response of the companion star (i.e., for smaller values of
414: $\vert\xi_{\rm ad}\vert$), the rate of orbital period change is also
415: smaller. This is expected, given our assumption that the companion
416: star always fills its Roche lobe for mass transfer to occur and,
417: therefore, a small change in the radius of the star caused by mass
418: loss will be accompanied by a small change in the orbital
419: period. Because the weakest companion-star response corresponds to the
420: lowest mass stars (see Deloye \& Bildsten 2003) and the latter also
421: produce the lowest rate of orbital period change, using the lowest
422: allowed value of $\vert\xi_{\rm ad}\vert$ for the lowest companion mass in each
423: binary will provide the most stringent limit.
424:
425: \begin{figure}[t]
426: \centerline{
427: \psfig{file=f3.ps,angle=0,width=10truecm}}
428: \figcaption[]{\footnotesize The rate of
429: orbital period evolution ($\tau_{\rm P}^{-1}\equiv \dot{P}/P$) as a
430: function of the mass-loss parameter $\beta$, for different values of
431: the angular momentum of the wind, $j_{\rm w}$. The binary parameters
432: are the same as in Figure~1, the neutron-star sensitivity is set to
433: $s=0.2$, the adiabatic index for the companion star is set to
434: $\xi_{\rm ad}=0$ and the Brans-Dicke parameter is set to $\omega_{\rm
435: BD}=1000$. The lower half of the plot corresponds to an orbital period
436: that decreases with time ($\dot{P}/P<0$), whereas the upper half
437: corresponds to an orbital period that increases with time
438: ($\dot{P}/P>0$).
439: \label{fig:mloss}}
440: \end{figure}
441:
442: {\em Properties of the mass loss.---}The last and hardest to constrain
443: uncertainty in the calculation of the predicted rate of orbital period
444: change arises from the properties of the mass loss from the binary,
445: i.e., the parameters $\beta$ and $j_{\rm w}$. Even though the
446: companions to the neutron stars in the ultracompact binaries are white
447: dwarfs and lie deep in the gravitational potential wells of the neutron
448: stars, mass from their surfaces may be lost from the binary systems
449: because of the ablation caused by the intense X-ray irradiation.
450:
451: The dependence of the predicted rate of change of the orbital period
452: on the amount ($\beta$) and strength ($j_w$) of mass loss is shown in
453: Figure~\ref{fig:mloss}. Clearly, when there is very little mass loss,
454: i.e., when $\beta\rightarrow 1$, the orbital period increases with
455: time and the predicted rate depends very weakly on $j_{\rm w}$. The
456: situation is very different, however, when most of the mass is lost
457: from the system, i.e., when $\beta\rightarrow 0$.
458:
459: In order to understand this dependence, I rewrite
460: equation~(\ref{Pdot}) taking the limit $q\gg 1$, which is appropriate
461: for the ultracompact binaries under study. In this case, ${\cal
462: A}\rightarrow 1/2$ and
463: \begin{equation}
464: \frac{\dot{P}}{P}\simeq 3 \left[\frac{\xi_{\rm ad}-\frac{1}{3}}
465: {\frac{5}{3}+\xi_{\rm ad}-
466: j_{\rm w}(1-\beta)}\right]\frac{\dot{J}_{\rm rad}}{J}\;.
467: \label{Pdotaprx}
468: \end{equation}
469: In the absence of mass loss, equation~(\ref{Pdotaprx}) shows that,
470: even though angular momentum is lost from the companion star due to
471: the emission of gravitational waves, i.e., $\dot{J}_{\rm rad}<0$, the
472: period of the binary increases, as long as $\xi_{\rm ad}<1/3$. This is a direct
473: consequence of the assumption that the companion star always fills its
474: Roche lobe for mass transfer to occur and of the fact that the radii
475: of these low-mass white dwarfs increase with decreasing mass. However,
476: if a significant amount of angular momentum is removed by the wind in
477: addition to the gravitational radiation, i.e., if
478: \begin{equation}
479: j_{\rm w}(1-\beta)>\frac{5}{3}+\xi_{\rm ad}\;,
480: \end{equation}
481: then the orbit of the binary shrinks and the orbital period decreases.
482:
483: It is important to note, however, that, whether there is mass loss or
484: not, there is always a lower limit on the absolute value of the rate
485: of change of the orbital period, since
486: \begin{equation}
487: \left\vert\frac{\dot{P}}{P}\right\vert\simeq
488: \frac{1-3\xi_{\rm ad}}{\vert\frac{5}{3}+\xi_{\rm ad}+j_{\rm w} \beta-j_{\rm w}\vert}
489: \left\vert\frac{\dot{J}_{\rm rad}}{J}\right\vert \ge
490: \frac{1-3\xi_{\rm ad}}{\frac{5}{3}+\xi_{\rm ad}+j_{\rm w} \beta+j_{\rm w}}
491: \left\vert\frac{\dot{J}_{\rm rad}}{J}\right\vert
492: \label{Pdotmin}
493: \end{equation}
494: Equation~(\ref{Pdotaprx}) shows that the minimum positive rate of
495: change of the orbital period corresponds to the limit
496: $\beta\rightarrow 1$, whereas the maximum negative rate of change of
497: the orbital period corresponds to the limit $\beta \rightarrow 0$ and
498: $j_{\rm w}\rightarrow 1$ (see also Fig.~\ref{fig:mloss}). Moreover,
499: the absolute values of these two limits are comparable in size. This
500: fact allows for stringent constraints to be placed on the Brans-Dicke
501: parameter $\omega_{\rm BD}$ by the measurement of an orbital period
502: derivative, independent of whether the latter has a positive or
503: negative value.
504:
505: \section{RESULTS AND DISCUSSION}
506:
507: In \S3, I studied the rate of orbital period evolution in an
508: ultracompact binary and its dependence on the various model
509: parameters. When the binary period is increasing with time, I found
510: that the rate of evolution is minimized for the lowest neutron-star
511: mass (taken here to be $\ge 1.3M_\odot$), the largest inclination
512: ($\sin i=1$), the largest adiabatic index for the companion star (take
513: here to be $\xi_{\rm ad}\le0$), and the case of no mass-loss
514: ($\beta=1$). On the other hand, when the binary period is decreasing
515: with time, the absolute value of this rate is minimized for the same
516: stellar parameters as before but at the limit of complete mass loss
517: ($\beta=0$) and with the wind carrying the orbital angular momentum
518: ($j_{\rm w}=1$).
519:
520: \begin{figure}[t]
521: \centerline{
522: \psfig{file=f4.ps,angle=0,width=10truecm}}
523: \figcaption[]{\footnotesize The limiting rate for the evolution of the
524: orbital period ($\tau_{\rm P}^{-1}\equiv \dot{P}/P$) as a function of the
525: Brans-Dicke parameter $\omega_{\rm BD}$, for the orbital parameters of
526: the five known millisecond accreting pulsars. The lower half of the
527: plot corresponds to an orbital period that decreases with time
528: ($\dot{P}/P<0$), whereas the upper half corresponds to an orbital
529: period that increases with time ($\dot{P}/P>0$). \label{fig:constr}}
530: \end{figure}
531:
532: Figure~\ref{fig:constr} shows the limiting rate of orbital period
533: evolution as a function of the Brans-Dicke parameter $\omega_{\rm
534: BD}$, for the five known accreting millisecond pulsars. For each
535: source, the region between the two corresponding curves is
536: excluded. As a result, a measurement of either a positive or a
537: negative rate of orbital period evolution can be used in placing a
538: lower bound on the Brans-Dicke parameter.
539:
540: The limiting curves become insensitive to the deviation of Brans-Dicke
541: gravity from general relativity for $\omega_{\rm BD}\gtrsim 10^4$.
542: This number represents the tightest constraint that can be achieved
543: with this method and is approximately an order of magnitude larger
544: than the constraints imposed by solar-system and double-neutron-star
545: tests (Will 2001). In deriving these constraining curves, I assumed no
546: prior knowledge of any of the binary parameters other than the ones
547: that can be inferred from X-ray timing. These constraints can be
548: improved by measuring the masses of the neutron star and the companion
549: star, by identifying the nature of the companions (and hence
550: constraining their adiabatic indices $\xi_{\rm ad}$), and by placing
551: constraints on the mass loss from the systems. The tightest limits can
552: be achieved for an eclipsing ultracompact binary, with a relatively
553: long orbital period, and limited mass loss.
554:
555: \begin{figure}[t]
556: \centerline{
557: \psfig{file=f5.ps,angle=0,width=10truecm}}
558: \figcaption[]{\footnotesize The limiting rate for the evolution of the
559: orbital period ($\tau_{\rm P}^{-1}\equiv \dot{P}/P$) of
560: SAX~J1808.4$-$3658, in general relativity, as a function of the
561: orbital inclination. The lower half of the plot corresponds to an
562: orbital period that decreases with time ($\dot{P}/P<0$), whereas the
563: upper half corresponds to an orbital period that increases with time
564: ($\dot{P}/P>0$). \label{fig:incl}}
565: \end{figure}
566:
567: Even when the binary parameters are not favorable, and hence the
568: resulting constraints on the Brans-Dicke parameter $\omega_{\rm BD}$
569: are not stringent, the analysis presented here can be used in
570: constraining the orbital inclination of the ultracompact binaries and
571: the properties of mass loss. As an example, Figure~\ref{fig:incl}
572: shows the limiting rate of evolution of the orbital period of the
573: source SAX~J1808.4$-$3658, in general relativity, as a function of the
574: orbital inclination. The region between the two curves is not allowed
575: for any neutron-star mass in the range $1.3-2.2 M_{\odot}$ and for any
576: mass-loss mechanism from the binary. As a result, a measurement of the
577: rate of orbital period evolution can also be used in placing a lower
578: bound on the inclination of the binary system.
579:
580: The measurements required to place a stringent constraint on the
581: Brans-Dicke parameter $\omega_{\rm BD}$ can be achieved using the
582: Rossi X-ray Timing Explorer by measuring the orbital periods of the
583: binaries over a long period of time, i.e., between successive
584: outbursts. Indeed, the source SAX~J1808.4$-$3658 has shown four
585: outbursts within six years and the accuracy of the measurement of the
586: orbital period in the 1998 outburst alone is comparable to the value
587: needed to achieve a useful constraint (see Chakrabarty \& Morgan
588: 1998). However, the properties of mass loss from the binary system may
589: not remain constant over a long period of time. If the mass loss is
590: driven by irradiation of the companion star, then during an outburst
591: the mass loss will be significant and the orbital period may be
592: decreasing with time, whereas in between outbursts the mass loss may be
593: negligible and the orbital period will be increasing with time. The
594: net result will be an artificially reduced rate of orbital period
595: evolution and thus an artificially stringent constraint on Brans-Dicke
596: gravity. The effects of a variable rate of mass loss can be minimized
597: if an orbital period derivative can be measured in a single, long
598: outburst. Given the short durations of the outbursts so far observed
599: from the five known sources, such a measurement is unlikely with the
600: capabilities of RXTE. It can be, however, one of the key scientific
601: goals of the next X-ray timing mission (see, e.g., Markwardt 2004).
602:
603: \acknowledgements
604: It is a pleasure to thank Duncan Galloway for help in compiling the
605: data in the table, Deepto Chakrabarty for useful discussions, and
606: Feryal \"Ozel for a critical reading of the manuscript. I am also
607: grateful for their hospitality to the Astrophysics Group at Sabanci
608: University, in Istanbul, Turkey, where this work was completed.
609:
610: {\em Note added.---\/}While this work was in its final stages, a new
611: accreting millisecond pulsar, IGR~J00291$+$5934, was discovered in an
612: ultracompact binary (Markwardt et al.\ 2004, ATEL \#353, \#360). This
613: new source has very similar orbital parameters with the five previously
614: known accreting millisecond pulsars and can also be used in placing
615: constraints on Brans-Dicke gravity.
616:
617: \begin{references}
618:
619: \reference{CM98} Chakrabarty, D., \& Morgan, E.\,H.\ 1998, \nat, 394, 346
620:
621: \reference{CG01} Chou, Y., \& Grindlay, J.\,E.\ 2001, \apj, 563, 934
622:
623: \reference{Getal02} Galloway, D.~K., Chakrabarty, D.,
624: Morgan, E.\,H., \& Remillard, R.\,A.\ 2002, \apjl, 576, L137
625:
626: \reference{M04} Markwardt, C.\,B.\ 2004, in X-ray Timing 2003: Rossi
627: and Beyond, eds.\ P.\ Kaaret, F.\,K.\ Lamb, and J.\,H.\ Swank (AIP)
628:
629: \reference{Metal02} Markwardt, C.\,B., Smith, E, \& Swank, J.\,H.\
630: 2003, ATEL 122
631:
632: \reference{Metal02} Markwardt, C.\,B., \& Swank, J.\,H.\ 2003, IAUC 8144
633:
634: \reference{Metal02} Markwardt, C.\,B., Swank, J.\,H., Strohmayer, T.\,E.,
635: in't Zand, J.\,J.\,M., \& Marshall, F.\,E.\ 2002, \apjl, 575, L21
636:
637: \reference{MRG88} Morgan, E.\,H., Remillard, R.\,A., \& Garcia, M.\,R.\ 1988,
638: \apj, 324, 851
639:
640: \reference{PK94} Phinney, E.\,S., \& Kulkarni, S.\,R.\ 1994, \araa, 32, 591
641:
642: \reference{Retal87} Rappaport, S., Ma, C.\,P., Joss, P.\,C., \& Nelson,
643: L.\,A.\ 1987, \apj, 322, 842
644:
645: \reference{Tetal91} Tan, J.~et al.\ 1991, \apj, 374, 291
646:
647: \reference{TC99} Thorsett, S., \& Chakrabarty, D.\ 1999, \apj
648:
649: \reference{Ketal93} van der Klis, M., Hasinger, G., Verbunt, F.,
650: van Paradijs, J., Belloni, T., \& Lewin, W.\,H.\,G.\ 1993, \aap, 279, L21
651:
652: \reference{V94} Vaughan, B.\,A.\ et al.\ 1994, \apj, 435, 362
653:
654: \reference{V93} Verbunt, F.\ 1993, \araa, 31, 93
655:
656: \reference{W01} Will, C.\ M. 2001, Living Rev. Relativity 4,(Online article:
657: cited on 15 Aug 2001;\\ http://www.livingreviews.org/lrr-2001-4)
658:
659: \reference{WZ89} Will, C.\,M., \& Zaglauer, H.\,W.\ 1989, \apj, 346, 366
660:
661: \end{references}
662:
663: \end{document}
664:
665:
666:
667:
668:
669:
670:
671:
672:
673: