1: \documentclass[12pt,preprint]{aastex}
2: \shorttitle{Bifurcation periods in LMXBs} \shortauthors{MA \& Li.}
3:
4: \begin{document}
5:
6: %% LaTeX will automatically break titles if they run longer than
7: %% one line. However, you may use \\ to force a line break if
8: %% you desire.
9:
10: \title{The bifurcation periods in low-mass X-ray binaries: the effect of magnetic braking
11: and mass loss}
12:
13:
14: \author{Bo Ma and Xiang-Dong Li}
15: \affil{Department of Astronomy, Nanjing University, Nanjing 210093,
16: China} \email{xiaomabo@gmail.com, lixd@nju.edu.cn}
17:
18:
19: %% Mark off your abstract in the ``abstract'' environment. In the manuscript
20: %% style, abstract will output a Received/Accepted line after the
21: %% title and affiliation information. No date will appear since the author
22: %% does not have this information. The dates will be filled in by the
23: %% editorial office after submission.
24:
25: \begin{abstract}
26: The bifurcation period in low-mass X-ray binaries is the initial
27: orbital period which separates the formation of converging systems
28: (which evolve with decreasing orbital periods until the donor
29: becomes degenerate) from the diverging systems (which evolve with
30: increasing orbital periods until the donor star loses its envelope
31: and a wide detached binary is formed). We calculate systematically
32: the bifurcation periods of binary systems with a $1.4M_\sun$ neutron
33: star and a $0.5-2 M_\sun$ donor star, taking into account different
34: kinds of magnetic braking and mass loss mechanisms. Our results show
35: that the saturated magnetic braking can considerably decrease the
36: values of bifurcation period compared to the traditional magnetic
37: braking, while the influence of mass loss mechanisms on bifurcation
38: periods is quite weak. We also develop a semi-analytical method to
39: compute the bifurcation period, the result of which agrees well with
40: the numerical method in the leading order.
41: \end{abstract}
42:
43: \keywords{binaries: close -- stars: evolution -- X-rays: binaries}
44:
45: \section{Introduction}
46:
47: %Binary and millisecond pulsars (BMSPs) with a low-mass companion
48: %have been suggested to be descendants of \citep[e.g.][]{alp82}.
49: %Formation of BMSPs can be described by the ``recycled" scenario
50: %\citep{bha91}. This model has gained strong support in recent years
51: %from the discoveries of quasi-periodic kiloHertz oscillations as
52: %well as millisecond X-ray pulsations in a number of LMXBs
53: %\citep[][for a review]{kli06}
54:
55: One interesting and important topic in the secular evolution of
56: low-mass X-ray binaries (LMXBs) is the so-called ``bifurcation
57: period" $P_{\rm bif}$, the initial binary orbital period which
58: separates the formation of converging systems (which evolve with
59: decreasing orbital periods until the donor becomes degenerate) from
60: the diverging systems (which evolve with increasing orbital periods
61: until the donor star loses its envelope and a wide detached binary
62: is formed) \citep{tutukov85}. The first systematic investigations on
63: the bifurcation period were done by \citet{pylyser88,pylyser89}.
64: Neglecting mass loss from the binary system and assuming angular
65: momentum loss due to magnetic braking \citep[MB;][]{verbunt81} and
66: gravitational radiation \citep[GR;][]{landau}, these authors found
67: that the bifurcation period is in the range $P_{\rm bif}\sim
68: 0.4-0.7$ day for LMXBs, and strongly depends on magnetic braking
69: efficiency. \citet{ergma98} included mass loss from the binary
70: system and re-calculated the bifurcation period for two mass
71: configurations ($M_1/M_\sun$, $M_2/M_\sun)= (1.4,1)$ and $(1.4,1.5)$
72: and two chemical compositions ($Z=0.003$, $0.03$). They pointed out
73: that the mass loss from the binary system also plays an impotent
74: role besides magnetic braking in determining the value of $P_{\rm
75: bif}$, while the chemical composition could only cause small change
76: in $P_{\rm bif}$. Their bifurcation periods are $P_{\rm bif}\sim
77: 0.85-1.05$ day under conservative mass transfer, and $1.6-1.7$ times
78: larger if moderate non-conservative mass transfer is assumed.
79: \citet{pod02} found a bifurcation period around $18$ hr for a
80: $1.4M_\sun$ NS and a $1M_\sun$ companion star, where they defined
81: the bifurcation period as the orbital period when the Roche lobe
82: overflow just began, instead of the initial orbital period.
83:
84: \citet{sluys05a,sluys05b} also investigated the bifurcation period
85: in LMXBs focusing the formation of ultra-compact X-ray binaries
86: (UCXBs), and specified the bifurcation period as ``the longest
87: initial period that leads to UCXBs within a Hubble time ($13.7$
88: Gyr)". UCXBs are bright X-ray sources with very short orbital
89: periods ($P \la 1$ h). The donor has to be a compact source like a
90: white dwarf or a compact core of an evolved giant star to fit in the
91: small Roche lobe size. Such sources may be formed through dynamical
92: processes including stellar collisions and common envelop evolution
93: \citep{clark75,rasio00,lombard06}. An alternative scenario for the
94: formation of such sources is through stable mass transfer in X-ray
95: binaries with a low- or intermediate-mass donor star, which may
96: explain the negative derivative of the 11-min source in NGC 6624
97: \citep{klis93,chou01}. It has been found that systems with initial
98: orbital period just below the bifurcation period may form UCXBs
99: \citep{nelson86,tutukov87,pylyser88,pod02,sluys05a}. \citet{pod02}
100: showed that the closer the initial orbital period to the bifurcation
101: period from below, the smaller the minimum orbital period will be
102: achieved. So the value of bifurcation period is crucial to
103: understanding the formation of UCXBs \citep{sluys05b}.
104:
105: In this paper we make a systematic investigation on the bifurcation
106: period for binary systems containing an NS with a main-sequence (MS)
107: companion of mass from $0.5M_\sun$ to $2M_\sun$. This work was
108: motivated by recent progress in studies on mass and angular momentum
109: loss mechanisms in LMXB evolution. In previous works the MB law
110: originally postulated by \citet{verbunt81} and \citet{rappaport83}
111: was usually adopted. However, this law predicts too fast spin-down
112: of low-mass MS stars, contradicted with the observation of rapid
113: rotators in young open clusters \citep{sills00,andronov03}.
114: Obviously a modification of the MB law will have significant
115: influence on the period evolution \citep{sluys05b}. Additionally, there is strong
116: evidence that during LMXB evolution the mass transfer is highly
117: non-conservative. Recent measurements of the masses of binary and
118: millisecond pulsars indicate that a large fraction of the
119: transferred mass may be lost from the systems rather accreted by the
120: NS \citep[][and references therein]{bassa06,steeghs07}\footnote{The
121: massive ($\sim 1.9 M_{\sun}$) NS discovered in the globular cluster
122: M5 \citep{freire08} may reflect a bimodal distribution of the
123: initial masses of NSs (rather heavy accretion during the previous
124: LMXB evolution), as already predicted by hydrodynamical core
125: collapse simulations \citep{timmes96}.}. Theoretically, possible
126: ways of mass loss have been suggested, including ``evaporation" of
127: the donor \citep{ruderman89} or ``radio-ejection" of the transferred
128: material \citep{burderi01,burderi02,antona06} due to the pulsar
129: radiation/wind impinging on. In the latter case, the matter is lost
130: from the system at the inner Lagrangian ($L_1$) point, carrying away
131: angular momentum and altering the period evolution.
132:
133: This paper is organized as follows. \S 2 briefly describes the
134: stellar evolution code, the binary models, and the physical
135: assumptions, especially the MB laws and the mass loss mechanisms.
136: Then we present the calculated results in \S 3. Our discussion and
137: conclusions are given in \S 4.
138:
139:
140: %% In our model, once mass transfer has commenced, it is sustained by either (1) systemic angular momentum
141: %%losses (e.g., magnetic braking or
142: %% gravitational radiation) or (2) expansion of the donor star due to nuclear and/or thermal evolution.
143:
144: \section{Evolution code and binary model}
145: \subsection{The stellar evolution code}
146: We use an updated version of the stellar evolution code originally
147: developed by \citet[][see also Han et al. 2004, Pols et al.
148: 1995]{eggleton71,eggleton72} to calculate the evolutions of binaries
149: consisting of an NS (of mass $M_1$) and an MS secondary (of mass
150: $M_2$). For the secondary star we assume a solar chemical
151: composition ($X=0.70$, $Y=0.28$, and $Z=0.02$), the ratio of mixing
152: length to pressure scale height $\alpha= 2.0$, and convective
153: overshooting parameter to be $0.12$. The opacity table is from
154: \citet{rogers92}, \citet{alexander94} and \citet{hubbard69}. The
155: effective radius of the Roche lobe for the secondary is taken from
156: \citet{eggleton83},
157: \begin{equation}
158: R_{\rm{L,2}}=\frac{0.49q^{-2/3}}{0.6q^{-2/3}+\ln(1+q^{-1/3})}a
159: \end{equation}
160: where $q=M_2/M_1$ is the mass ratio, and $a$ is the orbital
161: separation. Mass transfer rate via Roche lobe overflow is evaluated
162: as $-\dot{M}_2=RMT\cdot max(0,(R_2/R_{\rm
163: L,2}-1)^3)M_{\sun}$yr$^{-1}$, and we adopt $RMT=10^3$ in the
164: calculations.
165:
166: \subsection{Mass and Angular momentum loss mechanisms}
167: For LMXBs the timescale of tidal synchronization is much shorter
168: than the characteristic evolutionary timescale of the binary, so we
169: can assume that the spin of the secondary star and the binary
170: orbital revolution are always synchronized. Assuming rigid body
171: rotation of the secondary star and neglecting the spin angular momentum
172: of the neutron star, the total angular momentum of the binary system
173: can be expressed as
174: \begin{eqnarray}
175: J&=&I_2\omega+J_{\rm orb} \nonumber \\
176: &=&I_2\omega+G^{2/3}M_1M_2(M_1+M_2)^{-1/3}\omega^{-1/3}
177: \end{eqnarray}
178: where $I_2$ is the moment of inertia of the secondary star, $\omega$
179: is the angular velocity of the binary.
180:
181: We consider three kinds of mechanisms of angular momentum loss. The
182: first is the angular momentum loss due to gravitational radiation
183: \citep{landau}
184: \begin{equation}
185: \frac{{\rm d}J_{\rm{GR}}}{{\rm d}t}=-\frac{32}{5}\frac{G^{7/2}}{c^5}
186: \frac{M_1^2M_2^2(M_1+M_2)^{1/2}}{a^{7/2}},
187: \end{equation}
188: where c is the light speed. This mechanism is important only in
189: very short period binary systems.
190:
191: The second angular momentum loss mechanism is for non-conservative
192: mass transfer. We assume that a fraction $\alpha$ of the transferred
193: mass is accreted by the NS, and the remaining mass is ejected out of
194: the binary as isotropic winds from the NS, carrying away the
195: specific angular momentum of the NS,
196: \begin{equation}
197: \frac{{\rm d}J_{\rm{ML}}}{{\rm
198: d}t}=-(1-\alpha)\dot{M}_2(\frac{q}{1+q})^2a^2\omega.
199: \end{equation}
200: In our numerical calculations we have set $\alpha=0$. Alternatively,
201: if the NS is spun up to be a millisecond pulsar, its radiation
202: pressure may be strong enough to halt the transferred matter at the
203: $L_1$ point and quench the accretion. This ``radio ejection'' may
204: cause almost all the matter from the secondary to be lost from the
205: binary \citep{burderi01,burderi02}. The corresponding rate of
206: angular momentum loss is
207: \begin{equation}
208: \frac{{\rm d}J_{{\rm ML}}}{{\rm d}t}=-\dot{M}_2 a_{L1}^2\omega
209: \end{equation}
210: where $a_{L1}$ is the distance from the $L_1$ point to the center
211: of mass of the binary system.
212:
213: The third angular momentum loss mechanism is MB. For a low-mass
214: MS star with deep convection zone, stellar winds which are magnetically
215: coupled with the star can decelerate the stellar spin efficiently, thus
216: carrying away the orbital angular momentum because of tidal synchronization.
217: The widely used formula for such MB effect was postulated by
218: \citet{verbunt81} and \citet{rappaport83} as
219: \begin{equation} \label{oldmb}
220: \frac{{\rm d}J_{\rm{MB}}}{{\rm d}t}=-3.8\times10^{-30}M_2R_2^4\omega^3 \,\rm{dyn\,cm}.
221: \end{equation}
222: However, observations of rapid rotators in young open clusters
223: suggest a modification of the MB law at high rotation rate
224: \citep{sills00},
225: \begin{eqnarray}\label{newmb}
226: \frac{{\rm d}J_{\rm{MB}}}{{\rm d}t} & = & -K \omega^3
227: \left(\frac{R_2}{R_{\sun}}\right)^{0.5}
228: \left(\frac{M_2}{M_{\sun}}\right)^{-0.5}, \hspace{20pt} \omega \leq \omega_{\rm{crit}}, \nonumber \\
229: \frac{{\rm d}J_{\rm{MB}}}{{\rm d}t} & = & -K \omega_{\rm{crit}}^2 \omega
230: \left(\frac{R_2}{R_{\sun}}\right)^{0.5}
231: \left(\frac{M_2}{M_{\sun}}\right)^{-0.5}, \ \omega >
232: \omega_{\rm{crit}},
233: \end{eqnarray}
234: where $K=2.7\times10^{47}$ gcm$^2$s \citep{andronov03},
235: $\omega_{\rm{crit}}$ is the critical angular velocity at which the
236: angular momentum loss rate reaches a saturated state, given by
237: \citep{krishnamurthi97}
238: \begin{equation}
239: \omega_{\rm{crit}}(t) = \omega_{\rm{crit}\sun} \frac{\tau_{\rm{t}_0,\sun}}{\tau_{\rm{t}}},
240: \end{equation}
241: where $\tau_{\rm{t}_0,\sun}$ and $\tau_{\rm{t}}$ are the global
242: turnover timescales for the convective envelope of the Sun at its
243: current age and of the secondary star at age $t$, respectively. They
244: can be calculated by integrating the inverse local convective
245: velocity over the surface convective envelope \citep{kim96}:
246: \begin{equation}
247: \tau_{\rm t}=\int_{R_b}^{R_2}\frac{{\rm d} r}{v},
248: \end{equation}
249: where $R_b$ is the radial distance from the center of the star to
250: the bottom of the surface convective envelope, and $v$ is the local
251: convective velocity from mixing-length theory \citep{bohm58}. Our
252: calculation gives $\tau_{\rm{t}_0,\sun}\simeq 28.4$ d, slightly
253: larger than $\tau_{\rm{t}_0,\sun}\simeq 13.8$ d in
254: \citet{sluys05b}, but consistent with the results of \citet{kim96}
255: and \citet{jung07}. See \citet[page 46]{eggleton06} for the discussion
256: of a possible reason for different values of $\tau_{\rm{t}_0,\sun}$ calculated.
257:
258: Following the suggestion of \citet{pod02}, we also add an ad hoc
259: factor
260: \[
261: \exp(-0.02/q_{\rm{conv}}+1) \; \rm{if}\;q_{\rm{conv}}<0.02,
262: \]
263: in Eqs.~(6) and (7), where $q_{\rm{conv}}$ is the mass fraction of
264: the surface convective envelop, to reduce the MB effect when the
265: convective envelope becomes too small.
266:
267: \subsection{Binary models}
268: To examine the influence of mass and angular momentum loss
269: mechanisms on the period evolution, we construct four models with
270: various mass and angular momentum loss combinations: (1) model 1 -
271: conservative mass transfer with traditional MB law (Eq.~[6]); (2)
272: model 2 - conservative mass transfer with saturated MB law
273: (Eq.~[7]); (3) model 3 - non-conservative mass transfer with mass
274: loss from $L_1$ point (Eq.~[5]) and saturated MB law (Eq.~[7]); and
275: (4) model 4 - non-conservative mass transfer with mass loss from the
276: NS (Eq.~[4]) and saturated MB law (Eq.~[7]). In all the four
277: models, the initial NS mass is set to be $M_{1, {\rm i}}=1.4M_\sun$,
278: and the initial mass of the secondary $M_{2,{\rm i}}$ ranges from
279: 0.5 to $2.0M_\sun$.
280:
281: \section{Numeric Results}
282: \subsection{The bifurcation periods}
283: Throughout this paper we define the bifurcation period $P_{\rm bif}$
284: as the initial binary orbital period $P_{\rm i}$ with a zero-age
285: main-sequence (ZAMS) companion star that separates converging from
286: diverging systems. We use $P_{\rm f}$ to denote the final orbital
287: period after the mass transfer. Another definition of the
288: bifurcation period used by \citet{pod02} is the orbital period when
289: the Roche lobe overflow just begins, which is expressed as $P_{\rm
290: rlof}$ in this paper.
291:
292: The results of the bifurcation periods for the four models described
293: in \S 2.3 are summarized in Fig.~\ref{bifur} and Table~\ref{tab1}.
294: We also draw the minimum initial period $P_{\rm ZAMS}$ that
295: corresponds to a lobe-filling ZAMS donor star in Fig.~\ref{bifur}.
296: Several features are noted for the bifurcation periods in
297: Fig.~\ref{bifur}. First, the bifurcation periods for all the four
298: models decrease with increasing initial secondary mass from
299: $0.5M_\sun$ to $1.3M_\sun$. Second, in models with saturated MB,
300: there exists an upper limit of the initial secondary mass, beyond
301: which no bifurcation period exists. This upper limit is in the range
302: $\sim 1.2-1.3\; M_\sun$ for model 3, and $\sim 1.3-1.4\; M_\sun$ for
303: models 2 and 4. Third, comparing the bifurcation periods of model 1
304: with those of models $2-4$ indicates that the MB law plays the most
305: important role in determining the values of the bifurcation periods:
306: different MB laws can change the bifurcation periods by as much as
307: $\sim 60\%$, compared to $\sim 14\%$ (see Table~\ref{tab1}) caused
308: by different mass loss mechanisms.
309:
310: In Table~\ref{tab1} we also present $P_{\rm rlof}$ following
311: \citet{pod02}. In model 1, we get $P_{\rm rlof}\simeq 18.3$ hr for
312: $M_{2,i}=1M_\sun$, which is in close line with the result of
313: \citet{pod02} ($17.7$ hr with $Z=0.001$, $Y=0.27$), where the difference
314: could be explained as the difference between the metallicities we used.
315: When we use saturated MB, $P_{\rm rlof}$ decreases to $\sim 11$ hr.
316:
317: According to the calculated orbital period evolutions, LMXBs can be
318: classified into three categories: the diverging systems with $P_{\rm
319: f} \gg P_{\rm i}$, %%(where the subscripts i and f denote the initial
320: %and final evolutionary phases),
321: the converging systems with $P_{\rm f} \ll P_{\rm i}$, and the
322: parallel systems with $P_{\rm f} \sim P_{\rm i}$. As an example, we
323: present the calculated results for a $1.4M_\sun+1.0M_\sun$ binary in
324: model 3, to illustrate the three kinds of evolutionary sequences in
325: Fig.~\ref{tp}. The corresponding bifurcation period is found to be
326: $1.25$ day, and the initial orbital periods are chosen to be $P_{\rm
327: i}=1.20$, $1.25$, and $1.40$ days, which represent the converging,
328: parallel, and diverging systems respectively.
329: %%In Figs.~\ref{tp}$-$\ref{MT} we show the evolution of
330: %the orbital periods with time, the evolution of the secondary masses
331: %with the period, the evolutionary tracks of the secondary in the H-R
332: %diagram, and the evolution of the mass transfer rates with time,
333: %respectively. From the H-R diagram we see that the secondary star in
334: %the diverging and parallel systems ends nearly at the same place,
335: %but considerably more luminous than in the converging systems.
336:
337: %To examine the effect of various angular momentum loss mechanisms,
338: %in Fig.~\ref{AT} we compare the angular momentum loss timescales due
339: %to MB $\tau_{\rm MB} = -({\rm d}\ln J/{\rm d}t)_{\rm MB}^{-1}$, mass
340: %loss $\tau_{\rm ML} = -({\rm d}\ln J/{\rm d}t)_{\rm ML}^{-1}$, and
341: %gravitational radiation $\tau_{\rm GR} = -({\rm d}\ln J/{\rm
342: %d}t)_{\rm GR}^{-1}$ as a function of the donor mass. The shorter
343: %angular momentum loss timescale corresponds to more efficient
344: %driving of mass transfer. It is seen that MB dominates the angular
345: %momentum loss of the converging system ($P_{\rm i}=1.20$ day), while
346: %mass loss becomes more important in most of the evolutionary phase
347: %of the diverging system ($P_{\rm i}=1.40$ day).
348:
349: \subsection{Effect of MB and mass loss}
350: \citet{pylyser88} have emphasized the effect of MB on the evolution
351: of LMXBs. Comparing the results of models 1 and 2 presented in
352: Table~\ref{tab1}, we find that the bifurcation periods with
353: traditional MB law are smaller (larger) than those with saturated MB
354: law, when the initial secondary star mass $M_{\rm 2,i}$ is less
355: (larger)
356: than $0.7M_\sun$. %To illustrate the effect of different MB in more
357: %detail, we present the evolutionary tracks with both traditional and
358: %saturated MB for binary systems with $M_{2,\rm i}=0.5 M_\sun$ in
359: %Fig.~\ref{MBa}, and with $M_{2,\rm i}= 1M_\sun$ days in
360: %Fig.~\ref{MBb}, respectively. Although the initial orbital periods
361: %are the same, different MB laws lead to completely different period
362: %evolution.
363:
364: %This is because that the saturated magnetic braking law is weaker than the old magnetic
365: %braking law for star mass more than $0.7M_\sun$. As a consequence of this, for $M_{2,\rm i}>
366: %1.3M_\sun $, the saturated magnetic braking law is too weak to draw the binary system together
367: %to form a converging system, while the old magnetic braking law could (see Tab.~\ref{tab1}).
368:
369:
370: %\subsection{Effect of mass loss}
371: Our results suggest that mass loss also influences the value of
372: $P_{\rm bif}$, though in an less important way compared with MB. The
373: bifurcation periods in non-conservative models 3 and 4 are lower than
374: those in model 2, in which conservative mass transfer has been
375: assumed. This result is consistent with \citet{sluys05b}, but
376: contradicted with \citet{ergma98}.
377: %To see this more clearly, we
378: %removed MB and calculated the evolutionary sequences for an LMXB
379: %with $M_{2,\rm i} = 1.0 M_\sun$ and $P_{\rm i}=0.4$, 0.5, 0.6 day in
380: %models 3 and 4, respectively. The results are shown in
381: %Fig.~\ref{massloss}. We see that the diverging system in model 3
382: %($P_{\rm i}=0.5$ day) evolves to converge in model 4. This is
383: %because the mass loss from the NS surface can take more specific
384: %angular momentum away from the binary system than from the $L_1$
385: %point, thus shrinking the orbit more efficiently.
386:
387: It is also interesting to see whether an UCXB can form with
388: saturated MB. For an LMXB with an initial orbital period below the
389: bifurcation period, mass transfer is mainly driven by the loss of
390: angular momentum. The orbital period will decrease with the donor
391: mass until a minimum period is reached. \citet{paczynski81} found a
392: minimum period about $80$ min without MB, while \citet{pod02} showed
393: that minimum orbital periods less than $11$ min could be reached for
394: binaries with an initial orbital period very close to the bifurcation
395: period if traditional MB is included, but in a time longer than the age of the
396: universe. \citet{sluys05b} further investigated this ``magnetic capture"
397: scenario for the formation of UCXBs. Our calculations show that when the
398: initial orbital period is close to the bifurcation period, ultra-compact
399: systems ($P<1$ h) can indeed form with saturated MB, but also in a time
400: longer than the age of the universe. For example, for an LMXB with
401: $M_{2,\rm i}=1.3M_\sun$ and $P_{\rm i}=0.46$ day in model 4, a final
402: period of $P_{\rm f}=22$ min can be reached after $\ga 15$ Gyr of
403: mass transfer. All the works done by previous authors show that a more
404: efficient angular momentum loss mechanism is required to produce UCXBs
405: within $13.7$ Gyr in this scenario.
406: % An example of such evolution is shown in Fig.~\ref{UCXB}
407: %for an LMXB with $M_{2,\rm i}=1.3M_\sun$ and $P_{\rm i}=0.46$ day in
408: %model 4. A final period of $P_{\rm f}=22$ min can be reached after
409: %$\ga 15$ Gyr of mass transfer. Obviously a more efficient angular
410: %momentum loss mechanism is required to produce UCXBs.
411:
412: \subsection{Semi-analytical Method}
413: In this subsection we will try to use a semi-analytical method to
414: understand our numerical results. First from Eq.~(2), we have the
415: following equation
416: \begin{equation}
417: \frac{3}{2}\frac{\dot{J}}{J}=\frac{3}{2}(\frac{\dot{M_1}}{M_1}+\frac{\dot{M_2}}{M_2})-
418: \frac{1}{2}\frac{\dot{M_1}+\dot{M_2}}{M_1+M_2}+\frac{1}{2}\frac{\dot{P}}{P}.
419: \end{equation}
420: If we assume a fraction $\alpha$ of the mass lost by the donor is
421: accreted by the NS, i.e., $\dot{M_1}=-\alpha\dot{M_2}$, we can write
422: the period derivative as
423: \begin{equation} \label{pdot}
424: \frac{\dot{P}}{P}=3\frac{\dot{J}}{J}-A(M_1,M_2,\alpha)\frac{\dot{M_2}}{M_2},
425: \end{equation}
426: where
427: \begin{equation}
428: A(M_1,M_2,\alpha)=\frac{3M_1^2+2(1-\alpha)M_1M_2-3\alpha
429: M_2^2}{M_1(M_1+M_2)}.
430: \end{equation}
431: Our analysis is limited to binary evolution with $M_2 < M_1$. In
432: this case it is clearly seen that mass transfer increases the
433: orbital period and angular momentum loss decreases the orbital
434: period. The bifurcation period is decided by the balance of these
435: two factors.
436:
437: Keeping the orbital period unchanged (i.e. $\dot{P}\simeq 0$), we
438: calculate the maximum mass transfer rates for orbital periods from
439: $0.35$ day to $0.95$ day. This period interval covers the whole
440: range of the bifurcation periods obtained in this work and in
441: \citet{pod02}. We show the calculated mass transfer rates in
442: Fig.~\ref{mdot2}, and find that they can be fitted by an approximate
443: expression as
444: \begin{equation} \label{mdot}
445: -\dot{M_2}(P)\simeq 2.73\times 10^{-8} (P/{\rm day})^{6.41(\pm
446: 0.11)}M_\sun \rm{yr}^{-1}.
447: \end{equation}
448: Then we calculate the mean mass transfer rates with constant $P$ and
449: find that they lie between $\dot{M}_2(P)/2$ and $\dot{M}_2(P)$. This
450: means that if we use Eq.~(\ref{mdot}) to calculate $P_{\rm rlof}$, it
451: will deviate no more than $\sim 10\%$ from the true value. The mean mass
452: transfer rates here are calculated as follows. We fix the binary period $P$ in
453: a constant value in our code, and then evolve the donor from its initial mass
454: $M_{\rm 2,i}$ to the time when it loses half of its initial mass $0.5M_{2,i}$.
455: This mass transfer process takes a time of $T_{\rm 1/2}$. Then we use
456: $0.5M_{\rm2,i}/T_{1/2}$ as the mean $\dot{M}_2$ for this period $P$.
457: If we assume that angular momentum loss is dominated by saturated MB when
458: $P<10$ d, from Eqs.~(2) and (\ref{newmb}) we have
459: \begin{eqnarray}
460: \frac{\dot{J}}{J_{\rm orb}} & = & -K \omega_{\rm crit}^2
461: (\frac{R_2}{R_{\sun}})^{0.5}(\frac{M_2}{M_{\sun}})^{-0.5}
462: (\frac{4\pi^2}{G})^{2/3}\frac{(M_1+M_2)^{1/3}}{M_1 M_2}P^{-4/3} \nonumber \\
463: &\simeq & -6.2\times 10^{-11}(P/{\rm day})^{-1} {\rm yr}^{-1}.
464: \end{eqnarray}
465: Here we adopt $K=2.7\times10^{47}$ gcm$^2$s \citep{andronov03},
466: $\omega_{\rm crit}=2.9\times 10^{-5}$ Hz \citep{sills00},
467: $M_1=1.4M_\sun$, $M_2=1M_\sun$, and replace the radius of the
468: secondary $R_2 $ with its Roche lobe radius $R_{\rm L,2}$ from
469: Eq.~(1). Combining Eqs.~(\ref{pdot})-(14) with $\dot{P}=0$ we obtain
470: $P_{\rm rlof}\simeq 12.4$ hr for conservative mass transfer
471: ($\alpha=1$), and $P_{\rm rlof} \simeq 10.7$ hr for non-conservative
472: mass transfer ($\alpha=0$). These values agree well with our
473: numerical results ($\sim 10.8-11.1$ hr). If the traditional MB law
474: is used, similarly, from Eq.~(\ref{oldmb}) we get
475: \begin{eqnarray}
476: \frac{\dot{J}}{J_{\rm orb}} & = & -3.8\times 10^{-30}M_2R_2^4\omega^3
477: G^{-2/3}\frac{(M_1+M_2)^{1/3}}{M_1 M_2}\omega^{1/3} \nonumber \\
478: &\simeq & -4.6\times 10^{-9}(P/{\rm day})^{-2/3} {\rm yr}^{-1},
479: \end{eqnarray}
480: for $M_1=1.4M_\sun$ and $M_2=1M_\sun$. Combining Eqs.~(\ref{pdot})-(13), (15)
481: with $\dot{P}=0$ we get $P_{\rm rlof}\simeq22.3$ hr for conservative mass
482: transfer ($\alpha=1$), which is about $20\%$ larger than $\sim 18.3$ hr
483: from our numerical calculations and $\sim 17.7$ hr in \citet{pod02}.
484: The main reason for this difference is that we use the constant value
485: $1.4M_\sun$, $1M_\sun$ for $M_1$, $M_2$ in Eq.~(12), which should change
486: with time to $\sim 2.2M_\sun$, $\sim 0.2M_\sun$. This will decrease
487: the coefficient in Eq.~(12) and increase the value of $P_{\rm rlof}$ by
488: $\sim 10-20\%$. For donor mass $\ge 1.2M_\sun$, this will increase
489: the value of $P_{\rm rlof}$ by as much as $\sim 30\%$. So it is better to
490: use $\alpha=0$ instead of $\alpha=1$ for donors mass $\ge 1.2M_\sun$, which could yield
491: more accurate results (from our numerical results we find that the deviation
492: of $P_{\rm rlof}$ between conservative and non-conservative mass transfer is
493: smaller than $10\%$).
494: %The ad hoc
495: %term we add in Eq.~(6) and Eq.~(7) is will also increase the value
496: %of the semi-analytical result, but its degree of influence is too
497: %hard to estimate.
498:
499: Using Eq.~(11)-(15), we also compute the semi-analytical results of
500: $P_{\rm rlof}$ for $0.5-2M_\sun$ donors, and compare them with our
501: numerical results of models 1 and 4 in Fig.~4, where the semi 4 results
502: are calculated with $\alpha=0$. When we calculate the semi 1 results in
503: Fig.~4, for the reasons mentioned above and below, we use $\alpha=1$ for
504: $M_{\rm 2,i} \le 1.1M_\sun$ and $\alpha=0$ for $M_{\rm 2,i} \ge 1.2M_\sun$,
505: where $\alpha=1$ should be used. A few points need to be noted for the
506: semi-analytical results in Fig.~4. Firstly with the above mentioned equations it
507: is impossible to compute the $P_{\rm rlof}$ for binaries with $M_{\rm 2,i}\ge 1.4M_\sun$
508: under conservative mass transfer, since both terms in the right side of
509: Eq.~(11) are negative when $\alpha=1$ and $M_{\rm 2,i}\ge1.4\,M_\sun$,
510: and there will be no solutions for $\dot{P}=0$. We instead adopt $\alpha=0$
511: when $M_{\rm 2,i}>1.4M_\sun$ (from our numerical results we find that the
512: deviation of $P_{\rm rlof}$ between conservative and non-conservative mass
513: transfer is smaller than $10\%$). Secondly, Eq.~(13) is derived only for
514: $1M_\sun$ donor star rather donors in the whole mass range ($0.5-2\,M_\sun$),
515: because in the latter case it is impossible to find a unified
516: expression of the mass transfer rate like Eq.~(13). As seen in
517: Fig.~4, the difference between the semi-analytical and numerical
518: results is generally smaller than $20\%$ except for donors smaller
519: than $0.7M_\sun$. The reasons for the big discrepancies when
520: $M_{2,i}<0.7M_\sun$ are discussed in \S 4.
521: %From the above derivation, we could see
522: %that the difference of bifurcation period between our result and
523: %that of \citet{pod02} is from the difference of the strength of MB
524: %laws used.
525:
526: \section{Discussion and Conclusions}
527:
528: Motivated by new ideas about MB and mass loss in LMXB evolution, we
529: have made a systematic investigation on the bifurcation periods in
530: binary models, taking into account different MB laws and mass loss
531: mechanisms. We find that the strength of MB is the dominant factor
532: in determining the value of bifurcation periods compared with mass
533: loss. The stronger MB, the larger the bifurcation periods. This also
534: results in an upper limit for the secondary masses beyond which no
535: converging systems exist.
536:
537:
538: %When the mass loss is at the bifurcation period
539: %according to the definition of \citet{pod02}, the period should
540: %remain nearly unchanged during the mass transfer. So we could use
541: %the relation to compute the equilibrium period $P_{\rm eq}$ which
542: %should be close to $P_{\rm bif}$. Here A is order of 1. The angular
543: %momentum loss $\dot{J}$ is mainly from the MB, which will increase
544: %as the period decrease. When period is unchanged, $\dot{m_2}$ is
545: %from the donor expansion.
546:
547: %The first term in Eqs.~\ref{pdot} is negative, which will increase as the period decrease, and the second
548: %term is positive, which will decrease as the period decrease. So there could only be one period solution
549: %for this equation. The if we take Eqs.~\ref{newmb} and Eqs.~\ref{mdot} into Eqs.~\ref{pdot}, we could
550: %compute the $P_{\rm bif}$ for $m_2=1M_\sun$. If the first term is from the old magnetic braking law, it
551: %is stronger than the saturated magnetic braking law, so $P_{\rm bif}$ will be larger than.
552:
553:
554: In our calculations we assume either fully conservative (models 1
555: and 2) or non-conservative (models 3 and 4) mass transfer to
556: constrain the bifurcation period distribution in different mass
557: transfer modes. From the expression of $A(M_1,M_2,\alpha)$ we always
558: have
559: \begin{equation}
560: A(M_1,M_2,1)=3-\frac{3M_2}{M_1} <A(M_1,M_2,0)=
561: 3-\frac{M_2}{M_1+M_2},
562: \end{equation}
563: which means that non-conservative mass transfer contributes more to
564: the increase of the orbital period than conservative mass transfer.
565: This explains why we generally have a lower bifurcation period in
566: non-conservative mass transfer models (models 3 and 4) than in
567: conservative mass transfer model (model 2) under the same MB law.
568: The real situation may lie between these two extreme cases. For
569: binary systems with donors $M_{\rm 2,i}\sim 0.5-0.8 M_\sun$, it
570: would take more that 13.7 Gyr before mass transfer begins via Roche
571: lobe overflow. So the bifurcation period for these system seems
572: meaningless, unless there exist some unknown mechanisms of loss of
573: orbital angular momentum.
574:
575: In our semi-analytical analysis in \S 3.3 we use the condition
576: $\dot{P}\sim 0$ to derive the values of $P_{\rm rlof}$. This
577: expression seems different from $P_{\rm f}\simeq P_{\rm rlof}$, which
578: is the original definition of bifurcation period.
579: %We argue here that
580: %these two expressions are roughly the same except for two cases. The
581: %first is for binaries with $M_{\rm 2,i}\ge1.4 M_\sun$ under
582: %conservative mass transfer ($\alpha=1$), which has been discussed in
583: %\S 3.3. The second is for $M_{\rm 2,i}<0.8M_\sun$. For $0.8M_\sun <
584: %M_{\rm 2,i}< 1.4 M_\sun$, we find that $\dot{P}$ always scales with
585: %$P$ from Eq.~(11) and (13)-(15). This means that if initially
586: %$\dot{P}>0$ ($<0$), $\dot{P}$ will become larger (smaller) during
587: %the evolution, leading to monotonic increase (decrease) of the
588: %period, as seen in Fig.~\ref{tp}.
589: %So for these systems $P_{\rm f} \sim P_{\rm rlof}$ is approximately
590: %equivalent with $\dot{P} \sim 0$. When it comes to the binary systmes
591: %with $M_{\rm 2,i}<0.8M_\sun$, several main reasons account for this
592: %big discrepancy between the semi-analytical result and numerical result.
593: %s the original definition of bifurcation period.
594: We argue here that these two expressions are roughly the same except for binaries
595: with $M_{\rm 2,i}\ge1.4 M_\sun$ under conservative mass transfer ($\alpha=1$),
596: the reason of which has been given in \S 3.3. For $M_{\rm 2,i}< 1.4 M_\sun$,
597: we find that $\dot{P}/P$ always scales with $P$ from Eq.~(11) and (13)-(15).
598: This means that if initially $\dot{P}>0$ ($<0$), $\dot{P}/P$ will become
599: larger (smaller) during the evolution, leading to monotonic increase
600: (decrease) of the period, as seen in Fig.~\ref{tp}. So for these systems
601: $P_{\rm f} \sim P_{\rm rlof}$ is approximately equivalent with $\dot{P} \sim 0$.
602: Several rough assumptions in this semi-analytical method contribute to the
603: discrepancies between the semi-analytical results and the numerical results in Fig.~4, especially
604: for $M_{\rm 2,i}<0.7M_\sun$. First is the use of $\dot{P}\sim0$ as the
605: definition of $P_{\rm rlof}$, which may not work well sometimes. Second is the
606: use of Eq.~(13), which is most suitable for binaries with $M_{2,i}=1M_\sun$ as
607: pointed out in \S 3.3. Third is the assumption we made that magnetic braking
608: law is the dominated mechanism for the angular momentum loss, while the true
609: case is that the MB may not work sometimes (for example when the convective envelop
610: is too small). Fourth is that we use a constant initial value of $M_{\rm 1,i}$,
611: $M_{\rm 2,i}$ for $M_1$, $M_2$ in Eq.~(12) and Eq.(14)-(15), while in the true case
612: $M_1$, $M_2$ should change with time. This will cause a big problem for $\alpha=1$
613: when $M_{\rm 2,i} > 1M_\sun$, which has been pointed out in \S 3.3. Fifth reason is
614: the use of $\omega_{\rm crit,\sun}$ in Eq.~(14) as the value of $\omega_{\rm crit}$
615: for all the donors ranging from $0.5M_\sun$ to $1.3M_\sun$. At last we conclude
616: that (1)$\dot{P}$ is a fair definition of bifurcation period, and (2)the period
617: evolution during the mass transfer phase is in first approximation sufficiently
618: well described by the balance of mass transfer and angular momentum loss caused
619: by MB. For these rough assumptions made in this semi-analytical method, its
620: results agree with the numerical results only in the leading order.
621: %When $M_{\rm 2,i}<0.8M_\sun$, the
622: %decrease of the orbital period before the RLOF will take too long a
623: %time. This accounts for the discrepancy between the semi-analytical
624: %and numerical results when $M_{\rm 2,i}<0.8M_\sun$ in Fig.~4.
625:
626: %%So if initially we have approximately $\dot{P}\sim0$, we would finally have $P_{\rm f} \sim P_{\rm i}$.
627:
628: Our numerical calculations show that there is an upper limit for the
629: donor mass beyond which no converging systems will form.
630: \citet{pylyser88} found that, in the case of $M_{1, \rm i} =
631: 4.0M_\sun$, there is no converging system existing if $M_{2,\rm i}>
632: 1.7 M_\sun $, and concluded that for any given initial accretor mass
633: there exists a maximum initial secondary mass for the formation of
634: converging systems. From our calculations with $M_{1,\rm
635: i}=1.4M_\sun$, we find an upper limit for the initial secondary mass
636: $M_{2,\rm i}$ between $1.2$ and $1.4 M_\sun$ under saturated MB.
637: The reason is that for binaries with a MS donor of initial mass
638: $> 1.4M_\sun$, the bifurcation period is shorter than the minimum ZAMS
639: period, so that these systems will diverge.
640: %The reason is that, for a MS star of mass $>1.4M_\sun$, the initial
641: %orbital period has to be larger than $\sim 12$ hr, otherwise the
642: %star cannot fit into its Roche lobe. On the other hand, the above
643: %analysis shows that $\sim 12$ hr is the equilibrium period at which
644: %mass transfer balances angular momentum loss for donor stars around
645: %$1M_\sun$. Hence the binary system should have an initial period
646: %above this period, causing it to evolve to diverge.
647: For traditional MB, this upper limit is $> 2M_\sun$, beyond the range of donor
648: masses we adopt.
649:
650: %\clearpage
651:
652: %\begin{equation} \n
653: %\frac{\dot{J}}{J_{\rm orb}} = -K \omega_{\rm crit}^2
654: %(\frac{R_2}{R_{\sun}})^{0.5}(\frac{M_2}{M_{\sun}})^{-0.5}
655: %(\frac{4\pi^2}{G})^{2/3}\frac{(M_1+M_2)^{1/3}}{M_1 M_2}P^{-4/3}
656: %\end{equation}
657:
658: %\begin{eqnarray}
659: % -K \omega_{\rm crit}^2 & = & [2.7\times 10^{47} \times (2.9\times 10^{-5})^2] = 2.27\times10^{38} \nonumber \\
660: %(\frac{R_2}{R_{\sun}})^{0.5} & = & (\frac{R_{\rm L,2}}{R_{\sun}})^{0.5} =
661: %(\frac{\frac{0.49q^{-2/3}}{0.6q^{-2/3}+\ln(1+q^{-1/3})}a}{R_\sun})^{0.5}=
662: %(\frac{0.408a}{R_\sun})^{0.5} \nonumber \\
663: %&=& (\frac{0.408*(\frac{G(M_1+M_2)}{\omega^2})^{1/3}}{R_\sun})^{0.5} = 3.15\times10^{-2}P^{1/3} \nonumber \\
664: %(\frac{M_2}{M_{\sun}})^{-0.5} &=& 1 \nonumber \\
665: %(\frac{4\pi^2}{G})^{2/3} &=& 7.05\times10^5 \nonumber \\
666: %\frac{(M_1+M_2)^{1/3}}{M_1 M_2}P^{-4/3} &=& 3.01\times10^{-56}P^{-4/3} \nonumber
667: %\end{eqnarray}
668:
669: %\begin{eqnarray}
670: %\frac{\dot{J}}{J_{\rm orb}} & = & 2.27\times10^{38}\times 3.15\times10^{-2}P^{1/3}\times 7.05\times10^5\times 3.01
671: %\times10^{-56}P^{-4/3} \nonumber \\
672: %&=& 1.66\times 10^{-13} (P/s)^{-1} s^{-1} \nonumber \\
673: %&=& 1.66 \times 10^{-13} \times 365 }\times (P/{\rm day})^{-1} {\rm yr}^{-1}= -6.06\times 10^{-11}(P/{\rm day})^{-1} {\rm yr}^{-1}
674: %\end{eqnarray}
675: %Here we adopt $K=2.7\times10^{47}$ gcm$^2$s \citep{andronov03},
676: %$\omega_{\rm crit}=2.9\times 10^{-5}$ Hz \citep{sills00}, $M_1=1.4M_\sun$,
677: %$M_2=1M_\sun$, $a$ is the orbital separation, and replace the radius of the secondary $R_2 $with
678: %its Roche lobe radius $R_{\rm L,2}$.
679:
680: \acknowledgments We are grateful to an anonymous referee for helpful
681: comments. This work was supported by the Natural Science Foundation
682: of China under grant numbers 10573010 and 10221001. B.M. thank P.
683: P. Eggleton, Xuefei Chen, and Xiaojie Xu for their help during this
684: work.
685:
686:
687:
688: \begin{thebibliography}{}
689: \bibitem[Alexander \& Ferguson(1994)]{alexander94} Alexander, D.~R., \& Ferguson, J.~W.\ 1994, \apj, 437, 879
690: %\bibitem[Alpar et al.(1982)]{alp82} Alpar, M. A., Cheng, A. F., Ruderman, M. A., \& Shaham, J. 1982,
691: %Nat, 300, 728f
692: \bibitem[Andronov et al.(2003)]{andronov03} Andronov, N., Pinsonneault, M., \& Sills, A.\ 2003, \apj, 582, 358
693: %\bibitem[Barnes \& Sofia(1996)]{barnes96} Barnes, S., \& Sofia, S.\ 1996, \apj, 462, 746
694: \bibitem[Bassa et al.(2006)]{bassa06} Bassa, C. G., van Kerkwijk, M. H., Koester, D. \& Verbunt, F. 2006,
695: \aap 456, 295
696: %\bibitem[Bhattacharya \& van den Heuvel(1991)]{bha91}Bhattacharya, D. \& van den Heuvel, E. P. J. 1991, Phys. Rep., 203,
697: %1
698: \bibitem[B{\"o}hm-Vitense(1958)]{bohm58} B{\"o}hm-Vitense, E.\ 1958, Zeitschrift fur Astrophysik, 46, 108
699: \bibitem[Burderi et al.(2001)]{burderi01} Burderi, L., et al.\ 2001, \apjl, 560, L71
700: \bibitem[Burderi et al.(2002)]{burderi02} Burderi, L., D'Antona, F., \& Burgay, M.\ 2002, \apj, 574, 325
701: \bibitem[Chou \& Grindlay(2001)]{chou01} Chou, Y., \& Grindlay, J.~E.\ 2001, \apj, 563, 934
702: \bibitem[Clark et al.(1975)]{clark75} Clark, G.~W., Markert, T.~H., \& Li, F.~K.\ 1975, \apjl, 199, L93
703: \bibitem[D'Antona et al.(2006)]{antona06} D'Antona, F., Ventura, P.,Burderi, L., Di Salvo, T., Lavagetto, G.,
704: Possenti, A. \& Teodorescu, A. 2006, \apj, 640, 950
705: \bibitem[Eggleton(1971)]{eggleton71} Eggleton, P.~P.\ 1971, \mnras, 151, 351
706: \bibitem[Eggleton(1972)]{eggleton72} Eggleton, P.~P.\ 1972, \mnras, 156, 361
707: \bibitem[Eggleton(1983)]{eggleton83} Eggleton, P.~P.\ 1983, \apj, 268, 368
708: \bibitem[Eggleton(2006)]{eggleton06} Eggleton, P.\ 2006, Evolutionary Processes in Binary and Multiple Stars, by Peter Eggleton,
709: pp.~.~ISBN 0521855578.~Cambridge, UK: Cambridge University Press, 2006.,
710: \bibitem[Ergma \& Sarna(1996)]{ergma96} Ergma, E., \& Sarna, M.~J.\ 1996, \mnras, 280, 1000
711: \bibitem[Ergma et al.(1998)]{ergma98} Ergma, E., Sarna, M.~J., \& Antipova, J.\ 1998, \mnras, 300, 352
712: \bibitem[Frank, King, \& Raine(2002)]{frank02} Frank, J., King, A. R., \& Raine, D. 2002, Accretion Power in
713: Astrophysics (Cambridge: Cambridge Univ. Press)
714: \bibitem[Freire et al.(2008)]{freire08} Freire, P. C. C., Wolszczan, A., van den Berg, M., \& Hessels, J. W.
715: T. 2008, \apj, 679, 1433
716: \bibitem[Jung \& Kim(2007)]{jung07} Jung, Y.~K., \& Kim, Y.-C.\ 2007, Journal of Astronomy and Space Sciences, 24, 1
717: \bibitem[Han et al.(1994)]{han94} Han, Z., Podsiadlowski, P., \& Eggleton, P.~P.\ 1994, \mnras, 270, 121
718: \bibitem[Hubbard \& Lampe(1969)]{hubbard69} Hubbard, W.~B., \& Lampe, M.\ 1969, \apjs, 18, 297
719: %\bibitem[Kalogera \& Baym(1996)]{kalogera96} Kalogera, V., \& Baym, G.\ 1996, \apjl, 470, L61
720: \bibitem[Kim \& Demarque(1996)]{kim96} Kim, Y.-C., \& Demarque, P.\ 1996, \apj, 457, 340
721: \bibitem[Krishnamurthi et al.(1997)]{krishnamurthi97} Krishnamurthi, A., Pinsonneault, M.~H., Barnes, S.,
722: \& Sofia, S.\ 1997, \apj, 480, 303
723: \bibitem[Landau \& Lifshitz(1975)]{landau} Landau, L.~D., \& Lifshitz, E.~M.\ 1975, Course of theoretical physics -
724: Pergamon International Library of Science, Technology, Engineering and Social Studies, Oxford: Pergamon Press,
725: 1975, 4th rev.engl.ed.,
726: %\bibitem[Liu et al.(2007)]{liu07} Liu, Q.~Z., van Paradijs, J., \& van den Heuvel, E.~P.~J.\ 2007, \aap, 469, 807
727: \bibitem[Lombardi et al.(2006)]{lombard06} Lombardi, J.~C., Jr., Proulx, Z.~F., Dooley, K.~L.,
728: Theriault, E.~M., Ivanova, N., \& Rasio, F.~A.\ 2006, \apj, 640, 441
729: \bibitem[Nelson et al.(1986)]{nelson86} Nelson, L.~A., Rappaport, S.~A., \& Joss, P.~C.\ 1986, \apj, 311, 226
730: %\bibitem[Nice et al.(2005)]{nice05} Nice, D.~J., Splaver, E.~M., Stairs, I.~H., L{\"o}hmer, O., Jessner, A., Kramer,
731: %M., \& Cordes, J.~M.\ 2005, \apj, 634, 124
732: \bibitem[Paczynski \& Sienkiewicz(1981)]{paczynski81} Paczynski, B., \& Sienkiewicz, R.\ 1981, \apjl, 248, L27
733: \bibitem[Podsiadlowski et al.(2002)]{pod02} Podsiadlowski, P., Rappaport, S., \& Pfahl, E.~D.\ 2002, \apj, 565, 1107
734: \bibitem[Pols et al.(1995)]{pols95} Pols, O.~R., Tout, C.~A., Eggleton, P.~P., \& Han, Z.\ 1995, \mnras, 274, 964
735: \bibitem[Pylyser \& Savonije(1988)]{pylyser88} Pylyser, E., \& Savonije, G.~J.\ 1988, \aap, 191, 57
736: \bibitem[Pylyser \& Savonije(1989)]{pylyser89} Pylyser, E.~H.~P., \& Savonije, G.~J.\ 1989, \aap, 208, 52
737: \bibitem[Rappaport et al.(1983)]{rappaport83} Rappaport, S., Verbunt, F., \& Joss, P.~C.\ 1983, \apj, 275, 713
738: \bibitem[Rasio et al.(2000)]{rasio00} Rasio, F.~A., Pfahl, E.~D., \& Rappaport, S.\ 2000, \apjl, 532, L47
739: %\bibitem[Ritter \& Kolb(2003)]{ritter03} Ritter, H., \& Kolb, U.\ 2003, \aap, 404, 301
740: \bibitem[Rogers \& Iglesias(1992)]{rogers92} Rogers, F.~J., \& Iglesias, C.~A.\ 1992, \apjs, 79, 507
741: \bibitem[Ruderman et al.(1989)]{ruderman89} Ruderman, M., Shaham, J., \& Tavani, M.\ 1989, \apj, 336, 507
742: \bibitem[Sills et al.(2000)]{sills00} Sills, A., Pinsonneault, M.~H., \& Terndrup, D.~M.\ 2000, \apj, 534, 335
743: %\bibitem[Spruit \& Ritter(1983)]{spruit83} Spruit, H.~C., \& Ritter, H.\ 1983, \aap, 124, 267
744: \bibitem[Steeghs \& Jonker(2007)]{steeghs07} Steeghs, D. \& Jonker, P. G. 2007, \apj, 669, L85
745: \bibitem[Timmes et al.(1996)]{timmes96} Timmes, F. X., Woosley, S. E., \& Weaver, T. A. 1996, \apj, 457, 834
746: \bibitem[Tutukov et al.(1985)]{tutukov85} Tutukov, A.~V., Fedorova, A.~V., Ergma, E.~V., \& Yungelson, L.~R.\ 1985,
747: Soviet Astronomy Letters, 11, 52
748: \bibitem[Tutukov et al.(1987)]{tutukov87} Tutukov, A.~V., Fedorova, A.~V., Ergma, E.~V., \& Yungelson, L.~R.\ 1987,
749: Soviet Astronomy Letters, 13, 328
750: %\bibitem[van der Klis(2006)]{kli06} van der Klis, M. 2006, in Compact Stellar X-Ray Sources, ed. W. H.
751: %G. Lewin, \& M. van den Klis (Cambridge: Cambridge Univ. Press), 39
752: \bibitem[van der Klis et al.(1993)]{klis93}
753: van der Klis, M., Hasinger, G., Verbunt, F., van Paradijs, J., Belloni, T., \& Lewin, W.~H.~G.\ 1993, \aap, 279, L21
754: \bibitem[van der Sluys et al.(2005a)]{sluys05a} van der Sluys, M.~V., Verbunt, F., \& Pols, O.~R.\ 2005a, \aap, 431, 647
755: \bibitem[van der Sluys et al.(2005b)]{sluys05b} van der Sluys, M.~V., Verbunt, F., \& Pols, O.~R.\ 2005b, \aap, 440, 973
756: \bibitem[Verbunt \& Zwaan(1981)]{verbunt81} Verbunt, F., \& Zwaan, C.\ 1981, \aap, 100, L7
757:
758:
759:
760:
761: \bibitem[Webbink et al.(1983)]{webbink83} Webbink, R.~F.,
762: Rappaport, S., \& Savonije, G.~J.\ 1983, \apj, 270, 678
763:
764: \end{thebibliography}
765:
766: %\clearpage
767:
768: \clearpage
769:
770: \begin{deluxetable}{rrrrrrrrrr}
771: \tablecolumns{10} \tablewidth{0pc} \tablecaption{Calculated results
772: of the bifurcation periods for different binary models \label{tab1}}
773: \tablehead{ \colhead{} & \multicolumn{4}{c}{$P_{\rm bif}$ (day)}
774: & \colhead{} &
775: \multicolumn{4}{c}{$P_{\rm rlof}$ (hr)} \\
776: \cline{2-5} \cline{7-10} \\
777: \colhead{$M_{2,\rm{i}}$ } & \colhead{Model 1} & \colhead{ Model 2} & \colhead{ Model 3} &
778: \colhead{ Model 4} &\colhead{} & \colhead{ Model 1} & \colhead{ Model 2} & \colhead{ Model 3} &
779: \colhead{ Model 4}
780: }
781: \startdata
782: 0.5$M_\sun$ &3.20 &4.08 &4.12 &4.01 & & 31.8 &16.5 &18.2 &13.6 \\
783: 0.6$M_\sun$ &3.02 &3.41 &3.40 &3.35 & & 25.6 &11.3 &11.1 &10.0 \\
784: 0.7$M_\sun$ &2.86 &2.81 &2.79 &2.77 & & 21.5 &10.0 &9.8 &9.6 \\
785: 0.8$M_\sun$ &2.77 &2.41 &2.38 &2.37 & & 19.0 &10.1 &9.9 &9.8 \\
786: 0.9$M_\sun$ &2.73 &2.14 &2.09 &2.10 & & 17.9 &10.7 &10.3 &10.4 \\
787: 1.0$M_\sun$ &2.28 &1.29 &1.25 &1.27 & &18.3 &11.4 &10.8 &11.1 \\
788: 1.1$M_\sun$ &1.51 &0.63 &0.59 &0.61 & &18.9 &11.7 &10.8 &11.2 \\
789: 1.2$M_\sun$ &1.01 &0.55 &0.48 &0.51 & &19.1 &11.8 &10.5 &11.0 \\
790: 1.3$M_\sun$ &0.86 &0.52 & &0.46 & & 19.2 &12.0 & &11.1 \\
791: 1.4$M_\sun$ &0.90 & & & & &20.7 & & & \\
792: 1.6$M_\sun$ &0.95 & & & & &22.3 & & & \\
793: 1.8$M_\sun$ &0.98 & & & & &22.8 & & & \\
794: 2.0$M_\sun$ &0.92 & & & & &21.6 & & & \\
795: \enddata
796: %\tablecomments{Here we give both of the initial orbital period $P_{\rm ini}$ in days and the period at the Roche lobe
797: %overflow $P_{\rm rlof}$ in hours following \citet{pod02}.}
798: \end{deluxetable}
799:
800: \clearpage
801:
802:
803: \begin{figure}
804: \epsscale{.80} \plotone{f1.eps} \caption{Bifurcation periods as a
805: function of the secondary mass in an LMXB for the four kinds of
806: models described in \S 2.3. The dotted line shows the minimum
807: initial period $P_{\rm ZAMS}$ that corresponds to a Roche lobe
808: filling zero-age main-sequence secondary star.\label{bifur}}
809: \end{figure}
810:
811: \clearpage
812:
813:
814: \begin{figure}
815: \epsscale{.80}
816: \plotone{f2.eps}
817: \caption{Period evolutions of an LMXB with $M_{\rm 2,i}=1.0M_\sun$ and
818: $P_{\rm i}=1.20, 1.25, 1.40$ day in model 3. Circle, triangle, and cross
819: mark the onset of the Roche lobe overflow (RLOF), the end of the RLOF,
820: and the end of the calculation, respectively.
821: \label{tp}}
822: \end{figure}
823:
824: %\begin{figure}
825: %\epsscale{.80} \plotone{f3.eps} \caption{The secondary mass--period
826: %relation for the evolutionary sequences same as in Fig.~\ref{tp}.
827: % \label{mp}}
828: %\end{figure}
829:
830: %\begin{figure}
831: %\epsscale{.80}
832: %\plotone{f4.eps}
833: %\caption{Same as in Fig.~\ref{tp}, but for the evolutionary tracks of the donor star in the
834: %H-R diagram .
835: % \label{HR}}
836: %\end{figure}
837:
838: %\begin{figure}
839: %\epsscale{.80} \plotone{f5.eps} \caption{The mass transfer rates as
840: %a function of the secondary mass for the evolutionary sequences same
841: %as in Fig.~\ref{tp}.
842: % \label{MT}}
843: %\end{figure}
844:
845: %\begin{figure}
846: %\epsscale{.80} \plotone{f6.eps} \caption{The angular momentum loss
847: %timescales due to MB (solid lines), mass loss (dashed lines) and
848: %gravitational wave (dotted lines) as a function of the secondary
849: %mass for the evolutionary sequences same as in Fig.~\ref{tp} (upper
850: %panel: $P_{\rm i}=1.20$; middle panel: $P_{\rm i}=1.25$ day; bottom
851: %panel: $P_{\rm i}=1.40$ day).
852: % \label{AT}}
853: %\end{figure}
854:
855:
856: %\begin{figure}
857: %\epsscale{.80}
858: %\plotone{f7.eps}
859: %\caption{Comparison of the period evolutions in models 1 and 2 for an LMXB with
860: %$M_{\rm 2,i}=0.5\,M_\sun$. The initial orbital periods are taken to be $P_{\rm i}=2.5$,
861: %3.5, and 4.5 days, respectively. The meaning of the symbols are same as in Fig.~\ref{tp}.
862: % \label{MBa}}
863: %\end{figure}
864:
865: %\begin{figure}
866: %\epsscale{.80}
867: %\plotone{f8.eps}
868: %\caption{Comparison of the period evolutions in models 1 and 2 for an LMXB with
869: %$M_{\rm 2,i}=1.0\,M_\sun$. The initial orbital periods are taken to be $P_{\rm i}=1.0$,
870: %1.5, and 2.5 days, respectively. The meaning of the symbols are same as in Fig.~\ref{tp}.
871: % \label{MBb}}
872: %\end{figure}
873:
874: %\begin{figure}
875: %\epsscale{.80}
876: %\plotone{f9.eps}
877: %\caption{Comparison of the period evolutions in models 3 and 4 for an LMXB with
878: %$M_{\rm 2,i}=1.0M_\sun$. The initial orbital periods are taken to be $P_{\rm i}=0.4$,
879: %0.5, and 0.6 day, respectively. The meaning of the symbols are same as in Fig.~\ref{tp}.
880: % \label{massloss}}
881: %\end{figure}
882:
883: %\begin{figure}
884: %\epsscale{.80}
885: %\plotone{f10.eps}
886: %\caption{Orbital period (upper panel) and donor mass (lower pane ) evolution for an
887: %LMXB with $M_{\rm 2,i}=1.3M_\sun$ and $P_{\rm i}=0.46$ day in model 4. An ultra-compact
888: %binary system ($P_{\rm f}=21$ min) is finally formed. \label{UCXB}}
889: %\end{figure}
890:
891: \begin{figure}
892: \epsscale{.80} \plotone{f3.eps} \caption{Maximum mass transfer
893: rates in an LMXB consisting of $1.4M_\sun$ neutron star and a
894: $1M_\sun$ secondary at fixed orbital periods from $0.35$ d to $0.95$
895: d. Crosses marks the calculated data and the solid line represents a
896: logarithmic fit. \label{mdot2}}
897: \end{figure}
898:
899: \begin{figure}
900: \epsscale{.80} \plotone{f4.eps} \caption{Comparison of the
901: semi-analytical results of $P_{\rm rlof}$ with the numerical values
902: for model 1 and model 4. The dotted line shows the minimum initial
903: period $P_{\rm ZAMS}$ that corresponds to a Roche lobe filling
904: zero-age main-sequence secondary star. Here the semi 1 results
905: are calculated with $\alpha=1$ for $M_{\rm 2,i} \le 1.1M_\sun$, and $\alpha=0$
906: for $M_{\rm 2,i} \ge 1.2M_\sun$ where $\alpha=1$ should be used.
907: The reasons why we do this are given in the text. \label{bifurolf}}
908: \end{figure}
909:
910: \clearpage
911:
912:
913:
914:
915: \clearpage
916:
917:
918:
919:
920: %\begin{table}
921: %\begin{center}
922: %\caption{Calculated results of the bifurcation periods (in units of
923: %days) for different binary models.\label{tab1}}
924: %\begin{tabular}{ccccc}
925: %\tableline\tableline
926: %$M_{2,\rm{i}}$\tablenotemark{a} & Model 1 & Model 2 & Model 3 & Model 4 \\
927: %\tableline
928: %0.5$M_\sun$ &3.20 &4.08 &4.12 &4.01 \\
929: %0.6$M_\sun$ &3.02 &3.41 &3.40 &3.35 \\
930: %0.7$M_\sun$ &2.86 &2.81 &2.79 &2.77 \\
931: %0.8$M_\sun$ &2.77 &2.41 &2.38 &2.37 \\
932: %0.9$M_\sun$ &2.73 &2.14 &2.09 &2.10 \\
933: %1.0$M_\sun$ &2.28 &1.29 &1.25 &1.27 \\
934: %1.1$M_\sun$ &1.51 &0.63 &0.59 &0.61 \\
935: %1.2$M_\sun$ &1.01 &0.55 &0.48 &0.51 \\
936: %1.3$M_\sun$ &0.86 &0.52 & &0.46 \\
937: %1.4$M_\sun$ &0.90 & & & \\
938: %1.6$M_\sun$ &0.95 & & & \\
939: %1.8$M_\sun$ &0.98 & & & \\
940: %2.0$M_\sun$ &0.92 & & & \\
941: %\tableline
942: %\end{tabular}
943: %% Any table notes must follow the \end{tabular} command.
944: %\tablenotetext{a}{$M_{2,\rm{i}}$ -- initial mass of the secondary
945: %star. See the text for more information about the four models.}
946: %\end{center}
947: %\end{table}
948:
949:
950: %\begin{table}
951: %\begin{center}
952: %\caption{The bifurcation periods (in units of hours) following the
953: %definition of \citet{pod02} for different binary
954: %models.\label{tab2}}
955: %\begin{tabular}{ccccc}
956: %\tableline\tableline
957: %$M_{2,\rm{i}}$ \tablenotemark{a} & Model 1 & Model 2 & Model 3 & Model 4 \\
958: %\tableline
959: %0.5$M_\sun$ &31.8 &16.5 &18.2 &13.6 \\
960: %0.6$M_\sun$ &25.6 &11.3 &11.1 &10.0 \\
961: %0.7$M_\sun$ &21.5 &10.0 &9.8 &9.6 \\
962: %0.8$M_\sun$ &19.0 &10.1 &9.9 &9.8 \\
963: %0.9$M_\sun$ &17.9 &10.7 &10.3 &10.4 \\
964: %1.0$M_\sun$ &18.3 &11.4 &10.8 &11.1 \\
965: %1.1$M_\sun$ &18.9 &11.7 &10.8 &11.2 \\
966: %1.2$M_\sun$ &19.1 &11.8 &10.5 &11.0 \\
967: %1.3$M_\sun$ &19.2 &12.0 & &11.1 \\
968: %1.4$M_\sun$ &20.7 & & & \\
969: %1.6$M_\sun$ &22.3 & & & \\
970: %1.8$M_\sun$ &22.8 & & & \\
971: %2.0$M_\sun$ &21.6 & & & \\
972: %\tableline
973: %\end{tabular}
974: %% Any table notes must follow the \end{tabular} command.
975: %\tablenotetext{a}{$M_{2,\rm{i}}$ -- initial mass of the secondary star.}
976: %\tablenotetext{b}{This value is consistent with the result of \citet{pod02} (about $18$ hr), though
977: %we have assumed conservative mass transfer in model 1, while \citet{pod02} assumed mildly non-conservative
978: %mass transfer. See the text for more information.}
979: %\end{center}
980: %\end{table}
981:
982:
983:
984:
985:
986:
987:
988: \end{document}
989: