1: \documentclass[11pt,twoside]{article}
2: \usepackage{asp2004}
3: \usepackage{psfig}
4: \usepackage{epsf}
5: \usepackage{graphics}
6: \usepackage{lscape}
7: \markboth{B\"uning and Ritter}{Mass transfer cycles in compact binaries}
8: \pagestyle{myheadings}
9: \setcounter{equation}{0}
10: \setcounter{figure}{0}
11: \setcounter{footnote}{0}
12: \setcounter{section}{0}
13: \setcounter{table}{0}
14:
15: \begin{document}
16:
17: \title{Irradiation-driven Mass Transfer Cycles in Compact Binaries}
18:
19: \author{Andreas B\"uning and Hans Ritter}
20: \affil{Max-Planck-Institut f\"ur Astrophysik,
21: Karl-Schwarzschild-Str. 1,
22: D-85741 Garching bei M\"unchen}
23:
24: \begin{abstract}
25: We elaborate on the analytical model of Ritter, Zhang, \& Kolb (2000)
26: which describes the basic physics of irradiation-driven mass transfer
27: cycles in semi-detached compact binary systems. In particular, we take
28: into account a contribution to the thermal relaxation of the donor
29: star which is unrelated to irradiation and which was neglected in
30: previous studies. We present results of simulations of the evolution
31: of compact binaries undergoing mass transfer cycles, in particular
32: also of systems with a nuclear evolved donor star. These computations
33: have been carried out with a stellar evolution code which computes
34: mass transfer implicitly and models irradiation of the donor star in
35: a point source approximation, thereby allowing for much more
36: realistic simulations than were hitherto possible. We find that
37: low-mass X-ray binaries (LMXBs) and cataclysmic variables (CVs) with
38: orbital periods $\la 6$hr can undergo mass transfer cycles only for
39: low angular momentum loss rates. CVs containing a giant donor or one
40: near the terminal age main sequence are more stable than previously
41: thought, but can possibly also undergo mass transfer cycles.
42:
43: \end{abstract}
44:
45: \section{Introduction}
46:
47: The possible importance of irradiating the donor star of a
48: semi-detached compact binary by accretion luminosity for its long-term
49: evolution has first been pointed out by Podsiadlowski (1991).
50: Subsequently, the stability of mass transfer with irradiation feedback
51: has been studied in some detail by King et al. (1996, 1997, hereafter
52: KFKR96 and KFKR97), and by Ritter, Zhang, \& Kolb (2000, hereafter
53: RZK00). In KFKR96 and KFKR97 it was also shown that mass transfer in
54: cataclysmic variables (CVs) and low-mass X-ray binaries (LMXBs) can
55: become unstable against irradiation feedback, and that in case of
56: instability mass transfer proceeds in cycles in which episodes of
57: irradiation-driven mass transfer alternate with low-states during
58: which mass transfer is essentially shut off. Evolutionary calculations
59: of mass transfer cycles which were based on a homology model for the
60: donor and a simplified irradiation model, have been presented in KFKR97
61: and RZK00. Here we elaborate on the work done in previous studies in
62: two respects: First, we have developed a binary evolution code
63: which allows us to simulate the evolution of compact binaries with
64: mass transfer cycles more realistically than was possible in
65: previous studies, and second, we elaborate on the stability analysis of
66: KFKR96, KFKR97 and RZK00, thereby taking into account a contribution
67: to the thermal relaxation of the donor star which is unrelated to
68: irradiation and which has been neglected in previous studies.
69:
70: \section{Input Physics and Model Assumptions for the Numerical
71: Calculations}
72:
73: In the following we are briefly listing the main model assumptions and
74: details of the input physics adopted for our numerical
75: calculations. For details the reader is referred to B\"uning \&
76: Ritter (2004), hereafter BR04.
77:
78: \subsection{The Stellar Evolution Code}
79: Basically, we use the 1D stellar evolution code described by Schlattl,
80: Weiss, \& Ludwig (1997) and Schlattl (1999). For calculating binary
81: evolution, in particular for determining the mass transfer rate
82: essentially free of numerical noise, considerable refinements in the
83: calculation of the equation of state and the opa\-cities were necessary.
84: A detailed description of what has been done and how, is given in BR04.
85:
86: \subsection{Computing Mass Transfer}
87:
88: The mass loss rate from the donor star $-\dot M_2$ is computed
89: following Ritter (1988), i.e.\ from an explicit relation of the form
90: \begin{equation}
91: %
92: -\dot M_2 = \dot M_0 \;e^{-\frac{R_{\rm R,2}-R_2}{H_{\rm P}}}\,
93: %
94: \end{equation}
95: where $R_2$ and $R_{\rm R,2}$ are respectively the radius of the donor
96: star and the corresponding critical Roche radius, $H_{\rm P}$ the
97: effective photospheric pressure scale height of the donor, and
98: $\dot M_0\geq 0$ a slowly varying function of the donor's mass,
99: radius, and photospheric parameters. For our numerical computations
100: Eq.~(1) is formulated as an outer boundary condition for the donor
101: star and is thus solved implicitly with the stellar structure
102: equations. The numerical setup is similar to what has been used by
103: Benvenuto \& de Vito (2003) and is described in detail in B\"uning
104: (2003, hereafter B03).
105:
106: \subsection{Irradiation Physics}
107:
108: If the unperturbed donor star has a deep outer convective envelope,
109: i.e.\ if it is a cool star, irradiation, if not too strong, can be
110: treated as a local problem (for a detailed justification see RZK00).
111: Therefore, the problem reduces to specifying the intrinsic flux
112: $F_{\rm int}$, i.e.\ the true energy loss of the donor per unit
113: surface area and unit time as a function of the component of the
114: irradiating flux perpendicular to the stellar surface $F_{\rm irr}$.
115: For our numerical computations we use for $F_{\rm int}(F_{\rm irr})$
116: results tabulated by Hameury \& Ritter (1997) and additional data
117: kindly computed by Hameury (private communication) at our request.
118:
119: \subsection{Irradiation Model}
120:
121: The irradiation model links the momentary accretion luminosity
122: $L_{\rm accr}$ (liberated by accretion onto the compact star) to the
123: irradiating flux seen by a surface element of the irradiated donor
124: star. For our numerical computations we adopt the so-called point
125: source model which assumes that the spherical donor (of radius $R_2
126: \approx R_{\rm R,2}$) is illuminated by a point source with luminosity
127: $L_{\rm accr}$ at the position of the accretor, i.e.\ at the orbital
128: distance $a$. Because the accretion luminosity is not necessarily
129: radiated isotropically or staedily from the accretor, and because only
130: a part of the irradiating flux is absorbed below the donor's
131: photosphere, the link between $F_{\rm irr}$ and $L_{\rm accr}$
132: \begin{equation}
133: %
134: F_{\rm irr} = \alpha \frac{L_{\rm accr}}{4 \pi a^2} h(\theta)
135: =: \left< F_{\rm irr} \right> \, h(\theta)
136: %
137: \end{equation}
138: involves a dimensionless and a priori unknown efficiency parameter
139: $\alpha < 1$ which is the main unknown quantity in the irradiation
140: problem. In the framework of the point source model the flux
141: $F_{\rm irr}$ depends also on the substellar latitude $\theta$ of the
142: irradiated surface element. This is taken into account by the function
143: $h(\theta)$ in Eq.~(2) (for details see e.g.\ RZK00 or BR04).
144:
145: \section{Stability Analysis}
146:
147: The stability of mass transfer in the presence of irradiation has been
148: studied previously in some detail by KFKR96, KFKR97 and RZK00. Because
149: some of the numerical results which we have obtained with the
150: above-described stellar evolution program were at variance with
151: results of these stability analyses, we have carried out a more refined
152: stability analysis and found that an additional term in the stability
153: criterion which becomes particularly important for giant donors had
154: been ignored in ealier considerations. Space limitations do not allow
155: us to repeat our stability analysis in detail. For this we refer the
156: reader to BR04. Rather we wish to give here only the main result. The
157: criterion for the stability of mass transfer can be written as
158: follows:
159: \begin{equation}
160: %
161: \frac{ds}{d\ln{-\dot{M_2}}} < \frac{{\tau}_{\rm ce}}{{{\tau}^{'}}_{\rm d}}
162: + \frac{H_{\rm P}}{R_2}\, \delta
163: %
164: \end{equation}
165: Here $s$ is the effective fraction of the donor surface through which
166: energy outflow from its interior is totally blocked by irradiation
167: (for a detailed definition of $s$ see BR04), $\tau_{\rm ce}$ is the
168: thermal time scale of the donor's convective envelope,
169: ${{\tau}^{'}}_{\rm d}$ the time scale on which mass transfer is driven,
170: i.e.\
171: \begin{equation}
172: %
173: \frac{1}{\tau^{'}}_{\rm d} = {\left(\frac{\partial \ln R_2}{\partial t}\right)}_{\rm nuc}
174: + {\left(\frac{\partial \ln R_2}{\partial
175: t}\right)}_{\rm th}
176: -\, 2 \, \frac{\partial \ln J}{\partial t}\,
177: %
178: \end{equation}
179: is the sum of the driving terms resulting from nuclear evolution and
180: thermal relaxation of the secondary, and from systemic loss of orbital
181: angular momentum $J$. In addition, if homology is used for describing
182: the structure of the secondary, $\delta$ in (3) can be approximated as
183: \begin{equation}
184: %
185: \delta = 4\,(1 - s) {\left(\frac{R_2}{R_{\rm 2,e}}\right)}^3
186: + (n + 1) {\left(\frac{R_2}{R_{\rm2, e}}\right)}^{-(n+2)}
187: \sim (n + 5),
188: %
189: \end{equation}
190: where $R_{\rm 2,e}$ is the thermal equilibrium radius of the
191: irradiated donor (under stationary irradiation) and $n=(\partial
192: \ln {\varepsilon}_{\rm nuc}/\partial \ln T)$ is the temperature
193: exponent of the nuclear energy generation rate ${\varepsilon}_
194: {\rm nuc}$. In the case of low-mass main sequence donors which burn
195: hydrogen via the pp-chain $n \approx 5$. In the case of giant donors
196: one has to set $n = -3$ for self-consistency.
197:
198: The new result is the last term in Eq.~(3). It derives from the
199: previously neglected fact that the thermal relaxation term, i.e.
200: $(\partial R_2/\partial t)_{\rm th}$, not only has a non-vanishing
201: derivative with respect to $\Delta R = R_2 - R_{\rm R,2}$ but also
202: with respect to $\Delta R_{\rm e} = R_2 - R_{\rm 2,e}$. Since the
203: last term of (3) is always positive it stabilizes mass transfer. Its
204: consequences are first that mass transfer with irradiation feedback
205: is always stable for very small driving rates, i.e.\ very small mass
206: transfer rates, and second that it is important for giant donors where
207: $H_{\rm P}/R_2$ is much larger than for main sequence stars.
208: Therefore, contrary to earlier results obtained by KFKR97,
209: mass transfer from giant donors in CVs is much more stable than
210: previously thought.
211:
212: \section{Results}
213:
214: Here we are mainly interested in answering the question whether
215: irradiation-driven mass transfer cycles can occur in CVs or LMXBs.
216: A necessary condition for such mass transfer cycles to be possible is
217: violation of the stability criterion (3). As can be seen from (3),
218: this is possible only if both terms on the right-hand side of (3) are
219: sufficiently small. It follows then immediately that mass transfer
220: cycles are most likely to occur if the donor star has a relatively
221: shallow convective envelope (i.e.\ a small $\tau_{\rm ce}$), and/or
222: mass transfer is driven on a long timescale ${\tau^{'}}_{\rm d}$, and
223: if the relative photospheric scale height, i.e.\ $H_{\rm P}/R_2$,
224: is small. On the other hand, the left-hand side of (3)
225: is itself also a function of the mass transfer rate (via the
226: irradiating flux) which vanishes for very small and very large
227: mass transfer rates and which attains a maximum value of $\sim 0.1$
228: for irradiating fluxes $1 \la \left< F_{\rm irr}\right>/F_0 \la 10$,
229: where $F_0$ is the unperturbed flux of the donor. Therefore, for a
230: given donor star (i.e.\ given values of $\tau_{\rm ce}$ and
231: $H_{\rm P}/R_2$) there is always only a restricted range of mass
232: transfer rates, which also depends on the efficiency factor $\alpha$
233: defined in (2), for which mass transfer cycles can occur. Because
234: of (2) the left-hand side of (3) vanishes for both very small and very
235: large values of $\alpha$. For physical reasons we can exclude values
236: $\alpha \ga 1$. But apart from this upper limit $\alpha$ has to be
237: treated as a free parameter.
238:
239: From what has just been explained it has probably become clear that
240: irradiation--driven mass transfer cycles can occur only under special
241: conditions. Given the binary parameters, the rate of systemic angular
242: momentum loss $-\dot J$, and an adopted value of $\alpha$ one can
243: explore the range of instability by using (3) and an analytical
244: approximation for $\delta$, e.g.\ the one given in (5). In this way one
245: can narrow down the parameter space of interest before going to
246: extensive (and expensive) numerical calculations. For more details on
247: this point we refer the reader to BR04.
248:
249: \begin{figure}[ht!]
250: \plotone{Fig1.eps}
251: \caption{Mass transfer rate (in $M_{\sun}$yr$^{-1}$) of a CV undergoing
252: irradiation-driven mass transfer cycles (full line) as a
253: function of its orbital period $P$ (in hours). The
254: corresponding evolution without irradiation feedback is shown
255: as a dashed line. For the parameters underlying these
256: calculations see text.}
257:
258: \end{figure}
259:
260: As an example of such a numerical calculation we show in Fig.~1 the
261: variation of the mass transfer rate (in $M_{\sun}$yr$^{-1}$) as a
262: function of orbital period $P$ (in hours) of a CV undergoing mass
263: transfer cycles (full line). For this calculation we have
264: assumed the following parameters: The systemic loss of orbital angular
265: momentum is due to gravitational radiation only, i.e.\ $\dot J = {\dot
266: J}_{\rm GR}$. The donor star is a standard Pop.~I low-mass main
267: sequence star with an initial mass $M_2 = 0.5 M_{\sun}$, an age of
268: $10^{10}$yr and a central hydrogen mass fraction of $X_{\rm c} \approx
269: 0.62$. In this case $H_{\rm P}/R_2 \approx 10^{-4}$. The primary is a
270: white dwarf with a mass of $M_1 = 0.8 M_{\sun}$ and a radius of $R_1 =
271: 0.010 R_{\sun}$. The transferred mass is assumed to eventually leave
272: the binary system (e.g.\ via nova explosions) with a dimensionless
273: orbital angular momentum $\partial \ln J_{\rm orb}/\partial \ln (M_1
274: + M_2) = M_2/M_1$. The irradiation efficiency parameter is $\alpha =
275: 0.3$. The dotted line shows the corresponding evolution without
276: irradiation feedback.
277:
278: What this example shows, is first how numerically accurate the stellar
279: evolution code described in Sect.~2 can follow such an evolution.
280: Second, we see that irradiation-driven mass transfer cycles can occur
281: in CVs for not unrealistic values of $\alpha$ provided that the
282: braking rate is small. In a corresponding evolution in which the
283: absolute value of the angular momentum loss rate is much higher, e.g.\
284: according to the Verbunt \& Zwaan (1981) prescription, no mass
285: transfer cycles would occur. We note that the step in the amplitude of
286: the mass transfer cycles at an orbital period of $\sim 3$ hr is due to
287: the secondary becoming fully convective. Third, because the thermal
288: time scale of the fully convective donor star increases with
289: decreasing mass, mass transfer eventually becomes stable at $P \approx
290: 2.5$ hr.
291:
292: Based on numerous numerical simulations (detailed in B03 or BR04)
293: and on the more general considerations outlined above we can summarize
294: our results as follows:
295:
296: \begin{enumerate}
297: \item In agreement with earlier results we find that CVs which contain
298: a main sequence donor and in which the driving rate above the
299: period gap is as high as required for explaining the period gap
300: in the framework of the model of disrupted magnetic braking
301: (see e.g.\ Spruit \& Ritter 1983) are stable against
302: irradiation feedback except for the most massive donor stars
303: $M_2 \sim 1 M_{\sun}$. On the other hand, mass transfer cycles
304: can occur in short-period CVs if the driving rate is small,
305: i.e.\ no larger than a few times the gravitational braking rate.
306:
307: \item CVs containing a donor star which is near the terminal age main
308: sequence turn out to be more stable than has been anticipated
309: based on simple homology arguments.
310:
311: \item CVs containing an extended giant donor with nuclear
312: timescale--driven mass transfer are less likely to undergo mass
313: transfer cycles than has been anticipated based on the results
314: of KFKR97. The main reason for this discrepancy is the second
315: term on the right-hand side of the stability criterion (3),
316: i.e.\ the relatively large value of $H_{\rm P}/R_2$ associated
317: with such stars. If mass transfer cycles do occur they are
318: characterized by comparatively very short phases with high,
319: irradiation-driven mass trasnsfer rates which are followed by
320: extended periods without mass transfer during which the system
321: is essentially detached and reattachment is reached only on the
322: nuclear time scale of the giant.
323:
324: \item Adopting the point source irradiation model which takes into
325: account irradiation of surface elements near the terminator of
326: the donor we find that possibly also LMXBs can undergo mass
327: transfer cycles. Regarding the braking rate which is necessary
328: to drive cycles, basically the same restrictions apply as for
329: short--period CVs. We confirm also that LMXBs containing a giant
330: donor can undergo cyles.
331:
332: \item Mass transfer cycles in CVs do occur only if $0.1 \la \alpha \la
333: 1$ whereas in the case of LMXBs cycles do not occur if $\alpha
334: \ga 0.1$.
335:
336: \item For systems containing an unevolved main sequence or a giant
337: donor the results of our numerical computations and the
338: predictions from the analytic model for the stability boundaries
339: are in reasonable agreement.
340: \end{enumerate}
341:
342: \begin{references}
343:
344: \reference Benvenuto, O.G., \& de Vito, M.A. 2003, \mnras, 342, 50
345: \reference B\"uning, A. 2003, Ph.D. thesis, Ludwig--Maximilians
346: Universit\"at M\"unchen (B03)
347: \reference B\"uning, A., \& Ritter, H. 2004, \aap, 423, 281 (BR03)
348: \reference Hameury, J.-M., \& Ritter, H. 1997, \aaps, 123, 273
349: \reference King, A.R., Frank, J., Kolb, U., \& Ritter, H. 1996, \apj,
350: 467, 761 (KFKR96)
351: \reference King, A.R., Frank, J., Kolb, U., \& Ritter, H. 1997, \apj,
352: 482, 919 (KFKR97)
353: \reference Podsiadlowski, Ph. 1991, \nat, 350, 136
354: \reference Ritter, H. 1988, \aap, 202, 93
355: \reference Ritter, H, Zhang, Z.-Y., \& Kolb, U. 2000, \aap, 360, 959
356: (RZK00)
357: \reference Schlattl, H. 1999, Ph.D. thesis, Technische Universit\"at
358: M\"unchen
359: \reference Schlattl, H., Weiss, A., \& Ludwig, H.-G. 1997, \aap, 322,
360: 646
361: \reference Spruit, H.C., \& Ritter, H. 1983, \aap, 124, 267
362: \reference Verbunt, F., \& Zwaan, C. 1981, \aap, 100, L7
363:
364: \end{references}
365: \end{document}