1: %%%%%%%%%%%%%%%%%%%%%%% file template.tex %%%%%%%%%%%%%%%%%%%%%%%%%
2: %
3: % This is a template file for these proceedings
4: %
5: % Copy it to a new file with a new name and use it as the basis
6: % for your article
7: %
8: %%%%%%%%%%%%%%%%%%%%%%%% EDP Sciences %%%%%%%%%%%%%%%%%%%%%%%%%%
9: %
10: \documentclass{eas}
11: \usepackage{graphicx}
12: \usepackage{astron}
13: %
14: %%%%%%%%%%%%%--PREAMBLE--%%%%%%%%%%%%%%%%%%
15: %%-----------------------------
16: % ...........
17: % your macros
18: % ...........
19: %%-------------------------%%----
20: %%%%%%%%%%%%%%%--BODY--%%%%%%%%%%%%%%%%%%
21: %
22: %\TitreGlobal{The Title of this Volume}
23: %
24: \begin{document}
25:
26: %%-----------------------------
27: %% the top matter
28: %%-----------------------------
29: \title{Observational Evidence for Tidal Interaction \\ in Close Binary Systems}
30: %
31: \runningtitle{Observation of Tidal Interaction in Close Binary Systems}
32: \author{Tsevi Mazeh}\address{School of Physics of Astronomy, Sackler
33: Faculty of Exact Sciences, Tel Aviv University}
34:
35: %
36: %
37: \begin{abstract}
38:
39: This paper reviews the rich corpus of observational evidence for tidal
40: effects, mostly based on photometric and radial-velocity measurements.
41: This is done in a period when the study of binaries is being
42: revolutionized by large-scaled photometric surveys that are detecting
43: many thousands of new binaries and tens of extrasolar planets.
44:
45: We begin by examining the short-term effects, such as ellipsoidal
46: variability and apsidal motion. We next turn to the long-term
47: effects, of which circularization was studied the most: a transition
48: period between circular and eccentric orbits has been derived for
49: eight coeval samples of binaries. The study of synchronization and
50: spin-orbit alignment is less advanced. As binaries are supposed to
51: reach synchronization before circularization, one can expect finding
52: eccentric binaries in pseudo-synchronization state, the evidence for
53: which is reviewed. We also discuss synchronization in PMS and young
54: stars, and compare the emerging timescale with the circularization
55: timescale.
56:
57: We next examine the tidal interaction in close binaries that are
58: orbited by a third distant companion, and review the effect of
59: pumping the binary eccentricity by the third star. We elaborate on the
60: impact of the pumped eccentricity on the tidal evolution of close
61: binaries residing in triple systems, which may shrink the binary separation.
62:
63: Finally we consider the extrasolar planets and the observational
64: evidence for tidal interaction with their parent stars. This includes
65: a mechanism that can induce radial drift of short-period planets,
66: either inward or outward, depending on the planetary radial position
67: relative to the corotation radius. Another effect is the
68: circularization of planetary orbits, the evidence for which can be
69: found in eccentricity-versus-period plot of the planets already known.
70:
71: Whenever possible, the paper attempts to address the possible
72: confrontation between theory and observations, and to point out
73: noteworthy cases and observations that can be performed in the future
74: and may shed some light on the key questions that remain open.
75:
76: \end{abstract}
77: %
78: \maketitle
79: %%-----------------------------
80: %% your text
81: %%-----------------------------
82:
83: %\tableofcontents
84:
85: %\newpage
86:
87: %========================
88: %
89: \section{Introduction}
90: %
91: %%=======================
92:
93: Stars in close binary systems are subject to mutual tidal forces that
94: distort their stellar shape, breaking their spherical and axial
95: symmetry, which leads to different observational effects. This paper
96: reviews the rich corpus of observational evidence for the tidal
97: effects, which has been accumulated mostly by photometric and
98: radial-velocity studies of short-period binaries.
99:
100: One can divide the observational effects of the stellar distortion
101: caused by tidal interaction into three classes:
102:
103: \begin{itemize}
104:
105: \item
106: Direct observations of the distorted shape of the two components of
107: the binary system. One example is the ellipsoidal periodic modulation
108: of the photometric intensity of the binary with the orbital period,
109: caused by the rotation of the stars with the ellipsoidal shape.
110:
111: \item
112: Deviation of the orbital motion of the binary from pure Keplerian
113: orbit. This is caused by departures of the gravitational attraction
114: from the inverse-square law, due to the distortion of the two stars
115: from spherical symmetry. One example is the famous apsidal motion
116: observed in many eccentric eclipsing binaries.
117:
118: \item
119: Long timescale evolution of the binary orbital elements and the
120: stellar rotation, induced by the tidal interaction. The long-term
121: interaction will take place as long as the tides raised by the mutual
122: interaction vary in size, shape and location on the surface of the two
123: components. The tides do not vary only when
124:
125: \begin{itemize}
126:
127: \item
128: the binary motion is circularized,
129:
130: \item
131: the stellar rotation is synchronized with the binary orbital
132: motion, and
133:
134: \item
135: the stellar rotation axis is aligned with the normal to the orbital plane
136: of motion.
137:
138: \end{itemize}
139: %
140: Observational evidence for this long-term evolution can be found, for
141: example, through the circular orbits of short-period binaries.
142:
143: \end{itemize}
144:
145: The timescales of the three classes are substantially different. The
146: ellipsoidal effect varies with the orbital period, which is of the
147: order of a few days. The timescale of the deviations of the binary
148: motion from the Keplerian orbit is in the range between tens and
149: thousands of years, whereas the last class timescale is of the order
150: of millions or billions of years. Therefore, we can witness close
151: binaries for which the effects of the first and the second class are
152: in action, while observation of the third class of effects can be made
153: only retroactively, in recognition of the results of a long timescale
154: processes.
155:
156: Observing one or more of these effects allows us to learn about the
157: response of each star to the tidal forces exerted by its
158: companion. This response depends on the stellar internal structure, and therefore observations of these effects can provide us
159: with a unique opportunity to study that structure in a
160: manner which is not possible with single stars.
161:
162: In the last few decades numerous observations of the tidal interaction in
163: close binaries have been accumulated. In many cases, observations of
164: binaries performed for other reasons could not have been accounted for
165: without the tidal effects. One example is the lightcurves of eclipsing
166: binaries, which are usually observed in order to derive the
167: geometrical elements of the systems. However, the lightcurves of these
168: close systems can not be fully understood without taking into account
169: the distortion of the two stars. In other cases, special observational
170: effort has been made in order to detect some evidence for tidal
171: effects. One example is the search for the circularization cutoff
172: period in samples of coeval binaries. Both types of cases give us an
173: opportunity to confront the tidal theory, based on stellar structure
174: and evolution, with the observations. This paper will review both
175: types of observations.
176:
177: Sections 2 \& 3 review the observational evidence for the
178: ellipsoidal modulation and the apsidal motion of eccentric binaries.
179: Section 4, 5, and 6 discuss the evidence for long-term
180: circularization, synchronization and alignment of short-period
181: binaries. Section 7 covers the tidal interaction in triple systems,
182: when the third distant companion can have an impact on the tidal
183: evolution of the close binary.
184:
185: The study of tidal interaction between the recently discovered
186: extrasolar planets and their parent stars has gained rapid momentum
187: since 1995, following the discovery of the first confirmed extrasolar
188: planet. Section 8 reviews some aspects of this fascinating new field.
189:
190: This paper has been written at a juncture when the study of close
191: binaries has been undergoing a transitional phase, the result of
192: large-scaled photometric surveys which are starting to produce large
193: samples of eclipsing binaries. In acknowledgment of the new era,
194: Section 9 reviews some of the recent results from two of these
195: surveys. Finally, Section 10 summarizes the previous sections and
196: discusses some aspects of future studies of tidal interaction in close
197: binaries.
198:
199: This paper is too short and my knowledge too incomplete to cover all
200: the studies in the different aspects of the subject. I made a great
201: effort, though, to cover in each of the sections at least the open
202: questions in that subfield. I apologize to those whose work was not
203: included in this review. I have made an effort to keep the notation
204: of each section consistent with the other sections of this
205: review. This forced me to adopt some notations that are not so common
206: in the different fields covered by this paper. Each section was
207: written as a separate piece of study, so many definitions were
208: repeated throughout this paper.
209:
210: Finally, I hope that this review will be of some use to young
211: researchers in the field. It would fulfill my highest expectations if
212: some research is initiated as a result of reading this review and
213: considering some of questions that it raises.
214:
215: %=============================
216: %
217: \section{Ellipsoidal effect}
218: %
219: %=============================
220:
221: Ellipsoidal variables are close binary systems whose lightcurves show
222: periodic variations due to the distortion of the components caused by
223: their mutual attraction. Usually, this term is used (e.g., Morris \&
224: Naftilan \nocite{mornaf93} 1993) to denote binaries whose orbital
225: inclinations are not large enough to render them eclipsing systems,
226: although the effect is seen in eclipsing binaries as well.
227:
228: When calculating the ellipsoidal effect (e.g., Kopal 1959, Morris
229: \nocite{morris1985}1985, Morris \& Naftilan 1993) one usually assumes
230: that the two stars are in equilibrium in the binary rotating reference
231: frame, which means that the stars are synchronized and aligned and
232: that the orbit is circularized. Models of stellar evolution and
233: circularization and synchronization (e.g., Zahn 1977) have shown that
234: for binaries with periods less than a few days these conditions are
235: reached on a much shorter timescale than the lifetimes of the
236: stars. Extensive observational evidence supports this theoretical
237: result (see Sections 4--6). For longer periods, the ellipsoidal
238: effect is small anyway.
239:
240: To derive the ellipsoidal effect, Morris \& Naftilan (1993) expanded
241: the periodic variation into discrete Fourier series with terms that
242: depend on the ratio $R_*/a$, where $R_*$ is the stellar radius and $a$
243: is the semi-major axis of the binary orbit. Assuming $R_*/a$ is small,
244: the leading term to present the variation of the {\it primary}, with a
245: radius $R_1$, of the
246: order of $(R_1/a)^3$, has a semi-amplitude of:
247: \begin{equation}
248: \left( \frac{\Delta F_1}{F_1}\right)_{\mathrm{ellips}} =
249: 0.15\frac{(15+u_1)(1+g_1)}{3-u_1}\frac{M_2}{M_1}
250: \left(\frac{R_1}{a}\right)^3
251: \ \sin^2 i \ .
252: \end{equation}
253: In the above expression, $g_1$ is the gravity darkening coefficient
254: of the primary and $u_1$ is its limb-darkening coefficient, and $i$ is
255: the orbital inclination.
256:
257: For a period of a few days and in main-sequence stars, the amplitude
258: of the ellipsoidal effect is a few percent, and therefore the effect
259: can be easily measured. For a spectroscopic binary, known by its
260: radial-velocity modulation, measuring the amplitude of the ellipsoidal
261: effect can substantially add to our knowledge of the system, as this
262: amplitude can add some constraints on the stellar radius, mass and the
263: binary inclination. If we can estimate the stellar radius from its
264: spectral type and the semi-major axis of the binary from its period,
265: we can estimate the inclination of the binary, which is a crucial
266: parameter for accurately obtaining the stellar masses of double-lined
267: spectroscopic binaries.
268:
269: Observational astronomy is always complex, and the ellipsoidal effect
270: is no exception. First, when calculating the expected amplitude of the
271: effect, one must take into account the variation of the secondary star
272: as well. In fact, as the intensities of the two stars are both
273: modulated with half the orbital period, the observed effect is the
274: weighted average of the two stars. Therefore, deriving the inclination
275: from the ellipsoidal amplitude is not a trivial matter and might
276: necessitate observations in a few wavelengths. Second, other stellar
277: modulations might also be present, like in the case of the $\delta$
278: Scuti star XX Pyx (Aerts \etal\ 2002)\nocite{aerts2002}. Fortunately,
279: the ellipsoidal effect has a clear periodicity of half the orbital
280: period, which is different, for example, from the stellar spot
281: modulation that varies, for a synchronized system, with the orbital
282: period.
283:
284: %---------------------------------------------------------------
285: \subsection{Interesting systems: Black holes in X-ray binaries}
286: %----------------------------------------------------------------
287:
288: One field in which the ellipsoidal variation was found to be critical is
289: the study of X-ray binaries, which are close binaries with a compact
290: object --- a white dwarf, a neutron star or a black hole, together
291: with a main-sequence or a giant optical companion. The X-ray
292: luminosity, generated around the compact object, is fuelled by mass
293: transfer from the companion onto the compact object. In many of these
294: systems, the companion fills its Roche lobe and therefore its shape is
295: substantially distorted and the corresponding ellipsoidal effect is
296: relatively large.
297:
298: In almost all the X-ray binaries, the ellipsoidal effect is crucial to
299: putting strong constraints on the orbital inclination and the mass of
300: the compact object. The mass of the compact object is the only way to
301: identify stellar black holes (=BH), and therefore the ellipsoidal
302: variability is an essential step in studying these general-relativity
303: objects (e.g., Orosz 2003\nocite{orosz2003}). Although one has to take
304: into account the unknown contribution of the accretion disc to the
305: luminosity of the system, careful analysis can give some constraints
306: on the mass of the compact object (e.g., Charles \& Coe
307: 2006\nocite{charles2006}). Two such examples are V616 Mon (e.g.,
308: Harrison \etal\ \nocite{harrison2007}2007) and XTE J118+480 (Gelino
309: \etal\ \nocite{gelino2006}2006).
310:
311: An interesting case is M33 X-7, the first stellar BH found in the
312: neighbouring galaxy M33 (Pietsch \etal\ \nocite{pietsch06}2006). With
313: a period of 3.45 days, the ellipsoidal amplitude was measured to be of
314: about 4\% (Shporer \etal\ \nocite{shporer07}2007), and contributed to
315: the derivation of a mass of $15.65 \pm 1.45$ $M_{\odot}$ (Orosz \etal\
316: \nocite{orosz2007}2007).
317:
318: %===========================
319: %
320: \section{Apsidal motion}
321: %
322: %===========================
323:
324: Apsidal motion, the precession of the orbital ellipse in the binary
325: plane of motion, can be measured by the detection of a small change of
326: $\omega$ --- the angle denoting the direction of the line of apsides
327: in the orbital plane. In a Keplerian motion the ellipse and its
328: orientation are constants of motion, and only deviation from
329: Newtonian inverse-square law attraction can cause the ellipse to
330: precess. The deviation results from the asymmetrical stellar shape,
331: caused by the tidal forces of its companion. The precession period is
332: longer than the orbital period by many orders, and therefore one can
333: first derive the osculating orbital parameters, ignoring the apsidal
334: precession, and only afterward search for variations of $\omega$ over a
335: longer timescale.
336:
337: The classical theoretical derivation of the precession rate ignores
338: the general relativity effect and assumes spin-orbit alignment of the
339: two stars (see discussion of the two effects in Subsections 3.1 and 3.2).
340: To calculate theoretically the expected rate of apsidal motion one has
341: first to derive the stellar shape induced by the tides, and then
342: to calculate the resulting advance of the line of apsides. For an
343: historical review of the theoretical calculation which started back in
344: 1928 by Russell \nocite{russell1928} (1928), see Claret \& Gim{\'e}nez
345: \nocite{claret1993} (1993). The apsidal precession period due {\it
346: only} to the primary star, $U_{tidal,1}$, taking into account only the
347: linear part of the stellar quadruple moment and up to the second
348: order in the eccentricity $e$, is given by the formula:
349: %
350: \begin{equation}
351: \frac{P}{U_{tidal,1}} \simeq k_{2,1}\left(\frac{R_1}{a}\right)^5
352: \left[15f_2(e)q+\left(\frac{\Omega_{1,rot}}{n_{orbit}}\right)^2
353: (1+q) \right]
354: \end{equation}
355: %
356: where $P$ is the orbital period, $q$ is the mass ratio --- $M_2/M_1$,
357: where $M_1$ is the primary mass and $M_2$ is the secondary mass,
358: $\Omega_{1,rot}$ is the {\it rotational} frequency of the primary,
359: $n_{orbit}=2\pi/P$ is the averaged orbital frequency, $R_1$ is the radius of
360: the primary, $a$ is the orbital semi-major axis, and $f_2(e) \simeq
361: (1+\frac{3}{2}e^2)$ is a function of $e$ of order unity. The parameter
362: $k_{2,1}$ is the second-order coefficient of the internal structure of
363: the primary, and reflects the stellar radial concentration (e.g.,
364: Cowling 1938\nocite{cowling1938}; Sterne \nocite{sterne1939a}
365: 1939a; b; c\nocite{sterne1939b}\nocite{sterne1939c}; Kopal
366: \nocite{kopal1959}1959). For infinite degree of concentration of the
367: primary (mass point) $k_{2,1}$ becomes zero, while for a homogeneous
368: configuration $k_{2,1}=0.75$ (e.g., Claret \& Gim{\'e}nez
369: \nocite{claret1993} 1993). Typical models yield $k_{2,1}$ of the order
370: of 0.01--0.001, depending on the stellar mass and age. The precession
371: is in the direction of the orbital motion, and therefore
372: $\dot{\omega}$ is always positive. Such a precession is sometimes
373: termed direct precession.
374:
375: In most systems one must also take into account the secondary
376: contribution to the precession, and therefore one uses $\bar{k}_2$ ---
377: some weighted average of $k_{2,1}$ and $k_{2,2}$, the latter being the
378: structure parameter of the secondary (e.g., Claret
379: \nocite{claret1999}1999).
380:
381: Apsidal motion can be measured with relative accuracy in eccentric {\it
382: eclipsing} binaries, as $e$ and $\omega$ determine the time interval
383: between the primary and secondary minima (e.g., Guinan \& Maloney
384: \nocite{guinan1985} 1985). When the time interval is substantially
385: different from half the orbital period, small changes of $\omega$ can
386: be detected. Observational determination of timing of minima, performed
387: over a baseline of the order of ten years, can reveal a precession of a
388: degree or two, indicating a precession period of thousands of years.
389:
390: For double-lined {\it spectroscopic eclipsing} binaries, where the
391: masses and radii of the two stars can be derived from the lightcurve
392: and the radial-velocity curves, observation of the
393: apsidal precession period can directly tell us the value of
394: $\bar{k}_2$. Therefore, the apsidal motion is a way to confront the
395: theory of stellar structure with observations (e.g., Gim{\'e}nez et
396: al. 1987\nocite{gimenez1987}; Gim{\'e}nez\nocite{gimenez1990} 1990). There
397: are very few contemporary observations that can directly challenge
398: the stellar structure theory without relying on stellar
399: evolution. Solar neutrino experiment (e.g., Bahcall \& Ulrich
400: \nocite{bahcall1988}1988) and stellar seismology (e.g., Dziembowski \&
401: Pamyatnykh \nocite{dziembowski1991}1991) are two of the very few other
402: examples. This is why the apsidal motion, although limited to
403: eclipsing binaries with short periods, is of fundamental importance to
404: stellar astrophysics. For a compilation of the binaries with measured
405: apsidal motion up to their time see Petrova \& Orlov
406: (\nocite{petrova1999}1999). Very recently Bulut \& Demircan
407: (\nocite{bulut2007}2007) published a new catalog of 124 eclipsing
408: binaries, including the pertinent observed apsidal motions.
409:
410: Claret \& Gim{\'e}nez \nocite{claret1993} (1993; see also Claret
411: \nocite{claret1999} 1999; \nocite{claret2007} 2007) carefully
412: considered the observational result of 14 double-lined eclipsing
413: binaries with known absolute dimensions and accurately observed
414: apsidal motion rates. They also included in their study an additional
415: ten systems with lower quality data. Although there have been
416: arguments for some inconsistencies between the old stellar models and
417: the observed apsidal motion, Claret \& Gim{\'e}nez found that the
418: observations are consistent with their modern stellar models (see also
419: Claret \& Willems \nocite{claret2002} 2002 and Petrova \&
420: Orlov \nocite{petrova2002}2002; 2003). As in the case of the solar
421: neutrino (Bahcall \etal\ 2001\nocite{bahcall2001}), the consistency
422: between the theory and observation of the apsidal motion for most of
423: the binaries should be considered one of the great achievements of
424: stellar astrophysics.
425:
426: Apsidal motion can still refine our understanding of stellar
427: interiors. This is especially true of stars with convective cores,
428: where the present models include a somewhat arbitrarily chosen value of
429: the overshooting parameter (e.g., Claret \& Willems 2002). The value
430: of this parameter can be deduced from the apsidal motion precession
431: period, as was done in the study of V380 Cyg (Guinan \etal\
432: \nocite{guinan2000}2000), which consists of two B stars.
433:
434: Given the agreement between theory of stellar structure and
435: observations, we can now reverse the reasoning and use the observed
436: tidal apsidal precession to derive the masses of the systems for
437: spectroscopic binaries that do not show any eclipse (Jeffery
438: \nocite{jeffery1984}1984). For those systems, detection of the tidal
439: precession can provide missing information about the orbital
440: inclination, information that can be found in eclipsing binaries from
441: the depth, shape and duration of the eclipse. Although the measurement
442: of the precession in spectroscopic binaries is more difficult than in
443: eclipsing binaries, Benevenuto \etal\ (\nocite{benenuto2002}2002) used
444: this method to derive the masses of the two O-stars in the HD 93205
445: binary system.
446:
447: \subsection{Relativistic Precession}
448: %-------------------------------------
449:
450: Deviation from inverse-square law gravitational attraction in binary
451: systems appears also because of a general relativistic (=GR) effect that
452: in some cases can not be neglected. The resulting direct apsidal
453: precession amounts to
454:
455: \begin{equation}
456: U_{GR} = 1800 (1-e^2) \left(\frac{P}{\rm{day}} \right)^{5/3}
457: \left(\frac{M_1+M_2}{M_{\odot}} \right)^{-2/3} \,{\rm yrs} \ .
458: \end{equation}
459: %
460: (e.g., Weinberg 1972\nocite{weinberg1972}).
461: The combined precession period, $U_{tot}$, is
462: \begin{equation}
463: \frac{1}{U_{tot}}=\frac{1}{U_{GR}}+\frac{1}{U_{tidal}} \ .
464: \end{equation}
465: %
466: Therefore, in order to derive the tidal precession from any observed
467: precession we have, in principle, to subtract the GR
468: effect. Fortunately, the GR precession can be predicted accurately, as
469: it depends only on orbital parameters and does not depend on the
470: stellar radius and internal structure.
471:
472: In order to compare the two precessions, we note that although for
473: constant $R_1/a$ the apsidal motion goes linearly with $P$, for
474: constant primary stellar radius $R_1$ we get that
475:
476: \begin{equation}
477: {1 \over U_{tidal}} \propto P^{-13/3} \ \ \ {\rm while} \ \ \ {1 \over U_{GR}}\propto P^{-5/3} \ .
478: \end{equation}
479: %
480: Therefore, for binaries with long enough periods, the tidal
481: precession becomes dominant and the GR precession can be neglected
482: (see Figure~\ref{apsidal_motion_LMC} and the relevant discussion).
483:
484: %------------------------------
485: \subsection{The effect of spin-orbit misalignment on the precession rate}
486: %------------------------------
487: \label{DI_Her}
488: %----------------------
489:
490: Another effect that can change the apsidal precession rate is
491: spin-orbit misalignment of one or two of the stars in the binary
492: system. One possible example for this effect is DI Her, which is one
493: of the eclipsing binaries that has presented a persistent challenge to
494: the theory of apsidal precession (Martinov \& Khaliullin
495: \nocite{martynov1980} 1980) for more than 25 years. The system
496: consists of two bright B stars orbiting each other with a period of
497: 10.55 days and eccentricity of $e=0.49$. For an historic account of
498: the study of the system see Guinan \& Maloney
499: (\nocite{guinan1985}1985, hereafter GM85).
500:
501: Modern value for the apsidal motion of DI Her was observed (GM85) to
502: be $55,400$ yrs. However, the stellar models, which rely on the
503: observed stellar radii and rotations, yielded a much shorter tidal
504: precession period --- of the order of 20,000 yrs. The inconsistency
505: with the theory got worse, as the GR precession period, which does not
506: depend on any stellar model, was estimated to be even shorter ---
507: $15,400$ yrs. Together, the two effects amounted to a precession period
508: as short as 8,400 yrs, almost one order of magnitude shorter than the
509: observed apsidal period. Such a confrontation between theory and
510: observations can be very fruitful. It can indicate either that the
511: theory is wrong, or that we are ignoring effects that are important
512: for the system.
513:
514: Shakura (\nocite{shakura1985}1985; see also Company \etal\
515: \nocite{company1988}1988) suggested one possible way to 'save the
516: phenomenon' of DI Her, by assuming the rotation axis of at least one
517: of the two stars is not aligned with the orbital angular momentum of
518: the system. In such a case, the companion's gravitational force
519: exerted on the equatorial stellar rotational bulge induces a {\it
520: counter} precession of the orbit, the rate of which depends on the
521: angle of misalignment and the size of the equatorial bulge. The net
522: result of the misalignment could be a substantial reduction of the
523: apsidal precession rate. Reisenberger \& Guinan
524: (\nocite{reisenberger1989}1989), who studied Shakura's idea in depth,
525: concluded that in order to account for the observed slow precession of
526: DI Her one has to assume a misalignment of about $50^{\circ}$.
527:
528: DI Her is not the only binary for which the apsidal motion might
529: indicate some spin-orbit misalignment. Petrova \& Orlov
530: (\nocite{petrova2003}2003) discussed the possible misalignment effect in
531: broader context, and considered a few more interesting systems.
532:
533: The alignment of the rotational axes of the stars with the angular
534: momentum of their binary orbit is discussed in detail in
535: Section~\ref{section_alignment}. The conjecture that the stellar spin
536: axes are not aligned with the binary motion involves two interesting
537: theoretical assumptions. First, if this idea is true for DI Her and
538: the other systems, it means that these binaries were formed with
539: misaligned stellar rotation. Second, the present misalignment means
540: that the long-term tidal dissipation processes did not have enough
541: time to align the stellar spins (see Section 6 below). In fact,
542: Reisenberger \& Guinan (1989) estimated that the lifetime of the two
543: components of DI Her is substantially shorter than the alignment
544: timescale, and therefore one would expect DI Her to retain its
545: original spin-orbit misalignment.
546:
547: Spin-orbit misalignment can cause two additional interesting
548: phenomena, in addition to the counter apsidal precession. The binary
549: interaction causes the stellar rotation axis {\it and} the orbital
550: orbit to precess around the total angular momentum of the system with
551: the same period. Such a forced precession will change the stellar spin
552: relative to our line of sight, resulting in a change of the stellar
553: rotational broadening of the two stars. It might also change slightly
554: the inclination angle of the binary, which modifies the eclipse
555: parameters. Reisenberger \& Guinan (\nocite{reisenberger1989}1989)
556: followed Shakura's suggestion and tried to find observational hints
557: for these two modulations, but the evidence they have collected was
558: inconclusive. Since almost 20 years have passed since the study
559: of Reisenberger \& Guinan, an observational revisit of the system
560: might yield more reliable evidence that can either confirm or refute the
561: orbital and stellar precession, as the time baseline of the
562: inclination measurements will be then quite longer.
563:
564: We note that the advance of the observational techniques since the
565: mid-1980's renders the Shakura conjecture for DI Her easily
566: observable. During the eclipse one can observe the Rossiter-McLaughlin
567: effect (Rossiter \nocite{rossiter1924} 1924; McLaughlin
568: \nocite{mclaughlin1924} 1924 --- see Section~\ref{rm_effect} for an
569: extensive discussion), by which the stellar absorption lines display
570: systematic deviation from their normal rotational broadening profile,
571: because some parts of the stellar surface are being eclipsed. As
572: Section~\ref{V1143_RM} points out, such observations were recently
573: obtained for V1143 Cyg (Albrecht \etal\ 2007). It would be of great
574: interest to confirm Shakura's idea, particularly because of its
575: implications for binary formation, as well as because competing ideas
576: to solve the mystery of DI Her have been discussed in the literature
577: (Claret \nocite{claret1998}1998).
578:
579: %==========================
580: %
581: \section{Circularization}
582: %
583: %==========================
584:
585: We move now
586: to consider the observational evidence for long-term dissipative
587: processes taking place in short-period binaries. This section will
588: consider the evidence for circularization, while the next two sections
589: will review the evidence for synchronization and alignment.
590:
591: It is not easy to find observational evidence for the circularization
592: processes, as we are not able to follow any binary that goes through
593: the long-term circularization. Instead, we can only observe circular
594: binaries that we {\it guess} were formed with eccentric
595: orbits. However, like every retrospect line of evidence, such
596: observations are open to other interpretations, and the conclusions are
597: not unquestionable.
598:
599: On the theoretical side, the calculations of the circularization
600: processes and their effects are more complicated than those of the
601: ellipsoidal modulation and the precession period. This is so because
602: in the latter two processes one assumes an immediate stellar response
603: to the tidal forces, while the derivation of circularization timescale
604: involves a response {\it lag} between the tidal forces and the stellar
605: shape (e.g., Zahn 1966). As we shall see, this entails serious
606: disagreement between the theoreticians, which makes the observational
607: evidence even more important.
608:
609: To first approximation, the tidal interaction between the two
610: components of a binary reduces the orbital eccentricity $e$ such that
611: \begin{equation}
612: {de\over{dt}}=-Ce \, ,
613: \end{equation}
614: where the factor $C$ depends on the orbital separation, the internal structure of the two stars and
615: their rotation. As $C$ varies very slowly, usually on the timescale of
616: the stellar lifetime, we can assume in many cases that $C$ is
617: constant. Therefore, the eccentricity decays exponentially, and we can
618: define the circularization timescale $\tau_{circ}$ with the equation:
619:
620: \begin{equation}
621: -{d\ln e\over{dt}}={1\over{\tau_{circ}}}
622: \label{eq_circ}
623: \end{equation}
624: %
625: (Zahn 1975). This equation implies that within a few $\tau_{circ}$
626: the binary motion assumes a nearly circular orbit, independently of its
627: primordial eccentricity. We therefore expect any binary older than
628: three or four times its $\tau_{circ}$ to become circular.
629:
630: As shown by the seminal work of Zahn (1975), the circularization
631: timescale is very sensitive to the orbital semi-major axis $a$:
632:
633: \begin{equation}
634: \tau_{circ} \propto \left\{
635: \begin{array}{c c}
636: %\left(\frac{a}{R_1}\right)^{8}
637: \left(a/R_1\right)^{8}
638: & \ \ \mbox{for stars with } convective \mbox{ envelopes} \\
639: %\left(\frac{a}{R_1}\right)^{13/2}
640: \ \ \ \left(a/R_1\right)^{13/2}
641: & \mbox{for stars with } radiative \mbox{ envelopes} \\
642: \end{array} \right.
643: \label{circularization_eq}
644: \end{equation}
645: %
646: where $R_1$ is the radius of the primary and $a$ is the binary
647: semi-major axis, assuming the tidal dissipation in the secondary can be
648: neglected. Determination of the eccentricities of short-period
649: binaries, for which $\tau_{circ}$ is necessarily short, can therefore
650: support the circularization idea observationally, {\it if} we find
651: that all or at least most of the short-period binaries have circular
652: orbits. Furthermore, considering a sample of {\it coeval} binaries,
653: the eccentricity as a function of the binary period can tell us how
654: the relevant $\tau_{circ}$ depends on the binary separation and the
655: stellar structure.
656:
657: \subsection{The SB9 sample of spectroscopic binaries}
658: %----------------------------------------------------
659:
660: One indirect evidence for long-term circularization processes can
661: be obtained by considering the eccentricities of all known
662: spectroscopic binaries as a function of their orbital period. To do
663: that we use the 2751 systems compiled by the official IAU catalog of
664: {\it spectroscopic} binaries --- SB9 (Pourbaix \etal\
665: \nocite{pourbaix2005}2005), as of August 2007. The sample of the
666: catalog is extremely inhomogeneous and includes stars of different
667: spectral types and ages. In addition, the binaries compiled
668: were discovered by different surveys with different observational
669: instruments, rendering their discovery thresholds
670: different as well. Furthermore, the compilation of the catalog is in progress,
671: and therefore the catalog is not yet complete. Nevertheless, the catalog
672: presents the largest sample of known spectroscopic binaries, and
673: consequently we use it as a valuable source to study some features of
674: the population of binaries in general, and short-period ones in particular.
675:
676: Figure~\ref{e_log_p} shows the eccentricity of the SB9 binaries as a
677: function of their period. Koch \& Hrivnak (\nocite{koch1981}1981)
678: already plotted a similar plot, using a much smaller sample that
679: included only late-type spectroscopic and {\it eclipsing} binaries
680: known at that time. Similarly to the results of Koch \& Hrivnak,
681: Figure~\ref{e_log_p} clearly shows that all binaries in the catalog
682: with periods shorter than 0.35 days, with the exception of one binary,
683: have circular orbits. In addition, all the binaries with longer
684: periods except two, have their eccentricities below an 'upper
685: envelope' that starts at eccentricity zero and climbs up to an
686: eccentricity of 0.98 when moving from 0.35 to about 350 days. Just
687: below the upper envelope the space diagram is somewhat sparse, and
688: only below another stripe does the diagram density become high.
689:
690: Somewhat arbitrarily, we chose to present the upper envelope and the
691: sparse stripe by a function of the form
692:
693: \begin{equation}
694: f=E-A \exp(-(p B)^c) \ .
695: \end{equation}
696: %
697: The values of the parameters of the upper function are $E=0.98$,
698: $A=3.25$, $B=6.3$ and $C=0.23$, and those of the lower function are
699: $A=3.5$ and $B=3$.
700:
701:
702: \begin{figure}
703: \includegraphics[width=13cm]{SB9.eps}
704: \caption{The eccentricity of the spectroscopic binaries as a function
705: of the orbital period. All 2751 binaries of the SB9 catalog are
706: included. The red dashed line is the upper envelope, with an
707: equation $f=E-A \exp(-(p B)^c)$, where $E=0.98$, $A=3.25$, $B=6.3$
708: and $C=0.23$. The green line is the line under which one can find most
709: binaries, with $A=3.5$ and $B=3$}
710: \label{e_log_p}
711: \end{figure}
712:
713: One could argue that the paucity of eccentric short-period
714: binaries above the upper envelope is the outcome of binary formation
715: processes. However, the natural explanation for these two features --- the
716: circularized binaries and the existence of the envelope --- is that they
717: are the result of tidal circularization, which is substantially weaker
718: in long-period binaries. This approach assumes that binaries
719: were formed above the upper envelope in the diagram, but the
720: circularization processes pushed them down.
721:
722: If this is indeed the case, we still have to explain the existence of three
723: systems high above the upper envelope: the cataclysmic variable WZ Sge,
724: with a period of $0.057$ days and an eccentricity of $0.17$ (Walker \&
725: Bell \nocite{walker1980}1980), the binary pulsar PSR 1913+16, with a
726: period of $0.32$ days and an eccentricity of $0.62$ (Hulse \& Taylor
727: \nocite{hulse1975}1975) and HD 142315 (Levato \etal\
728: \nocite{levato1987}1987), a binary with a B8V primary, a period of
729: $1.264$ days and an eccentricity of $0.61\pm0.2$.
730:
731: In fact, these three cases might not be a problem for the theory of
732: circularization:
733:
734: \begin{itemize}
735:
736: \item
737: The eccentricity of WZ Sge might not be real. The derivation of the
738: radial velocities of WZ Sge was difficult, as the spectra were taken
739: during the 1978 outburst, and the observed lines probably came from
740: the outburst material around the white dwarf. In addition, the
741: velocities derived from the He I lines are consistent with zero
742: eccentricity (Walker \& Bell 1980).
743:
744: \item
745:
746: The binary pulsar is composed of two neutron stars with radii as small
747: as 1 km (Hulse \& Taylor 1975). Therefore the circularization
748: processes in that system are not in action at all.
749:
750: \item
751:
752: The eccentricity of HD142315 is in fact $0.62\pm 0.20$ (Levato \etal\
753: 1987), so the eccentricity is above the upper envelope by only one
754: sigma.
755: \end{itemize}
756:
757: Figure~\ref{e_log_p} can not be used for a quantitative study of the
758: circularization processes, especially because of the lack of homogeneity and
759: the incompleteness of the SB9 catalog. For such a study we will
760: consider more homogeneous and coeval samples of stars below. Before
761: diving into the quantitative consideration, we present in the next
762: subsection two systems that show almost without any doubt an evidence
763: for long-term circularization. We can acquire such an evidence only for
764: circular binaries that we have strong reason to believe had
765: eccentric orbits in the past.
766:
767: %--------------------------
768: \subsection{X-ray binaries}
769: %--------------------------
770:
771: One of the first observational evidences for the long-term
772: circularization processes taken place in short-period binaries was the
773: discovery of the circular orbits of the first two close X-ray binaries
774: --- Cen X-3 (Schreier \etal\ 1972\nocite{sc72}) and Her X-1
775: (Tananbaum \etal\ 1972\nocite{tan72}), detected by Uhuru, the first
776: X-ray satellite. These two X-ray pulsars were found in binaries with
777: periods of 1.7 and 2.09 days, respectively, and with extremely small
778: eccentricities. Later analysis of the data of Cen X-3 (Fabbiano \&
779: Schreier \nocite{fabsc77} 1977) indicated that its eccentricity is
780: $0.0008 \pm 0.0001$ (see also a review by Bildsten \etal\
781: \nocite{bildsten1997}1997). Such precise measurement of the
782: eccentricity was possible only because of the rapid X-ray pulsation
783: detected by the satellite, which enabled the observers to follow
784: the orbit of the X-ray pulsar in great detail. The X-ray pulsations come
785: from a rapidly rotating neutron star (Davidson \& Ostriker
786: 1973\nocite{davidson1973}), formed by a supernova explosion that very
787: probably produced substantial changes in the binary orbit, inducing a large
788: orbital eccentricity (Wheeler \etal\ 1974\nocite{wheeler74}; Chevalier
789: 1975\nocite{che75}). Therefore, the observed minute eccentricity had
790: to be associated with tidal circularization that was in action since
791: the supernova explosion (Lecar \etal \nocite{lecar76} 1976).
792:
793: In fact, the short periods of Cen X-3 and Her X-1 might themselves be
794: another indication for tidal evolution taken place in these systems
795: during their post-supernova phase. The reason for this is that the
796: commonly accepted paradigm assumes that the neutron star in these
797: systems was formed by a supernova type II, which occurred in the last
798: stages of evolution of the original primary star, when it reached the
799: supergiant phase (e.g., McCluskey \& Kondo
800: \nocite{mccluskey1971}1971). The present binary system, with a
801: separation of order of 0.1 AU, certainly could not accommodate a
802: supergiant. Therefore, the binary orbits of Cen X-3 and Her X-1 have
803: probably shrunk because of tidal evolution of the system (e.g.,
804: Sutantyo 1976\nocite{sutantyo1976}), which led to transfer of angular
805: momentum from the orbital motion to the present main-sequence star.
806: Similar ideas were proposed for other X-ray binaries (e.g., Verbunt
807: \nocite{verbunt1994}1994; van Kerkwijk \etal\ \nocite{van2000}2000;
808: Janssen \& van Kerkwijk \nocite{janssen2005}2005).
809:
810: %-----------------------------------------------------
811: \subsection{Evidence for circularization in early-type
812: binaries}
813: \label{subsection_early_type}
814: %-----------------------------------------------------
815:
816: The circularization occurring in Cen X-3 brings evidence for
817: tidal interaction in a binary with an early-type star, as the optical
818: counterpart, V779 Cen (Krzeminski \nocite{krzeminski1974}1974), is of
819: O6--7 II--III type (Ash \etal\ \nocite{ash1999}1999), with a mass of
820: about 20 $M_{\odot}$. This is particularly important, as such an early-type
821: star has a radiative envelope, whereas most of the stars in the SB9
822: catalog are cool stars with convective envelopes. Tidal dissipation
823: is expected to be very efficient in convective envelopes, where the
824: viscosity is high due to turbulent eddies (Zahn 1975). On the other
825: hand, in early-type stars with radiative envelopes the dissipative
826: processes are assumed to be radiative damping on the dynamical tide
827: (Zahn 1977). Therefore, the circular orbit of Cen X-3 is an evidence
828: for tidal interaction of a different kind from that which occurs in
829: most of the binaries in the SB9 catalog.
830:
831: The number of spectroscopic binaries with early-type stars is small
832: because early-type stars have few spectral lines in their spectra,
833: which makes it difficult to derive their radial velocities and
834: discover their binarity. Additionally, very hot stars are rare in the
835: solar neighbourhood, and therefore not many early-type stars are known
836: as spectroscopic binaries.
837:
838: In order to study the circularization processes in a larger sample of
839: early-type stars, a seminal work of Giuricin \etal\
840: (\nocite{giuricin1984}1984) considered about 200 spectroscopic and
841: {\it eclipsing} binaries with known O, B and A primaries. They
842: constructed their sample from the seventh catalog of spectroscopic
843: binaries (Batten \etal\ \nocite{batten1978}1978) and from the list of
844: eclipsing binaries of Cester \etal\ (\nocite{cester1979}1979) and Wood
845: \etal\ (\nocite{wood1980}1980). They plotted the eccentricity as a
846: function of the orbital period, and showed that the two features of
847: Koch \& Hrivnak (1981) do appear, although with different
848: parameters. Excepting only a few cases, all binaries with periods shorter
849: than $2$ days are circular, and there is an 'upper envelope' that goes
850: up to an eccentricity of $0.6$.
851:
852: Giuricin \etal\ (1984) used the fact that the lightcurves of eclipsing
853: binaries yield one additional parameter --- the fractional radius,
854: which is {\it not} available for spectroscopic binaries. This
855: parameter gives the stellar radius as a fraction of the semi-major
856: axis of the binary orbit. In fact, in early-type stars the
857: circularization timescale strongly depends on the fractional radius
858: (see Equation~\ref{circularization_eq}, taken from Zahn
859: 1975). Therefore, Giuricin \etal\ (1984) plotted the eccentricity as a
860: function of fractional radius for their sample, using a less accurate
861: spectral estimation for the spectroscopic binaries. They have
862: discovered that all binaries with fractional radius {\it larger} than
863: 0.3 are circular, and there is an eccentricity 'upper envelope' that
864: rises from zero to 0.8 when {\it reducing} the fractional radius from
865: 0.3 to 0.05. Similar results were found in modern photometric data of
866: early-type binaries in the LMC, as discussed in
867: Section~\ref{subsection_circ_LMC} (see Figure~\ref{LMC} for a similar
868: diagram).
869:
870: The most natural interpretation of these findings is that systems with
871: fractional primary radius larger than 0.3 have been circularized
872: during the lifetime of the systems. Note, however, that Giuricin
873: \etal\ found a few binaries with primary radius larger than 0.3 that
874: displayed small but significant eccentricities. They attributed these
875: eccentricities to unseen third distant companions, an effect suggested
876: by Mazeh \& Shaham (1979; see detailed discussion in
877: Section~\ref{subsection_MS}).
878:
879: %------------------------------------------------------------
880: \subsection{Circularization in coeval sample of binaries}
881: %-------------------------------------------------------------
882:
883: The first evidence for circularization of coeval binaries was pointed
884: out by the seminal work of Mayor \& Mermilliod (1984, hereafter MM84).
885: \nocite{mm84} They have plotted the eccentricities versus period of 33
886: solar-type main-sequence spectroscopic binaries known at that time in
887: a few open clusters, including the Hyades, Pleiades, Praesepe and
888: Coma. MM84 found that the eccentricities of these binaries display a
889: distinct change at a period of 5.7 days. All seven binaries with
890: periods shorter than this critical period have circular orbits, while
891: binaries with periods longer than 5.7 days have orbits with
892: significant eccentricities.
893:
894: \nocite{zahn66} \nocite{zahn75} \nocite{zahn77}
895:
896: The transition period between circular and eccentric binaries can
897: be directly observed for a given sample of coeval binaries, though
898: with considerable observational effort. In principle, one can follow
899: the radial velocities of a large coeval sample of similar stars,
900: identify the binaries, determine their orbital parameters, the period
901: and eccentricity in particular, and find the transition period. Since
902: the seminal work of MM84, a few radial-velocity surveys have detected
903: the transition-period effect. This was done for binaries with G-type
904: primaries found in the solar neighbourhood (Duquennoy \& Mayor
905: \nocite{dm91} 1991), in the open clusters M67 (Latham \etal
906: \nocite{latham92a}\nocite{latham92b} 1992a,b), NGC 188 (Mathieu \etal
907: \nocite{mathieu2004} 2004) and recently M35 (Meibom \& Mathieu
908: \nocite{meibom05} 2005, hereafter MM05), in PMS (Melo \etal\
909: \nocite{melo01} 2001) and in the halo and field stars (Latham \etal\
910: \nocite{latham02} 2002). These studies derived the transition periods
911: for samples with different ages, starting with the PMS binaries, with
912: an age close to zero, and moving up to the age of the Galactic halo,
913: which is probably older than 10 Gyrs. The expectation was that
914: transition periods lengthen in older samples, for which the
915: circularization processes have been in action for a longer period of
916: time. Therefore, the variation of the transition period as a function
917: of the sample age can be used as indirect evidence of the
918: circularization processes.
919:
920: Following this line of reasoning, Mathieu \& Mazeh (1988, hereafter
921: MM88) \nocite{mm88} even suggested that the cutoff period can be used
922: to estimate the sample age. However, it turned out that this idea can
923: not be easily applied, as the theoreticians could not agree upon the
924: details of the theory of circularization, as discussed below.
925:
926:
927: \subsubsection{The transition period}
928: \label{subsub_transition}
929: %------------------------------------
930:
931: Since the early days of circularization studies, the definition of the
932: observed transition period between the circular and the eccentric
933: orbits has been unclear. Is the transition period the longest period
934: for which a circular binary was found, or is it the shortest period
935: with an eccentric orbit? For example, Figure~2 shows the
936: eccentricities of 18 binaries with short periods that were found by
937: the seminal work of MM05 in the open cluster M35, out of their sample
938: of 32 binaries. We somewhat arbitrarily define as circular binary any
939: orbit with a derived eccentricity smaller than 0.05. This is indicated
940: by the horizontal dotted line in the figure. According to this
941: definition, the longest period with circular orbit is binary no.~4037,
942: the one with the period of 16.5 days. On the other hand, the shortest
943: period with an eccentric orbit is that of binary no.~0422, with 8.2
944: days (MM05). The two periods, delineated in the figure with vertical
945: dashed lines, bound the possible range of the transition period. Such
946: ambiguities exist for most observed samples.
947:
948:
949: \begin{figure}
950: \includegraphics[width=13cm]{MM05_range.eps}
951: %\includegraphics{MM05_range.eps}
952: %\qquad
953: \caption{The transition range between circular and eccentric binaries
954: in M35, after MM05. All binaries with eccentricities below the
955: horizontal dotted line are considered circular. The transition range
956: is between the shortest period with an eccentric orbit and the
957: longest period of circular binary}
958: \label{range_fig}
959: \end{figure}
960:
961: In fact, the width of the transition range is not surprising. Mathieu
962: and Mazeh (1988) already listed several factors that can blur the
963: transition period between circular and eccentric binaries, even for a
964: sample of binaries in which all systems have exactly the same primary
965: mass. A major factor is the secondary mass, which might be different
966: for the various binaries in the sample. After all, the gravitational
967: attraction of the secondary is the source of the tidal force exerted
968: on the primary, and therefore the tidal circularization timescale of
969: the binary strongly depends on the binary mass ratio $q=M_2/M_1$,
970: where $M_1$ is the primary mass and $M_2$ is the secondary mass. For
971: a given period and primary star, Mathieu \& Mazeh (1988) got:
972:
973: \begin{equation}
974: \tau_{circ}\propto q^{2/3}(1+{1\over{q}})^{5/3} \ .
975: \end{equation}
976: %
977: Therefore, the secondary mass can lengthen the timescale of the
978: circularization processes taken place inside the primary by a factor
979: of 4, when moving from $q=1$ to $q=0.1$. Another factor to take into
980: account is the circularization processes taking place in the secondary
981: itself. For equal masses, the secondary contributes to the
982: circularization processes as much as the primary. But when we move to
983: a binary with a mass ratio of, say, 0.5, the circularization processes
984: in the secondary are negligible, because of the small stellar
985: radius. Considering the two effects together, the secondary mass can
986: change the binary circularization timescale by a factor of 8, which
987: corresponds to a factor of 1.5 in circularization period, if we assume
988: that $\tau_{circ}\propto P^{16/3}$ (Zahn 1977).
989:
990: As radial-velocity observations typically are not capable of
991: determining the nature of the secondary star in single-lined
992: spectroscopic binaries, the spread of mass ratios in any sample of
993: spectroscopic binaries induces a spread of the typical circularization
994: timescales.
995:
996: Another factor that can change the time needed for a binary to reach
997: circularization is its primordial eccentricity. For example, a binary
998: with initial eccentricity of 0.75 needs twice the amount of time needed by
999: a binary with initial eccentricity of 0.2 to get to an eccentricity
1000: of 0.05. Therefore, a single circular binary with a long period could
1001: have been the result of a low initial eccentricity, and not
1002: necessarily because of the efficiency of the circularization
1003: processes.
1004:
1005: In addition, samples of spectroscopic binaries are not uniform, and
1006: the primary masses are not always the same. A small change in the
1007: primary mass can dramatically change its interaction with the varying
1008: tidal pull of its secondary.
1009:
1010: Lastly, triple systems can show the Mazeh \& Shaham \nocite{ms79}
1011: (1979) effect, by which a third distant star can pump eccentricity
1012: into the binary orbit (see Section~\ref{triple} below), even if the
1013: binary system starts with a circular orbit. Mayor \& Mazeh
1014: \nocite{mm87} (1987) have shown that many of the spectroscopic
1015: binaries might have third faint companions (see a more detailed
1016: discussion in Section~\ref{triple}). Furthermore, Tokovinin \etal\
1017: (\nocite{tokovinin2006}2006; see also Tokovinin
1018: 2004\nocite{tokovinin2004}) claimed that more than 90\% of the
1019: short-period binaries, with periods shorter than 3 days, have at least
1020: one additional companion. Therefore, the Mazeh \& Shaham effect can be
1021: quite frequent (see Section~\ref{subsection_MS} for a detailed account
1022: of the observational evidence for this effect), causing (small)
1023: eccentricities to appear in short-period binaries.
1024:
1025: Because of all these considerations, the blur of the transition
1026: between circular and eccentric periods in any sample is
1027: inevitable. Therefore, a transit {\it range} between the circular and
1028: the eccentric binary is not that surprising. Nevertheless, some
1029: researchers used the term `cutoff period' to denote the longest period
1030: of a circular orbit in the sample; see, for example, Melo \etal\
1031: (2001) who suggested that the transition period of PMS stars is 7.56
1032: days, based on one circular binary. Such an approach should be
1033: modified, given the different factors that might blur the transition.
1034:
1035: %--------------------------------------------------------
1036: \subsubsection{Derivation of the transition period ---
1037: Meibom \& Mathieu approach}
1038: %--------------------------------------------------------
1039:
1040: A major step in the discussion of the transition period was recently
1041: achieved by Meibom and Mathieu (2005), who came up with a new
1042: algorithm to derive a transition period for a sample of spectroscopic
1043: binaries. Instead of expecting the sample to be drastically divided
1044: between circularized and eccentric binaries, MM05 defined a new
1045: expected `averaged' eccentricity, which presents a smooth transition
1046: between the two parts of the sample. Based on extensive simulations
1047: of populations of binaries, they concluded that the final `averaged'
1048: eccentricity as a function of their period is of the shape
1049:
1050: \begin{equation}
1051: e(P)= \left\{
1052: \begin{array}{c c}
1053: 0 & \mbox{if} \ \ P \le P_{tran}\\
1054: \alpha(1-e^{\beta(P_{tran}-P)})^{\gamma} & \mbox{if} \ \ P >
1055: P_{tran} \ .\\
1056: \end{array} \right.
1057: \end{equation}
1058: %
1059: In this distribution the eccentricity assumed zero value up to a
1060: 'transition' period, $P_{tran}$, and then climbed exponentially to
1061: the averaged eccentricity of $\alpha$. The parameters $\beta$ and
1062: $\gamma$ controlled the steepness of the rise of the averaged
1063: eccentricity. Based on their simulation they adopted the values of
1064: $\beta$ and $\gamma$ to be $0.14$ and $1.0$, respectively. MM05 found
1065: that the mean value of the eccentricity of the long-period binaries in
1066: all samples they considered was $0.35$, and therefore they adopted
1067: this value for $\alpha$.
1068:
1069: To derive the transition period of a sample, MM05 suggested finding
1070: the $P_{tran}$ parameter of their function that best fits the sample
1071: periods and eccentricities. The advantage of their approach was that
1072: $P_{tran}$ did not depend on one binary alone, with the longest
1073: circular period or the shortest eccentric period, but, instead took
1074: the whole set of eccentricities and periods in the transition range
1075: into consideration.
1076:
1077: In the course of their thorough study, MM05 analysed eight available
1078: samples of coeval spectroscopic binaries with their approach, and
1079: derived transition periods for each of these samples. Their results
1080: are given in Table~\ref{table_circ}, where we round the uncertainties
1081: of the periods.
1082:
1083: %------------------------------------------------------------------
1084:
1085: \begin{table}
1086: \label{tab:circularization}
1087: \caption{Circularization Periods for different coeval samples derived
1088: by MM05}
1089: \vskip 1pc
1090: \begin{tabular}{lcr}
1091: Binary Population & $\log Age$ & Transition Period \\
1092: & (Gyrs) & (days)\ \, \ \ \ \\
1093: \hline
1094: PMS & -2.5 & $7.1\pm 1.2$ \ \ \\
1095: Pleiades & -1.0 & $7.2\pm 1.8$ \ \ \\
1096: M35 & -0.8 & $10.2\pm 1.2$ \ \ \\
1097: Hyades/Praesepe & -0.2 & $3.2\pm 1.2$ \ \ \\
1098: M67 & 0.6 & $12.1\pm 1.2$ \ \ \\
1099: NGC188 & 0.8 & $14.5\pm 1.8$ \ \ \\
1100: Nearby G primaries & 0.95 & $10.3\pm 2.3$ \ \ \\
1101: Halo & 1.00 & $15.6\pm 2.8$ \ \ \\
1102: \hline
1103: \end{tabular}
1104: \label{table_circ}
1105: \end{table}
1106: %--------------------------------------------------------------
1107:
1108: Each of the eight samples has similar binaries and is comprised of
1109: main-sequence G-type primaries. It should therefore be interesting
1110: to compare the derived transition period of the samples as a function
1111: of their age. We followed MM05 and plotted their result in
1112: Figure~\ref{circ_per_age}. Based on the discussion above
1113: with regard to the factors that can blur the transition between the
1114: circular and eccentric binaries, we did not use MM05 errors. Instead,
1115: we assign somewhat arbitrarily to each transition period an error of
1116: 50\% up and 33\% down, so the up and down errors are the same in
1117: logarithmic scale.
1118:
1119: \begin{figure}
1120: \includegraphics[width=13cm]{TransPerVsAge.eps}
1121: \caption{The transition period for 8 coeval systems as a function of
1122: their age, after MM05. Contrary to MM05, we assigned to each period an
1123: error of factor of 50\% (see text). The error bar of the transition
1124: period of the Hyades/praesepe sample was extended by a dotted line
1125: (see text).}
1126: \label{circ_per_age}
1127: \end{figure}
1128:
1129: Two features are emerging from the thorough analysis of MM05, as
1130: presented in Table~\ref{table_circ} and
1131: Figure~\ref{circ_per_age}. First, for seven out of the eight coeval
1132: samples discussed by MM05, the transition period is between 6 and 16
1133: days. Second, for the same seven samples, the transition period seems
1134: to be a monotonic increasing function of the sample age. The only
1135: outlier which is not consistent with these two features is the sample
1136: composed of the binaries found in the Hyades and Praesepe, two
1137: clusters of the age of 650 Myrs. The MM05 analysis of this sample
1138: yields a transition period of 3.2 days, but this transition period is
1139: strongly affected by two binaries. One binary is KW 181 (Mermilliod \& Mayor
1140: \nocite{mermi1999}1999), with a period of 5.9 days and an eccentricity
1141: of 0.36, and the other is vB 121, with a period of 5.75 days and an
1142: eccentricity of 0.35 (Griffin \& Gunn
1143: \nocite{griffin1978}1978). Without these two systems, the algorithm of
1144: MM05 would assign a transition period of about 7 days to the Hyades
1145: and Praesepe sample. To mark this fact, we lengthened arbitrarily the
1146: error bar of the symbol that presents the Hyades and Praesepe sample
1147: in the figure. We therefore conclude that the accumulated
1148: observational data are consistent with the assumption that most, if not
1149: all, transition periods of coeval samples are longer than 6 days, and
1150: there seems to be a tendency for longer transition periods in older
1151: clusters.
1152:
1153: %-------------------------------------------------------------
1154: \subsection{Confrontation of the theory with the observations}
1155: \label{confrontation}
1156: %-------------------------------------------------------------
1157:
1158: The analysis of MM05 makes it possible to compare the derived
1159: transition periods with the theory. Three theoretical models were
1160: suggested for the tidal dissipation acting in close binary systems:
1161:
1162: \begin{itemize}
1163:
1164: \item The theory of equilibrium tide, which assumes that the stellar
1165: tidal bulge, induced by its companion, lags at some angle after the orbital motion
1166: of the companion. The lag, which is due to dissipative
1167: processes inside the star, enables a transfer of energy and angular
1168: momentum from the orbital motion to the stellar rotation (Zahn
1169: 1966). This mechanism is mostly effective in stars with convective
1170: envelopes, and therefore the equilibrium tide was applied only to
1171: binaries with late-type stars (Zahn 1977).
1172:
1173: \item The theory of dynamical tides, which assumes that the tidal
1174: interaction is acting through the damping of gravity modes inside
1175: the stellar radiative zone (Zahn 1970; 1975; Savonije \& Papaloizou
1176: \nocite{savpap83} \nocite{savpap84} 1983, 1984). It was assumed that
1177: this mechanism is only effective in stellar radiative zones, and
1178: therefore was applied first to high-mass stars with radiative
1179: envelopes exclusively (Zahn 1977). However, in the last decade a few studies
1180: (e.g., Terquem \etal \nocite{ter98} 1998; Goodman \& Dickson
1181: \nocite{goodman1998}1998; Witte \& Savonije 2002; Savonije \& Witte
1182: 2002) \nocite{witte2002}\nocite{savonije2002} have shown that this
1183: mechanism can be effective in radiative cores of late-type stars
1184: as well. Very recently tidal interaction with inertial waves in
1185: convective envelopes of solar-type stars was also considered
1186: as a source for orbital evolution (Ogilvie \& Lin 2007; see also
1187: Ivanov \& Papaloizou 2007).
1188:
1189: \item The theory of hydrodynamic flows, which assumes that the tidal forces
1190: of the companion induce large-scale meridional flows of the stellar
1191: surface, which in turn exert torques on the companion (Tassoul
1192: \nocite{tass95} 1995). This theory was questioned by Rieutord \&
1193: Zahn \nocite{rz97} (1997) and was later defended by Tassoul \&
1194: Tassoul \nocite{tasstass97}(1997).
1195:
1196: \end{itemize}
1197:
1198: Each of these theories should account for the overall result presented
1199: above which gives {\it an absolute scaling} of the tidal
1200: circularization processes in a binary with G-type primary. The
1201: observationally derived transition periods for the different coeval
1202: samples imply that for a binary with a period of about 10 days the
1203: circularization timescale is of the order of $10^9$ yrs (=Gyr). The
1204: theories should be able to account for the possible mild increase of
1205: the transition period as a function of sample age, presented already
1206: in MM88, and discussed by MM05 in particular.
1207:
1208: The plot of MM05 is of great interest because the slope of the period
1209: increase can, in principle, distinguish between the different
1210: theories. This is particularly true if we consider several updated
1211: versions of the equilibrium tide theory. This theory was restudied by
1212: Zahn (\nocite{zahn1989}1989), who specifically considered the reduction
1213: of the turbulent viscosity when the tidal period becomes shorter than
1214: the convective turnover time. He concluded that the circularization
1215: timescale should be $\tau_{circ}\propto P^{16/3}$. In a follow-up paper,
1216: Zahn and Bouchet (1989) concluded that all the circularization of
1217: G-type binaries must occur during the pre-main-sequence (=PMS)
1218: phase, when the stars are substantially larger than they will be during their
1219: main-sequence phase. They further claimed that the circularization during
1220: the main-sequence phase is negligible, and they therefore predicted that the
1221: circularization period would be between 7.2 and 8.5 days for all
1222: samples, regardless of their respective ages.
1223:
1224: Not all studies agreed with Zahn (1989) approach to the reduction of
1225: the turbulent viscosity. Goldman \& Mazeh (\nocite{goldman1991}1991),
1226: for example, proposed a modified approach that resulted in
1227: $\tau_{circ}\propto P^{10/3}$. They also pointed out the evidence
1228: that was available at that time for the variation of the transition
1229: period with the sample age. Goldreich \& Nicholson
1230: (\nocite{goldreich1989}1989) and Goodman \& Oh (\nocite{goodoh97}1997)
1231: studies yielded power dependence which is in between that of Goldman
1232: \& Mazeh and that of Zahn (1989).
1233:
1234: It seems that the plot of MM05 (Figure~\ref{circ_per_age}) indicates
1235: that the circularization processes during the PMS phase do circularize
1236: binaries up to about 6--8 days, as Zahn and Bouchet suggested. It also
1237: suggests that transition periods probably become longer in older
1238: samples, in contrast to the other claim of Zahn \& Bouchet (Goldman \&
1239: Mazeh 1991; Mathieu \etal\ \nocite{mathieu1992}1992), although the
1240: slope of the figure is not conclusive enough to distinguish between
1241: the different theories. If this is true, we must find out why the
1242: theory of equilibrium tide, when applied to {\it main-sequence} stars,
1243: yields a circularization absolute efficiency which is too low to
1244: explain the observations (e.g., Goodman \& Oh 1997; Terquem \etal\
1245: 1998; Goldman \& Mazeh 1991; Sasselov \nocite{sasselov2003}2003). We
1246: therefore have to close this discussion with the banal statement that
1247: we need more data for a definite conclusion.
1248:
1249: %-------------------------------------------------------------------
1250: \subsection{Circularization of binaries with giant star component}
1251: \label{giant}
1252: %---------------------------------------------------------------------
1253:
1254: An additional line of evidence for tidal circularization was studied
1255: by Verbunt \& Phinney (\nocite{verbunt1995}1995, hereafter VP95), who
1256: considered the eccentricity as a function of orbital period of
1257: binaries containing a giant component found in open clusters. VP95
1258: noted that giants can serve as the simplest test cases for the
1259: circularization theory, because the orbital period of such systems is
1260: much larger than the eddy timescale, and therefore we do not need to
1261: worry about the reduction efficiency of the viscosity. Furthermore,
1262: giants are fully convective and therefore avoid the problem of tides
1263: in the radiative zone. We note that the large radii of the giants
1264: render the circularization during PMS phase irrelevant, another
1265: advantage of studying binaries with giants. It is true that the giants
1266: have an history of fast expansion, which causes the circularization
1267: timescale to change rapidly. However, VP95 claimed that from the
1268: cluster age and turnoff mass one can reconstruct the history of the
1269: giants and integrate the expected theoretical circularization effect.
1270:
1271: The sample of VP95 included 28 binaries from 12 open clusters, taken
1272: from the works of Mermilliod \& Mayor (\nocite{mm89}\nocite{mm90}1989;
1273: 1990), Mermilliod \etal\ (\nocite{mermilliod1989}1989) and Mathieu
1274: \etal\ (\nocite{mathieu1990}1990). They found that except two systems,
1275: all binaries with periods shorter than 200 days have circular orbits.
1276: However, as the systems are with different ages and masses, the
1277: orbital period is not the best parameter to compare the observed
1278: eccentricity with. When VP95 considered instead the {\it expected}
1279: eccentricity reduction, based on the history of the giants derived
1280: from their mass and position on their H-R diagrams, they concluded
1281: that all are consistent with Zahn theory.
1282:
1283: More than a decade has passed since the seminal work of VP95, and many
1284: more giants have been discovered since. A new study of the binaries
1285: with giants in open clusters and in the field is highly due, so we can
1286: learn more about the circularization processes and derive their
1287: absolute scaling of efficiency.
1288:
1289: %-----------------------------------------------------------------------
1290: \subsection{Highly eccentric binaries and the theory of circularization}
1291: %-----------------------------------------------------------------------
1292:
1293: Some of the main-sequence spectroscopic binaries were found to have
1294: very high eccentricities. The five most eccentric binaries in SB9,
1295: with an eccentricity larger or equal to 0.95, are compiled in Table 2,
1296: which lists their orbital periods and eccentricities. The two most
1297: eccentric systems might be a challenge to the circularization theory,
1298: as their periastron distance is quite small, and therefore we would
1299: expect them to circularize, or at least to substantially reduce their
1300: eccentricities on a relatively short timescale (Duquennoy \etal\
1301: 1992).
1302: \begin{table}
1303: \label{tab:eccentric}
1304: \caption{The five most eccentric spectroscopic binaries in SB9. The
1305: table includes the period and eccentricity of each system and also the
1306: `periastron period', $P_{peri-dist}$ --- the orbital period of a
1307: circular orbit with a radius that equals the actual periastron
1308: distance.}
1309: \vskip 1pc
1310: \begin{tabular}{lrccl}
1311: Binary & $e$ \ \ \ & Orbital Period & $P_{peri-dist}$&Reference\\
1312: & & (days) & (days)\\
1313: \hline
1314: HD 2909 & 0.949 & 2128 & 25 & Mazeh \etal\ 1995 \\
1315: & $\pm$0.002 & &\\
1316: HD 165590 & 0.958 & 7400 & 64 & Batten \etal\ 1979 \\
1317: & $\pm$0.001 & &\\
1318: HD 123949 & 0.972 & 9200 & 43 & Udry \etal\ 1998 \\
1319: & $\pm$0.057 & &\\
1320: Gliese 586A & 0.9752 & 890 & 3.5 & Duquennoy \etal\ 1992 \\
1321: & $\pm$0.0003 & \\
1322: 41 Dra & 0.9754 & 1250 & 4.8 & Tokovinin \etal\ 2003 \\
1323: & $\pm$0.0001\\
1324: %1024,1627,1678,1531,1878
1325: \hline
1326: \end{tabular}
1327: \end{table}
1328: %\nocite{mazeh1995}
1329: \nocite{tokovinin2003}
1330: %\nocite{duq1996}
1331: \nocite{duq1992}
1332: \nocite{griffin1984}
1333:
1334: \nocite{udry1998} \nocite{batten1979} \nocite{mazeh1995}
1335:
1336: To estimate the effectiveness of the tidal forces at periastron, I
1337: introduce the `periastron period', $P_{peri-dist}$ --- the orbital
1338: period of a circular orbit that has the same orbital radius as
1339: the actual periastron distance. It is easy to show that
1340: \begin{equation}
1341: P_{peri-dist}=P(1-e)^{3/2} \ .
1342: \end{equation}
1343: %
1344: Note that this is {\it not} the period that presents the orbital
1345: angular velocity at periastron. Table 2 lists the corresponding
1346: periastron periods of the five systems.
1347:
1348: >From the table we see that two of the systems, with eccentricities
1349: larger than 0.97, indeed have short periastron distances, which are
1350: equivalent to orbital periods shorter than 5 days. If these systems
1351: are of the age of a few Gyrs, we would naively expect them to
1352: circularize, or at least to reduce their eccentricities during their
1353: main-sequence lifetime.
1354:
1355: Goldman \& Mazeh (\nocite{goldman1994}1994), who studied the tidal
1356: history of Gliese 586A, argued that for highly eccentric binaries the
1357: tidal shear, which drives the turbulent viscosity, changes near
1358: periastron on a timescale shorter than the convective
1359: timescale. Therefore the turbulent viscosity (e.g., Hut 1982a;b)
1360: \nocite{hut1982a}\nocite{hut1982b} is substantially reduced, as was
1361: argued in Section~\ref{confrontation}. With their recipe for
1362: the reduction of turbulence, Goldman \& Mazeh were able to suggest a tidal
1363: history of Gliese 586A that leads to the present parameters of the
1364: system. Obviously, similar arguments can be applied to the other
1365: systems.
1366:
1367: Goldman \& Mazeh even argued that the highly eccentric binaries have
1368: the potential for distinguishing between the different recipes for
1369: viscosity reduction. Unfortunately, very eccentric binaries are
1370: difficult to discover, as most of their radial-velocity variation is
1371: concentrated in a very small part of the orbital period (e.g., Griffin
1372: 1984). Therefore only very few highly eccentric binaries are
1373: known. Nevertheless, I suspect that many of these binaries exist, and
1374: when they are discovered we will be able to use them as test cases for
1375: tidal interaction.
1376:
1377: %=============================
1378: %
1379: \section{Synchronization} %
1380: \label{section_synchro}
1381: %
1382: %=============================
1383:
1384: As stated in the introduction, tidal forces in binaries also tend to
1385: synchronize the rotation of the two stars with the orbital motion. The
1386: theory of tidal evolution (Zahn \& Bouchet \nocite{zahnbou89}1989;
1387: Witte \& Savonije 2002) predicts that the synchronization timescale
1388: will be two or three orders of magnitudes shorter than the
1389: circularization timescale. This has to do with the fact that the
1390: angular momentum associated with stellar rotation is much smaller than
1391: the orbital angular momentum. Consequently, the angular momentum
1392: transferred between the orbital motion and the stellar rotation in
1393: order to reach synchronization is much smaller than that which is needed to
1394: attain circularization. This means that we can expect binaries to
1395: reach synchronization a long time before they reach
1396: circularization. Similarly to circularization, we should also expect
1397: coeval sample of binaries to show a transition period between
1398: synchronized and non-synchronized binaries. Such an observed
1399: transition can serve as another confirmation of the theory of tidal
1400: interaction.
1401:
1402: However, both the observations and the theory of stellar
1403: synchronization are more complicated than those of tidal
1404: circularization. Even the term `stellar rotation' is not well defined,
1405: as the notion of a star rotating as a rigid body is obviously an
1406: oversimplification. For example, differential rotation as a function
1407: of stellar latitude (e.g., Lyytinen \etal\ \nocite{lyytinen2002}2002;
1408: Croll \etal\ \nocite{croll2006}2006; Walker \etal\ \nocite{walker2007}
1409: 2007) must be taken into account when we define stellar
1410: rotation. Therefore the following section adopts an observational
1411: approach, according to which the stellar rotation is the observed
1412: rotation, and which disregards complications induced by differential
1413: rotation.
1414:
1415: The stellar rotational period can be obtained either by photometric or
1416: spectroscopic observations. Photometric monitoring can reveal stellar
1417: {\it periodic} variability, which presumably is caused by stellar spots
1418: that come into and go from the facing hemisphere of the star,
1419: as the star rotates on its axis (e.g., Stauffer \etal\
1420: \nocite{stauffer1987}1987). Stellar spectra, with high enough
1421: resolution and S/N ratio, yield line broadening which can be modelled
1422: to yield stellar rotational velocity. With the knowledge of the size
1423: of the stellar radius and inclination angle it is possible to derive
1424: the stellar rotation period (e.g., Stauffer \etal\
1425: \nocite{stauffer1984}1984). However, as stellar radii and inclinations
1426: are usually not well known, the spectroscopic approach in most cases
1427: induces a large uncertainty of the rotational period. On the other hand, in
1428: spectroscopic binaries the spectra have already been obtained, in
1429: contrast to the photometry that usually necessitates additional
1430: observational effort.
1431:
1432: The variability method is dependent upon the longevity of the spots.
1433: We assume that although stellar spots indeed form and dissolve, they
1434: still maintain some coherence with a timescale much longer than the
1435: stellar rotation period. If this is the case, then the spots can
1436: maintain a periodic brightness variability, modulated with the
1437: rotation period. However, as stellar spots form and dissolve, the
1438: amplitude and phase of the modulation might not be constant (see, for
1439: example, a discussion of the stellar spots of V1794 Cyg by Jetsu
1440: \etal\ \nocite{jetsu1999}1999). This can cause some spread of the
1441: frequency power of the modulation over neighbouring frequencies,
1442: rendering the detection of the periodic modulation more
1443: difficult. Therefore many measurements are needed in order to detect
1444: the rotational frequency, preferably distributed over many rotational
1445: periods.
1446:
1447: Another complication faced by the photometric approach is the
1448: existence of other sources of variability, coming either from the star
1449: itself, like stellar pulsation, or from the photometric modulation of
1450: its companion. The latter is specifically true for binaries with
1451: comparable brightness, where the secondary modulation can have an
1452: amplitude that is comparable with that of the primary. In addition, the
1453: ellipsoidal variability, with half the binary period, can also be
1454: present in the photometric data (see, for example, the analysis of the
1455: RS CVn-type stars RT And by Zhang \& Gu \nocite{zhang2007}2007 and SZ
1456: Psc by Eaton \& Henry \nocite{eaton2007}2007). Therefore, the
1457: identification of stellar rotation by a periodic modulation
1458: should be done with extra caution, and only after other sources of
1459: periodic modulation have been excluded.
1460:
1461: On the theoretical side, whereas for the eccentricity evolution only
1462: the tidal interaction is in action, stellar rotation has its own
1463: evolution even in single stars. Contraction of PMS stars that are
1464: approaching the main-sequence phase and their interaction with their
1465: accretion disc (e.g., K\"onigl \nocite{konigl1991}1991; Shu \etal\
1466: \nocite{shu1994}1994) can change the stellar rotation substantially
1467: (for observational evidence see, for example, Choi \& Herbst
1468: \nocite{choi1996}1996 and Stassun \etal\ \nocite{sta99}1999). In
1469: addition, the rotational angular momentum itself can evolve during the
1470: main-sequence phase (e.g., Baliunas \etal\ \nocite{baliunas1995}1995), via
1471: magnetic coupling with the stellar wind, for example.
1472:
1473: Because of the difficulties of the theory and observation, less
1474: attention has been paid by the binary community to the study of tidal
1475: synchronization. Two exceptions are the works of Abt \etal\
1476: (\nocite{abt2002}2002) and Abt \& Boonyarak (\nocite{abt2004}2004),
1477: who studied samples of B and A stars known to reside in binaries. Abt
1478: \etal\ (2002) derived the rotational period of their samples from the
1479: stellar line broadening, and concluded that all B-type stars with
1480: orbital periods shorter than 2.4 days are synchronized, while binaries
1481: with periods between 2.4 days and 5.0 days are `nearly
1482: synchronized'. The corresponding period ranges for A-type stars are
1483: 4.9 and 10.5 days, or twice as large. They also found that the
1484: rotations of the primaries are synchronized earlier than their orbits
1485: are circularized. The maximum orbital period for circularized B
1486: binaries is 1.5 days and for A binaries is 2.5 days. Abt \etal\
1487: suggested that their finding is consistent with the theory of Zahn for
1488: early-type stars.
1489:
1490: Abt \& Boonyarak found that B and A stars in binaries with
1491: periods as long as 500 days have rotational period significantly
1492: shorter than the corresponding single stars. They therefore concluded
1493: that the synchronization processes can have impact on the stellar
1494: rotation even for relatively wide binaries with periods as long as 500
1495: days. The origin of these findings is still not clear, and needs
1496: further theoretical study.
1497:
1498: The results of these two studies are most interesting, despite the
1499: difficulties to derive the rotational stellar periods, specially
1500: because these two studies focus on early-type stars (see
1501: Subsection~\ref{subsection_early_type} for discussion of the very few
1502: studies of circularization of early-type stars). We wish that similar
1503: studies would have been performed for G and K stars, and the results
1504: confronted with the theory. We could benefit from enlarging the
1505: interaction between theory and observations to include both
1506: circularization and synchronization of both early- and late-type stars.
1507:
1508: Note that because the synchronization timescale is probably
1509: substantially shor\-ter than that of circularization, the synchronized
1510: binaries might be more common than the circularized orbits (see
1511: above). This means that we can find many synchronized binaries with
1512: eccentric orbits. However, for eccentric binaries the orbital angular
1513: velocity is not constant, and therefore the synchronization frequency
1514: is not well defined. There is no rotating frame of reference attached
1515: to the primary for which the secondary is at rest. Therefore,
1516: eccentric binaries need a new definition of synchronization which will
1517: be reviewed in the next subsection.
1518:
1519: %----------------------------------
1520: \subsection{Pseudo-synchronization}
1521: %----------------------------------
1522:
1523: As the tidal forces strongly depend on the distance between the two
1524: stars, it is clear that in eccentric binaries each star should
1525: respond mostly to the orbital angular velocity near the periastron,
1526: where its companion is the closest. In a seminal paper, Hut
1527: (\nocite{hut81}1981) has demonstrated that in the so-called weak friction approximation
1528: (see Zahn in this volume),
1529: each star achieves its equilibrium when it rotates with a frequency that is
1530: smaller but still close to the orbital periastron frequency. The equilibrium
1531: frequency, which Hut calls the pseudo-synchronization frequency,
1532: is
1533: \begin{equation}
1534: n_{pseudo}=\frac{1+\frac{15}{2}e^2+\frac{45}{8}e^4+\frac{5}{16}e^6}
1535: {(1+3e^2+\frac{3}{8}e^4)(1-e^2)^{3/2}}\,n_{orbit} \ ,
1536: \end{equation}
1537: where $n_{orbit}=2\pi/P$ is the {\it averaged} orbital
1538: frequency. Sometimes it is convenient to express the
1539: pseudo-synchronization frequency with the angular velocity at
1540: periastron --- $n_{peri}$:
1541: \begin{equation}
1542: n_{peri}=\frac{(1+e)^{1/2}}{(1-e)^{3/2}}\,n_{orbit} \ ,
1543: \end{equation}
1544: which leads to the expression
1545: \begin{equation}
1546: n_{pseudo}=\frac{1+\frac{15}{2}e^2+\frac{45}{8}e^4+\frac{5}{16}e^6}
1547: {(1+3e^2+\frac{3}{8}e^4)(1+e)^{2}}n_{peri} \
1548: \end{equation}
1549: (Hut 1981).
1550: To third order of the eccentricity $e$ one gets:
1551: \begin{equation}
1552: n_{pseudo}\simeq
1553: (1-2e+7.5e^2-13e^3)\,n_{peri}=
1554: (1+6 e^2)\,n_{orbit}
1555: \ .
1556: \end{equation}
1557: %
1558:
1559: Only subsequent to the work of Hut (1981) can we examine the eccentric
1560: orbit for its pseudo-synchronization. Of particular interest are
1561: coeval samples of binaries with a transition between the
1562: pseudo-synchronized and non-pseudo-synchronized binaries.
1563:
1564: %------------------------------------------
1565: \subsection{Synchronization of young stars}
1566: %------------------------------------------
1567:
1568: Two important studies of synchronization of short-period binaries
1569: have been performed recently. Marilli \etal\ (\nocite{mar07}2007)
1570: photometrically observed 40 PMS T Tauri stars that were discovered as
1571: X-ray sources by the ROSAT Survey of the Orion complex, and
1572: subsequently were observed optically with both low and high resolution
1573: spectroscopy. They obtained extensive photometric data and analysed
1574: them with the Lomb-Scargle periodogram (Lomb \nocite{lomb76}1976;
1575: Scargle \nocite{scar82}1982), deriving photometric modulation periods
1576: for 39 stars. Assuming the modulations are due to spots, these periods
1577: present the stellar rotational periods. Meibom \etal\
1578: (\nocite{mei06}2006) derived the rotation periods of 13 close binaries
1579: in M34 and M35, two young open clusters, with an age of 150 and 250
1580: Myrs, respectively.
1581:
1582: %------------------------------------------------------------------
1583:
1584: \begin{table}
1585: \label{tab:pseudo}
1586: \caption{The PMS and M35 young cluster eccentric systems with their
1587: rotation periods from Marilli \etal\ (in the first part of the
1588: table) and Meibom \etal\ (below the horizontal line in the
1589: table). For each system the pseudo-synchronized period is calculated
1590: according to the Hut (1981) recipe. The last system is the only circular system with a non-synchronized rotational period from M35.}
1591: \vskip 1pc
1592: \begin{tabular}{lcccc}
1593: Binary & Orbital & e\ \ & Pseudo-synch. &Rotational \\
1594: & Period & & Period & Period \\
1595: & (days) & & (days)& (days)\\
1596: \hline
1597: 044059.2-084005 & 13.56 & 0.22 & 10.5 & 2.75 \\
1598: 053043.1-043453 & 40.57 & 0.32 & 24.7 & 12.90 \\
1599: 053202.1-073153 & 46.85 & 0.47 & 18.4 & 3.44 \\
1600: 0539.8-0205 &18.74 & 0.39 & 9.4 & 4.48 \\
1601: \hline
1602: 06091557+2410422 & 8.17 &0.65 & 1.6 & 3.71 \\
1603: 06095563+2417454 &30.13 &0.27 & 20.8 & 2.84 \\
1604: 06085441+2403081 & 12.28 & 0.55 & 3.7 & 6.03 \\
1605: 06090257+2420447 & 10.28 & 0.001 & 10.3 & 2.3
1606: \end{tabular}
1607: \label{tab_synchro}
1608: \end{table}
1609: %--------------------------------------------------------------
1610:
1611:
1612: Eight of the stars observed by Marrilli \etal\ were found to be in
1613: spectroscopic binaries (Covino \etal\ \nocite{covino2001}2001). Four
1614: systems, with orbital periods shorter than 10 days, were found to be
1615: circular and synchronized. The other four binaries, with orbital
1616: periods between 10 and 50 days, with eccentric orbits, are listed
1617: in the upper part of Table~\ref{tab_synchro}. The table shows the
1618: orbital, rotational and pseudo-synchronization periods. It is easily observed that the rotational periods are substantially shorter than the
1619: pseudo-synchronization periods.
1620:
1621:
1622: \begin{figure}
1623: \includegraphics[width=11cm]{Marilli_Meibom.eps}
1624: \begin{centering}
1625: \caption{Rotational periods as a function of the orbital periods for
1626: eight PMS binaries (Marilli \etal\ 2007) and six young stars (Meibom
1627: \etal\ 2006). Circular binaries are encompassed by a circle. The
1628: dashed line is the locus of points for which the rotation period
1629: equals to the orbital period.}
1630: \label{fig_synchro}
1631: \end{centering}
1632: \end{figure}
1633:
1634: Six systems from the Meibom \etal\ (2006) study have periods shorter
1635: than 100 days, and of the six, three systems have eccentric
1636: orbits. These systems are included in Table~\ref{tab_synchro},
1637: together with the system 06090257+2420447, for which the orbit is
1638: circular but is very far from synchronization. All binaries in the two
1639: studies with periods shorter than 100 days, including those which are
1640: documented in the table and another six circular {\it and}
1641: synchronized binaries, are plotted in Figure~\ref{fig_synchro}.
1642:
1643: Both samples are extremely small, but one feature seems to emerge from
1644: the figure: the transition period between circular and eccentric
1645: orbits is between 8 and 10 days. It would be reasonable to expect that
1646: the transition between pseudo-synchronized and non-pseudo-synchronized
1647: binaries should appear on {\it much longer} period, as the
1648: synchronization processes are a few orders of magnitude more
1649: efficient. However, the striking feature of this figure is that the
1650: transition between pseudo-synchronized and non-pseudo-synchronized
1651: systems is at about the same period!
1652:
1653: To understand this surprising similarity between the two transition
1654: periods we must bear in mind that while the circularization
1655: processes worked undisturbed during the PMS phase with high efficiency
1656: (Zahn \& Bouchet 1989), the synchronization processes were drastically
1657: affected by the internal variation inside the stars which changed their
1658: radii and therefore their rotational period. The first 100 Myrs of M35
1659: after the stars settled on the main sequence were probably not enough
1660: to achieve pseudo-synchronization for binaries with periods
1661: longer than 8-10 days. In order to observe the strong
1662: potential of the synchronization processes, we apparently need to
1663: observe much older clusters.
1664:
1665: Note that for the two most eccentric systems in Table 3, with
1666: eccentricity of 0.55 and 0.65, the pseudo-synchronized period is {\it
1667: shorter} than the actual rotational period of the primary star. This
1668: is in contrast to the other five systems with more moderate
1669: eccentricities, for which the pseudo-synchronized period is {\it
1670: longer} than the rotational period. If this feature is validated
1671: by further observations, the difference between the highly eccentric
1672: binaries and the binaries with small or medium eccentricities could be a
1673: noteworthy feature (see, however, the discussion of V1143 Cyg in
1674: Section~\ref{V1143}. The observed longer rotational periods could be
1675: associated with the approximation used by Hut in his work, which did
1676: not take into account the reduction of the turbulent viscosity on a
1677: timescale shorter than the convective timescale (Goldman \& Mazeh
1678: 1994). It is possible that the pseudo-synchronization periods for some stars in
1679: highly eccentric orbits are longer than the ones derived by Hut's
1680: formula. It is therefore extremely important to derive stellar
1681: rotational periods for many more highly eccentric binaries and to compare
1682: them with the pseudo-synchronized periods.
1683:
1684: %-----------------------------
1685: \subsection{Interesting cases}
1686: %-----------------------------
1687:
1688:
1689: \subsubsection{V1143 Cyg}
1690: \label{V1143}
1691: %----------------------------------------
1692:
1693: V1143 Cyg is a bright, nearby, eclipsing double-lined spectroscopic
1694: binary, and therefore allows, in principle, for a determination of the
1695: masses, radii and luminosities of the two stars (e.g., Popper
1696: \nocite{popper1980}1980). The binary consists of two F5V stars
1697: orbiting each other with a period of 7.6 days and an eccentricity of
1698: 0.54 (Snowden \& Koch \nocite{snowden1969}1969). Apparently, at the
1699: age of 2 Gyrs, the system did not have enough time to circularize (Andersen
1700: \etal\ \nocite{andersen1987}1987). Andersen \etal\ derived the
1701: rotational broadening of the two components to be $V_A \sin i_A =
1702: 18\pm3$ and $V_B \sin i_B = 28\pm3$ for the primary and secondary,
1703: respectively. With the derivation of the two radii --- $1.346\pm0.023$
1704: and $1.323\pm0.00023$ solar radii for the primary and secondary, and
1705: assuming an inclination close to $90^{\circ}$, one gets for the rotation
1706: period of the two stars 3.8 and 2.3 days, respectively. Andersen
1707: \etal\ results are confirmed by the very recent detailed study of
1708: Albrecht \etal\ (2007).
1709:
1710: Andersen \etal\ (1987) pointed out that the pseudo-synchronization
1711: period, derived from the orbital period and the eccentricity, is 2.3
1712: days, exactly equal to the rotation period of the
1713: secondary. Therefore, V1143 Cyg is a good example of a binary that has
1714: reached pseudo-synchronization, at least for one star, but obviously
1715: has not yet reached circularization. Thus, V1143 Cyg could be
1716: considered one of the few binaries that strongly supports Hut's
1717: (1981) theory of pseudo-synchronization.
1718:
1719: However, the rotational period of the primary is a challenge to the
1720: theory. It is hard to conceive how the secondary could have been
1721: pseudo-synchronized while the primary was not, especially since
1722: the finding of Andersen \etal\ (1987) indicates that the primary and the
1723: secondary are extremely similar. Is it possible that the initial rotation of
1724: the primary and the secondary were so different that there was enough
1725: time to pseudo-synchronize the secondary but not the primary? I find
1726: such an explanation hard to believe. If it is indeed true, the disparity
1727: between the rotational periods of the two components remains a
1728: mystery. The primary of V1143 Cyg joins the group of highly eccentric
1729: binaries with rotational periods longer than Hut's
1730: pseudo-synchronization period, which was discussed above.
1731:
1732: Recent work on the inclination of the two stars of V1143 Cyg
1733: relative to the orbital plane is reviewed in Section~\ref{V1143_RM}.
1734:
1735: \subsubsection{KH 15D}
1736: \label{KH_15D}
1737: %---------------------
1738:
1739: KH 15D is a T Tau binary star in the young cluster NGC 2264 (Badalian
1740: \& Erastova \nocite{badalian1970}1970) that fades by three mag every
1741: 48 days (Kearns \& Herbst \nocite{kearns1998}1998). The best model to
1742: account for the many observed details of the periodicity of KH 15D
1743: assumes that the binary, with an orbital period of 48 days, is
1744: surrounded by a precessing circumbinary disk (e.g., Herbst \etal\
1745: \nocite{herbst2002}2002). Because of the binary motion, the primary
1746: and the secondary are being occulted periodically by the disk (e.g.,
1747: Hamilton \etal\ \nocite{hamilton2003}2003; Chiang \& Murray-Clay
1748: \nocite{chiang2004}2004; Winn \etal\ \nocite{winn2004}2004; 2006).
1749:
1750: Hamilton \etal\ (\nocite{hamil05}2005) measured a rotational period of
1751: 9.6 days for the primary star of KH 15D, based on photometric
1752: periodicity observed out of eclipse. {\it Assuming}
1753: pseudo-synchronization, Herbst \& Moran (\nocite{herbst2005}2005) and
1754: Hamilton \etal\ (2005) derived an eccentricity of
1755: $e=0.65\pm0.01$. This was probably consistent with the eccentricity of
1756: $0.68 \leq e \leq 0.8$, derived by Johnson \etal\
1757: (\nocite{johnson2004}2004) from Keck radial-velocity
1758: measurements. Indeed, Winn \etal\ (2006) derived a value of
1759: $e=0.574\pm0.017$ for the orbit of KH 15D in one of their models. If
1760: this value is correct, then the primary star in KH 15D system rotates
1761: with a frequency close to its pseudo-synchronization frequency, despite the long orbital period of 48 days.
1762: Additional spectroscopic and photometric observations are certainly needed in
1763: order to establish the rotational period of the primary, the
1764: eccentricity, and then the pseudo-synchronization of this fascinating
1765: system.
1766:
1767: Recent work on the stellar inclination of KH 15D relative to its
1768: orbital motion is reviewed in Section~\ref{KH_RM}.
1769:
1770:
1771: %=========================
1772: % %
1773: \section{Alignment} %
1774: \label{section_alignment} %
1775: % %
1776: %=========================
1777:
1778: As stated in the introduction, tidal forces in binaries tend also to
1779: align the rotation axes of the two stars with the orbital angular
1780: momentum. As argued for synchronization, the alignment timescale
1781: should be two or three orders of magnitudes shorter than the
1782: circularization timescale. This is so because the angular momentum
1783: transferred between the orbital motion and the stellar rotation in
1784: order to reach alignment is much smaller than that which is needed to attain
1785: circularization. This means that we can expect binaries to reach
1786: alignment long before reaching circularization, and we expect all
1787: circularized binaries to be aligned.
1788:
1789: Contrary to circularization and synchronization, we do not know if
1790: binaries are formed with misaligned rotation. On the contrary, one
1791: could argue that binary systems are formed with stellar spin aligned
1792: with the binary motion. This is so because both stellar rotation and
1793: orbital motion are relics of the same angular momentum of the
1794: primordial cloud, out of which the binary was formed. Common
1795: primordial spin-orbit misalignment can rule out some binary formation
1796: scenarios (e.g., Durisen \etal\ \nocite{durisen2001}2001; Bate et
1797: al. \nocite{bate2002}2002; Bonnell \& Bate \nocite{bonnell2005}2005;
1798: Machida 2005). \nocite{machida2005} Thus, we do not know if tidal
1799: interaction to attain alignment should set into action at all. It is
1800: therefore important for our understanding of binary formation to
1801: measure the present extent of the spin-orbit inclinations in the
1802: binary population.
1803:
1804: Observationally, the spin-orbit inclination, $i_{rel}$, is
1805: difficult to obtain, and consequently not much work has been done on binary
1806: alignment. One way to derive the relative inclination is to use
1807: stellar spectra with high enough resolution and S/N ratio that can
1808: yield detailed line profiles. The line profiles can be modelled to give
1809: projected stellar rotational broadening --- $V_*\sin i_*$, where
1810: $i_*$ is the inclination of the stellar rotation axis relative to
1811: our line of sight. With an {\it independent} knowledge of the size of
1812: the stellar radius {\it and} rotational period one can estimate the
1813: inclination angle $i_*$.
1814:
1815: Note that this approach allows us to measure the stellar spin inclination
1816: relative to our line of sight and {\it not} relative to the binary
1817: orbit. In order to derive $i_{rel}$ we also need to know $i_{binary}$
1818: --- the orbital inclination of the binary. This angle is not known for
1819: spectroscopic binaries, and can be derived directly from the
1820: observations only for eclipsing binaries. The difference between the
1821: two inclinations, $i_{binary}-i_*$, is an estimate of the {\it
1822: minimum} relative angle $i_{rel}$.
1823:
1824: In addition, we do not know the stellar radius and the
1825: rotational period for most stars. The stellar radii can be directly obtained only
1826: from observations of eclipsing binaries. Estimation of the
1827: stellar radii from the spectral type is not very
1828: accurate. The stellar rotational period is also not well established, and
1829: can be obtained either via special photometric monitoring of the system
1830: (see above), or by {\it assuming} synchronization. As a result of
1831: these drawbacks, the line broadening approach to studying binary
1832: alignment was not used extensively in the past. In fact, in some cases
1833: the derivation is reversed, and is used to estimate the rotational
1834: period (see Section~\ref{section_synchro}). Another example is the work of Beatty
1835: \etal\ (\nocite{beatty2007}2007), which derived the radius of the
1836: primary for the eclipsing single-lined binary HAT-TR-205-013 from the
1837: observed $V\sin i_*$, assuming synchronization and alignment.
1838:
1839: As discussed in Section~\ref{DI_Her}, spin-orbit misalignment can
1840: cause three noteworthy effects. First, it can change the apsidal
1841: precession of the binary by introducing counter apsidal precession. As
1842: in the case of DI Her, such an effect can, in principle, be measured
1843: if the actual precession is observed and the other precessions can be
1844: derived. Second, the binary tidal interaction will cause the stellar
1845: spins {\it and} the orbital plane to precess around the total angular
1846: momentum of the system with the same period. Such a forced precession
1847: will change the stellar spin relative to our line of sight, resulting
1848: in a change of the stellar rotational broadening of the two stars.
1849: Third, it might also slightly change the inclination angle of the
1850: binary, which can be detected by radial-velocity measurements, as well
1851: as by photometry if the binary is an eclipsing system. However, a
1852: study of the old observations of DI Her by Reisenberg \& Guinan (1989)
1853: is one of the few works that tried to use this indirect line of
1854: evidence for detecting spin-orbit misalignment (see, however, Mayor \&
1855: Mazeh (1987) for a different misalignment found in triple systems and
1856: the observational search for its effects).
1857:
1858: Evidence for significant spin-orbit misalignment should have
1859: implications for our understanding of other systems. For example, a
1860: forced precession induced by spin-orbit misalignment was already
1861: considered in the context of X-ray binaries. It was suggested by
1862: Roberts (\nocite{roberts1974}1974) to account for the 35-day
1863: periodicity of Her X-1 (Giacconi \etal\ \nocite{giacconi1973}1973), and for
1864: the 164-day precession of the relativistic jets of SS 433 (see Margon
1865: \nocite{margon1984}1984). This idea was never confirmed for Her X-1,
1866: and a different precession model of Katz (\nocite{katz1973}1973)
1867: became more popular. For SS 433 the evidence is also not conclusive
1868: (see a review of SS433 recent observations and theory by Fabrika
1869: \nocite{fabrika2004}2004). Therefore, our comprehension of the typical
1870: spin-orbit inclination in binaries is of great importance. It is a
1871: pity that the derivation of this inclination from indirect
1872: observations is so difficult.
1873:
1874: Another, more direct, method to derive the spin-orbit relative
1875: inclination angle is through the Rossiter-McLaughlin effect, which is
1876: relevant only for eclipsing binaries. This effect is described in
1877: detail in the next section.
1878:
1879: %------------------------------------------
1880: \subsection{The Rossiter-McLaughlin effect}
1881: \label{rm_effect}
1882: %------------------------------------------
1883:
1884: The Rossiter-McLaughlin (RM) effect is named after Rossiter
1885: (\nocite{rossiter1924}1924) and McLaughlin
1886: (\nocite{mclaughlin1924}1924) who observed the effect for $\beta$
1887: Lyrae and Algol --- two famous eclipsing binaries, within the same year.
1888:
1889: The RM effect is the varying shape of the stellar line profile during
1890: eclipse and is based on a combination of two very basic ideas. The
1891: first is stellar rotational broadening, which is due to the fact that different
1892: sections of the stellar disk, at different distances from the
1893: projected stellar rotational axis, have different Doppler shifts
1894: (e.g., Gray \nocite{gray2005}2005). The second is that during the
1895: progress of an eclipse, varying parts of the stellar disk of the
1896: eclipsed component are covered by the eclipsing star. Therefore, the
1897: line shape of the eclipsed star is bound to vary during the
1898: eclipse. In fact, the eclipsing body allows us, through observing the
1899: RM effect, to indirectly resolve the disk of the eclipsed star. Such an
1900: effect was observed, for example, by Rauch \& Werner
1901: (\nocite{rauch2003}2003) in their observations of AA Dor.
1902:
1903: The RM effect has one additional interesting feature. If the orbital
1904: motion of the companion is in the same direction as the stellar
1905: rotation, then during the first part of the eclipse the blue-shifted
1906: part of the stellar disk is eclipsed, while the red-shifted part of the stellar disk is
1907: eclipsed during the second part of the eclipse. If, on the other hand,
1908: the orbital motion and the stellar rotation are in opposite
1909: directions, then the red-shift part of the disk is eclipsed
1910: first. Thus, the RM effect can also detect the relative direction of
1911: stellar rotation.
1912:
1913: Dividing the eclipse into two periods --- one when the red-shifted
1914: side of the disc is eclipsed, and the other when the blue-shifted side
1915: is eclipsed, might be illuminating on a conceptual level. For example,
1916: if the stellar axis is aligned with the orbital motion, then we expect
1917: these two parts of the eclipse to be symmetrical for circular
1918: orbits. Asymmetry between the two parts clearly indicates a nonzero
1919: relative inclination.
1920:
1921: The RM effect recently attracted new interest because of its
1922: application to extrasolar planets (e.g., Queloz \etal\
1923: \nocite{queloz2000}2000; Winn \etal\ \nocite{winn2007}2007; Loeillet
1924: \etal\ \nocite{loeillet2007}2007). However, contrary to the RM effect
1925: on a transiting planet, in a binary system the derivation of the
1926: relative inclination is not easy, specifically due to the
1927: contribution of the secondary star to the total light of the
1928: system. In order to follow the distortion of the primary lines when
1929: the primary is eclipsed, for example, one must resolve the blending
1930: of the primary and secondary lines. This might be complicated whenever the
1931: secondary brightness is comparable to the primary brightness, because
1932: during the eclipse there is almost no shift between the velocities of
1933: the two stars, except for some very eccentric orbits. Such a problem
1934: does not exist when analysing the RM effect of transiting planets, as
1935: the planet's light is negligible. On the other hand, the amplitude of
1936: the effect in a binary is much larger than in that of a planet's
1937: transit, as a planet covers during its transit only a very small
1938: fraction of the stellar disk of its parent star.
1939:
1940: The RM effect can be detected even in spectra taken with low resolution,
1941: such that the profile changes can not be resolved. This is done by
1942: deriving the stellar radial velocities from the spectra obtained
1943: during the eclipse. The velocities are expected to have shifted from
1944: their orbital motion value, because the center of gravity of the lines
1945: are shifted by the eclipsing component. When the red-shifted part of
1946: the stellar disk is eclipsed the velocity is shifted to the blue and
1947: {\it vice versa}. The amplitude of the shift depends on the relative
1948: spin-orbit inclination, but also on the rotational velocity of the
1949: eclipsed star, the ratio of the two radii and the limb darkening of
1950: the stellar disk (Kopal \nocite{kopal1942}1942; 1959; Hosokawa
1951: \nocite{hosokawa1953}1953; Ohta \etal\ 2005; Winn \etal\
1952: \nocite{winn2005}2005; Gim{\'e}nez \nocite{gimenez2006} 2006).
1953:
1954: In fact, almost all observations of the RM effect were performed by
1955: observing the distortion of the radial-velocity curve during the
1956: eclipse. This includes the original work of Rossiter and
1957: McLaughlin, and the more recent work of Worek
1958: (\nocite{worek1996}1996), who observed AI Dra and V505 Sgr. In the next
1959: subsection we review two additional interesting observational studies
1960: of DE Dra and V1143 Cyg.
1961:
1962:
1963: %-----------------------------
1964: \subsection{Interesting cases}
1965: %-----------------------------
1966:
1967: \subsubsection{DE Dra}
1968: %---------------------
1969:
1970: DE Dra is an eclipsing single-lined spectroscopic binary (Furtig \&
1971: Meinunger \nocite{furtig1976}1976), consisting of a B9V primary in a
1972: 5.3 day circular orbit (Hube \nocite{hube1976}1976). Hube \& Couch
1973: (\nocite{hube1982}1982) obtained many spectra of the system during and
1974: outside the eclipse, in order to derive the radial-velocity modulation
1975: of the system. The obtained radial velocities clearly showed the RM
1976: effect during eclipse. The distortion of the radial-velocity curve
1977: was significantly asymmetric, suggesting a spin-orbit inclination of
1978: the primary. The analysis of Hube \& Couch of the distorted
1979: radial-velocity modulation indicated a rotational broadening of 180 km
1980: s$^{-1}$. They performed an independent line profile analysis
1981: which yielded a rotational velocity of only 135 km s$^{-1}$. It seems as if
1982: the inconsistency between these two values prevented Hube \& Couch from
1983: carrying their analysis of the RM effect further and deriving a firm value
1984: for the relative inclination of the primary axis.
1985:
1986: The observations and analysis of Hube \& Couch (1982) were not
1987: performed with modern techniques, so the statistical significance of
1988: their results is questionable. However, if true, this is the only
1989: binary that we know of that has a non-aligned rotational axis. This is
1990: most interesting because it can tell us something about the binary
1991: formation of DE Dra. In fact, the system has another feature which
1992: indicates that the binary has not yet reached tidal equilibrium. The
1993: primary rotational velocity, being either 180 or 135 km s$^{-1}$, is
1994: spinning much faster than the synchronization velocity, which is about
1995: 30 km s$^{-1}$. Therefore, we probably have in hand a binary system
1996: that has completed its circularization, if it was formed with
1997: eccentric orbit, but has not yet reached synchronization and
1998: alignment.
1999:
2000: The exceptional features of this system make further observations
2001: and analysis very desirable. New observations can confirm the
2002: non-alignment of the primary, resolve the secondary spectrum,
2003: study the secondary eclipse, and search for forced precession of the
2004: primary, which should manifest itself in varying inclination of the
2005: stellar axis. The latter feature would add new information that would
2006: contribute to our understanding of this remarkable system.
2007:
2008: \subsubsection{V1143 Cyg}
2009: \label{V1143_RM}
2010: %------------------------
2011:
2012: Just before the conclusion of this manuscript, Albrecht \etal\
2013: (\nocite{albrecht2007}2007) put a beautiful paper in {\it astro-ph}, which reports on a very careful study of the RM effect in the eclipsing
2014: binary V1143 Cyg. This binary (see Section~\ref{V1143}) consists of
2015: two F5V stars orbiting each other with a period of 7.6 days and an
2016: eccentricity of 0.54. As stated above, it is one of the very few
2017: binaries of which we know from observations that one of the
2018: components reached pseudo-synchronization for a highly eccentric
2019: orbit. However, only the secondary reached pseudo-synchronization,
2020: while the primary is still rotating with a period which is larger by
2021: 50\% than the pseudo-synchronization period (Andersen \etal\ 1987).
2022: The fact that the apsidal motion of the system (Gim{\'e}nez \&
2023: Margrave \nocite{gimenez1985}1985) {\it might} be slightly smaller
2024: than the theoretical expectation (see discussion above in
2025: section~\ref{DI_Her} for DI Her) motivated Albrecht \etal\
2026: (\nocite{albrecht2007}2007) to study the relative inclination of the
2027: two stars through the RM effect.
2028:
2029: Analysing high S/N ratio spectra in detail, Albrecht \etal\ derived
2030: the angle $\beta$ between the stellar spin axis and the orbital
2031: angular momentum, both projected onto the plane of the sky. Obviously,
2032: $\beta=0$ is consistent with perfect spin-orbit alignment. They found
2033: that $\beta_A=0.5\pm 4^{\circ}$ and $\beta_B=-3.9\pm 4^{\circ}$, for
2034: the primary and the secondary, respectively, to be consistent with a perfect
2035: alignment of both components.
2036:
2037: These results are most interesting, showing that the primary is
2038: aligned despite the fact that the binary is not synchronized. We would
2039: like to suggest that this might indicate that the V1143 Cyg system was
2040: formed with a rotation axis aligned with the orbital binary motion, at
2041: least for the primary. This is so because we {\it assume} that the
2042: timescale for pseudo-synchronization and alignment are similar, and
2043: therefore the non-synchronization of the primary indicates that there
2044: was not enough time during the lifetime of the system for the primary
2045: to reach synchronization or alignment. If this is true, observing the
2046: RM effect of V1143 Cyg told us something about the formation of the
2047: binary instead of its tidal evolution.
2048:
2049: \subsubsection{KH 15D}
2050: \label{KH_RM}
2051: %---------------------
2052:
2053: KH 15D is a T Tau binary with a period of 48 days, which is surrounded
2054: by a precessing circumbinary disk (e.g., Herbst \etal\
2055: \nocite{herbst2002}2002). The primary and the secondary are being
2056: occulted periodically by the disk (for a more detailed account of this
2057: interesting system see Section~\ref{KH_15D}).
2058:
2059: Winn \etal\ (2006) pointed out the possible RM effect induced by the
2060: disk when occulting the visible star of KH 15D or its halo. However,
2061: the optical spectra they obtained did not allow a quantitative analysis of
2062: this effect. Therefore, there is no information available for
2063: alignment or misalignment of the primary of KH 15D relative to its
2064: binary orbit. However, strong evidence was found for an inclined, warped,
2065: eccentric, precessing circumbinary disk (Chiang \& Murray-Clay; Winn
2066: \etal\ 2004; 2006) with a precession period of the order of 1000
2067: years. The cause of the misalignment between the binary plane of
2068: motion and its circumbinary disk is still a mystery, and might
2069: indicate some basic misaligned component of the angular momentum left
2070: from the formation era of the system.
2071:
2072:
2073: %=================================================
2074: %
2075: \section{Tidal Interaction in Close Triple System}
2076: \label{triple}
2077: %
2078: %=================================================
2079:
2080: As stated in the introduction, tidal interaction in close binaries
2081: acts until the binary reaches an equilibrium state, in which the orbit
2082: is circularized, the stellar rotations are synchronized and the
2083: stellar spins are aligned with the orbital motion. It is only when
2084: this state is reached that the stellar tides do not move relative to
2085: the surfaces of the two stars, and the stellar shapes are constant in
2086: the rotating frame of reference.
2087:
2088: However, when a binary is a part of a hierarchical triple system,
2089: which is composed of the binary and a third distant companion, such a
2090: stable configuration might not be possible. The reason for this is that the
2091: third star constantly injects eccentricity into the binary orbit by
2092: exerting its own `tidal forces' on the binary system as a whole. The
2093: origin of these tidal perturbations is the small difference between
2094: the gravitational acceleration of the two individual stars induced by
2095: the third star and the corresponding acceleration of the center of
2096: mass of the binary system. This difference does not depend on the
2097: actual radii of the two stars, but instead depends on the size of the
2098: binary system relative to its distance from the third star.
2099:
2100: One can classify the perturbation of the third star into three
2101: different effects.
2102:
2103: \begin{itemize}
2104: \item Modulation of the binary eccentricity.
2105:
2106: \item Apsidal precession of the binary.
2107:
2108: \item Nodal precession of the binary orbit around the total angular
2109: momentum of the triple system.
2110: \end{itemize}
2111: %
2112: The effects of the third star can appear on three timescales: the
2113: binary period, the third star period, and a long-term modulation
2114: (e.g., Brown 1936; Borkovits \& Forg\'{a}cs-Dajka 2008, this
2115: volume). In what follows we are mainly interested in the long-term
2116: modulation and its interrelation with the long-term tidal interaction
2117: that is occurring in the binary.
2118:
2119: It is advisable that one take the possible effects of an unknown
2120: distant companion into consideration when observing binary stars,
2121: because a third distant companion could easily escape the notice of
2122: observers in the course of studying binary systems. Unfortunately, in
2123: many cases no effort is made to search for a faint, distant companion
2124: after a binary is discovered and studied. Nevertheless, several lines
2125: of evidence indicate that the percentage of binaries that have a
2126: distant third companion is significant.
2127:
2128: Mayor \& Mazeh (1987) were the first to search systematically for
2129: third companions of spectroscopic binaries, using the precession of
2130: the nodes induced by the third companion. They looked in a small
2131: sample of spectroscopic binaries and found evidence that 25\% of the
2132: members of their sample have third companions.
2133: Other studies (Isobe
2134: \etal\ \nocite{Isobe1992}1992; Tokovinin \& Smekhov
2135: \nocite{tokovinin2002}2002; Pribulla \& Rucinski
2136: \nocite{pribulla2006}2006; D'Angelo \etal\ \nocite{dangelo2006}2006)
2137: also indicated a high frequency of triple systems. A recent search by
2138: Tokovinin \etal\ (\nocite{tokovinin2006}2006) found that 96\% of
2139: binaries with periods shorter than 3 days have a third companion! A
2140: major step forward in the observational study of triple systems is the
2141: construction of a new catalog by Tokovinin which
2142: includes all known stellar triple systems
2143: (\nocite{tokovinin1997}1997; see also the on-line version at http://www.ctio.noao.edu/~atokovin/stars/index.php).
2144:
2145: Note that the nodal precession of the orbital plane of a binary caused
2146: by a third star, an effect used by Mayor \& Mazeh (1987) to search for
2147: triple systems, might induce spin-orbit misalignment into the binary
2148: system (Fabrycky \& Tremaine 2007)\footnote{{\tt I thank D.~Fabrycky
2149: for pointing this point to me.}}.
2150: %
2151: Therefore, observational evidence for such misalignment might suggest
2152: a third companion. Spin-orbit misalignment due to a third star was
2153: suggested by Beust \etal\ (\nocite{beust1997}1997) for the eclipsing
2154: binary TY Cor Aus.
2155:
2156: This section will review the theory of long-term binary eccentricity modulation
2157: and apsidal motion induced by a third star and will point out its
2158: application for the tidal evolution of binaries that reside in triple
2159: systems. In fact, as suggested already by Mazeh \& Shaham (1979) and
2160: now supported by the observations and statistical analysis of
2161: Tokovinin \etal\ (2006), a third companion might have a major role
2162: in the formation of very short binaries through constant pumping
2163: of the binary eccentricity.
2164:
2165: %-----------------------------------------------------------------------
2166: \subsection{The binary modulations induced by a third distant companion}
2167: \label{subsection_triple_ecc}
2168: %-----------------------------------------------------------------------
2169:
2170: The zero-order approximation of the dynamics of hierarchical triple
2171: stellar systems assumes that the close binary moves in an orbit with
2172: its own semi-major axis $a_{1,2}$ and period $P_{1,2}$, while the
2173: third distant companion moves together with the binary center of mass
2174: in an orbit with $a_3$ and $P_3$. The hierarchy of the system relies
2175: on the ratio $a_{1,2}/a_3$ being small. Harrington
2176: (\nocite{harrington1968}1968; \nocite{harrington1969}1969) developed a
2177: theory of hierarchical triple stellar systems by expanding their
2178: Hamiltonian into a series of terms in the small parameter
2179: $a_{1,2}/a_3$. In order to obtain the long-term modulation, Harrington
2180: further averaged the Hamiltonian over the binary and the third star
2181: periods. The resulting Hamiltonian showed that the tidal forces of the
2182: third star vary the binary eccentricity $e_{1,2}$, the longitude of
2183: the periastron $g_{1,2}$ and the direction of the plane of motion with
2184: quasi-periodic modulation. The modulation period is of the order of
2185: $T_{mod}$:
2186:
2187: \begin{equation}
2188: T_{mod}=P_{1,2}\left(\frac{a_3}{a_{1,2}}\right)^3\,
2189: \frac{M_1+M_2}{M_3} \ ,
2190: \end{equation}
2191: %
2192: where $M_1$ and $M_2$ are the masses of the two components of the
2193: binary and $M_3$ is the mass of the distant companion (Mazeh \&
2194: Shaham 1979). One can then derive the equation of the binary
2195: eccentricity:
2196:
2197: \begin{equation}
2198: \frac{de_{1,2}}{dt}=\frac{2\pi}{T_{mod}}\gamma \sin(2g_{1,2})e_{1,2}
2199: \ ,
2200: \end{equation}
2201: %
2202: where $\gamma$ is a geometrical factor of order unity (Mazeh \& Shaham
2203: 1976; 1979). This equation conjugates the binary eccentricity
2204: variability with that of the longitude of the periastron, which is
2205: also being varied by the third star:
2206: \begin{equation}
2207: \frac{dg_{1,2}}{dt}=\frac{2\pi}{T_{mod}}\gamma [\delta+\cos(2g_{1,2})]
2208: \ ,
2209: \end{equation}
2210: %
2211: where $\delta$ is another geometrical constant of order unity. (For a
2212: different approach to the derivation of the apsidal motion induced by
2213: a third star see, for example, Brown \nocite{brown1936}
2214: \nocite{brown1937}1936; 1937). The two equations consider the three stars
2215: as point masses and ignore general relativistic effects.
2216:
2217: Further studies of the dynamics of triple stellar systems as three
2218: mass points include the work of Krymolowski \& Mazeh
2219: (\nocite{krymolowski1999}1999) and Ford \etal\
2220: (\nocite{ford2000}\nocite{ford2004}2000; 2004) who expanded the
2221: Hamiltonian to higher orders of $a_{1,2}/a_3$ and that of Beust
2222: (\nocite{beust2003}2003), who applied a symplectic integration to the
2223: stellar triple problem.
2224:
2225: In any specific triple system, the actual binary apsidal motion
2226: includes the precession induced by the binary tidal interaction and
2227: the general relativistic effect (see Section 3). Bozkurt \& De{\u
2228: g}irmenci (\nocite{bozkurt2007}2007) have just recently published a
2229: study of the apsidal motion of several known triple systems with
2230: detected apsidal motion. However, for all the systems they considered,
2231: the apsidal motion induced by the third star is several orders of
2232: magnitude slower than the observed one. The reason for this is that in all
2233: systems considered by Bozkurt \& De{\u g}irmenci, the existence of the
2234: third star was discovered by timing the eclipses minima and using the
2235: light-travel time technique. In principle, this is the same technique
2236: as that by which the apsidal motion was detected. Therefore, the
2237: orbital period of the third star, $P_{3}$, is necessarily of the same
2238: order of magnitude as the detected apsidal motion. Consequently, the
2239: apsidal motion induced by the third star is much longer than the
2240: detected one. This can be seen by writing Equation (7.1) as
2241:
2242: \begin{equation}
2243: T_{mod} \simeq P_{3}\left(\frac{P_3}{P_{1,2}}\right)\,
2244: \frac{M_1+M_2}{M_3} \ ,
2245: \end{equation}
2246: %
2247: which implies that $T_{mod}$ is much longer than $P_3$. In order to find triple
2248: systems with shorter apsidal binary motion, one need to look for triple
2249: systems with a much smaller $P_{3}/P_{1,2}$ ratio (see Tokovinin
2250: (1997) catalog of multiple systems). Two such systems are G:38-13
2251: (Mazeh \etal\ \nocite{mazeh1993}1993) and HD 109648 (Jha \etal\
2252: \nocite{jha2000}2000). In these systems the period of the
2253: modulation induced by the third star is of the order of 100 years.
2254:
2255:
2256: %---------------------------------------------
2257: \subsection{The eccentricity modulation
2258: combined with tidal circularization}
2259: \label{subsection_MS}
2260: %---------------------------------------------
2261:
2262: In many triple systems, $T_{mod}$ is of the order of the
2263: circularization timescale, $\tau_{circ}$, or {\it shorter}. In such
2264: cases one must take the modulations induced by the third star into
2265: account when considering the tidal evolution of close binaries,
2266: particularly their eccentricity.
2267:
2268: Mazeh \& Shaham (1979), for example, considered the long-term effect of
2269: the eccentricity modulation induced by the third star together with
2270: the binary tidal circularization (Equation~\ref{eq_circ})
2271:
2272: \begin{equation}
2273: \frac{de_{1,2}}{dt}=-\frac{1}{\tau_{circ}}e_{1,2}
2274: \ ,
2275: \end{equation}
2276: %
2277: which is also linear in $e_{1,2}$, and got, to first order, a
2278: periodic modulation combined with an exponential decay, as displayed
2279: schematically in Figure~\ref{fig_ecc_evolution}.
2280:
2281: \begin{figure}
2282: \includegraphics[width=13cm]{ecc_evolution.eps}
2283: \caption{Schematic presentation of the eccentricity evolution of a
2284: binary that resides in a triple system, after Mazeh \& Shaham
2285: (1979).}
2286: \label{fig_ecc_evolution}
2287: \end{figure}
2288:
2289: Figure~\ref{fig_ecc_evolution} presents a moderate third-star
2290: modulation of the eccentricity. However, this is not necessarily the
2291: case for all systems, as it all depends on the relative inclination of
2292: the two orbits. As pointed out by many studies (e.g., Krymolowski \&
2293: Mazeh \nocite{krymolowski1999}1999; Ford \etal\
2294: \nocite{ford2000}\nocite{ford2004}2000; 2004) the third star 'tidal
2295: forces' can inject a high eccentricity into the binary motion, if
2296: the inclination angle of the plane of motion of the third star
2297: relative to the binary plane of motion is higher than approximately
2298: $40^{\circ}$. This high-amplitude modulation was first noted by Kozai
2299: (\nocite{kozai1962}1962) for the motion of asteroids and was first applied
2300: to stellar systems by Mazeh \& Shaham (\nocite{mazeh1976}1976),
2301: who at that time were unaware of Kozai's work.
2302:
2303: As the third star modulation and the tidal circularization are both
2304: linear in $e$, one {\it could} conclude that at the end of the
2305: circularization process the modulation decays to zero. However, Mazeh
2306: and Shaham (1979) have demonstrated that the linearity of the third star
2307: modulation is true only to the first order of $a_{1,2}/a_3$. Upon
2308: integrating Newton's equations directly, they discovered that the
2309: eccentricity modulation is present even if the binary eccentricity
2310: starts with zero. The Mazeh \& Shaham effect was further theoretically
2311: studied and confirmed by Georgakarakos \nocite{georgakarakos2002}
2312: \nocite{georgakarakos2004}(2002; 2004). We therefore expect some
2313: short-period binaries that should have been circularized long ago to
2314: display small eccentricities.
2315:
2316: In fact, observational evidence for such an effect has already been
2317: pointed out by Mazeh (\nocite{mazeh1990}1990) in three short-period
2318: binaries, with significant eccentricities {\it and} strong indications
2319: for a third distant stellar companion. One of the examples found by
2320: Mazeh is the Hyades star BD $+23^{\circ}635$, which has an orbit of
2321: 2.4 days and an eccentricity of $0.057\pm0.005$ (Griffin \& Gunn
2322: \nocite{griffin1981}1981). Giuricin \etal\ (1984) also suggested that
2323: the small but significant eccentricities they found in systems with
2324: fractional primary radius larger than 0.3 (see
2325: Section~\ref{subsection_early_type} for details) must be due to
2326: unknown third distant companions. In all these cases, the tidal
2327: processes should have circularized the orbits long ago, and the
2328: interaction with the third star seems to be the only plausible
2329: conjecture that accounts for the observed eccentricity.
2330:
2331: The discovery of Mazeh \& Shaham (1979) that the binary eccentricity
2332: induced by the third star continues its modulation even after the
2333: complete tidal circularization may have implications for the evolution
2334: of close binaries. The eccentricity modulation implies that binaries
2335: with close enough companions will never get into a stable circular
2336: orbit, which means that the binary tidal dissipation will continue to
2337: work forever. The injected binary eccentricity invokes frictional
2338: forces inside the two stars that dissipate rotational energy and
2339: transfer angular momentum between the stellar rotation and binary
2340: orbital motion. This causes the semi-major axis of the binary to
2341: decrease at a rate of
2342: %
2343: \begin{equation}
2344: \frac{1}{a_{1,2}}\frac{da_{1,2}}{dt}=-\frac{1}{T_{circ}}Ae^2_{1,2} \ ,
2345: \label{eq_da_dt}
2346: \end{equation}
2347: %
2348: where $A$ is a geometrical factor of order unity. Mazeh and Shaham
2349: (1979) suggested, therefore, that in certain configurations the third
2350: distant companion can provoke the spiral-in of the close binary at a
2351: rate that can be derived by averaging Equation~\ref{eq_da_dt}. The
2352: third star motion can serve as the sink for the angular momentum of
2353: the shrinking binary.
2354:
2355: Obviously, the separation of the perturbations of the third star and
2356: the tidal friction into two equations is artificial, and was done only
2357: as a first approximation. Further studies, including Kiseleva \etal\
2358: (\nocite{kiseleva1998}1998), Eggleton \& Kiseleva-Eggleton
2359: (\nocite{eggleton2001}2001; \nocite{eggleton_kis2006}2006), Eggleton
2360: (\nocite{eggleton2006}2006), Borkovits \etal\
2361: (\nocite{borkovits2004}2004; \nocite{borkovits2005}2005;
2362: \nocite{borkovits2007}2007) and Fabrycky \& Tremaine
2363: (\nocite{fabrycky_tremaine2007}2007) combined the two effects into one
2364: set of coupled equations. Kiseleva-Eggleton, Eggleton and Fabrycky \&
2365: Tremaine further studied the Mazeh \& Shaham effect for cases where
2366: the eccentricity modulation is large, which occurs when the relative
2367: angle between the two orbital planes of motion is larger than $\sim
2368: 40^{\circ}$. They also concluded that some of the presently close
2369: binaries could have been formed by the combination of tidal binary
2370: friction and the third star induced modulation. The Tokovinin \etal\
2371: (2006) result, that 96\% of binaries with periods shorter than 3 days
2372: have a third companion, might be a strong evidence for the Mazeh \&
2373: Shaham conjecture.
2374:
2375:
2376: %==================================================
2377: %
2378: \section{Extrasolar planets and tidal Interaction}
2379: %
2380: %==================================================
2381:
2382: Since the discovery of 51 Peg b (Mayor \& Queloz
2383: \nocite{mayor1995}1995), more than 200 planets have been discovered
2384: (see, for example, the Extrasolar Planets Encyclopaedia\footnote{{\tt
2385: http://vo.obspm.fr/exoplanetes/encyclo/catalog-RV.php}}). As one can
2386: consider a star with a planet as a binary with an extremely small mass
2387: ratio, we can apply the theory of tidal interaction to these systems,
2388: and therefore expect to find some signs of its action. In particular,
2389: we expect the planets with short enough orbital periods to attain
2390: circularization, alignment, and planetary synchronization. This
2391: section will review in brief some observational lines of evidence that
2392: might indicate that tidal interaction between extrasolar planets and
2393: their parent stars took place during the lifetime of those systems.
2394:
2395: %---------------------------------------------
2396: \subsection{Radial drift by tidal interaction}
2397: %---------------------------------------------
2398:
2399: One possible outcome of the tidal interaction between a planet and its
2400: parent star is a radial drift of the planet, resulting from the
2401: transformation of angular momentum between the stellar rotation and
2402: the planetary orbital motion. The angular momentum of the planetary
2403: orbital motion is proportional to $m_p\sqrt{a_p}$, where $a_p$ is the
2404: planetary orbital radius and $m_p$ is the planetary mass, and is small
2405: relative to the stellar angular momentum for planets with orbital
2406: periods of a few days. Therefore, transfer of a small amount of
2407: angular momentum from or into the planetary orbital motion can
2408: substantially change the planetary orbital radius, an effect which is
2409: unique to planetary motion. This subsection discusses the two possible
2410: directions of the radial drift and their astrophysical implications.
2411:
2412: \subsubsection{Stopping the migration}
2413: \label{subsubsection_stopping}
2414: %-------------------------------------
2415:
2416: One possible effect of the radial drift was suggested already by Lin
2417: \etal\ (\nocite{lin1996}1996) for 51 Peg b, whose orbital radius is
2418: about 0.05 AU. The idea is based on the paradigm that Jupiter-sized
2419: planets can not be formed close to their parent stars, at a distance
2420: of $\sim0.05$ AU, but instead must be formed behind the 'ice line' of
2421: the protoplanetary disk, at a distance of $\sim5$ AU (e.g.,
2422: Bodenheimer \& Pollack \nocite{bodenheimer1986}1986; but see, however,
2423: Boss \nocite{boss1997}1997 for a heretical approach). At that distance
2424: the icy particles initially form a rocky core of several earth masses,
2425: which subsequently accretes gas to form a giant planet. In order to
2426: save the core-accretion model, Lin \etal\ suggested that 51 Peg b
2427: migrated from the location of its formation into its present radius
2428: through interaction with the material in the disk, an idea that had
2429: already been discussed a decade before the discovery of 51 Peg b
2430: (e.g., Goldreich \& Tremaine \nocite{goldreich1980}1980; Papaloizou \&
2431: Lin \nocite{papaloizou1984}1984; Lin \& Papaloizou
2432: \nocite{lin1986}1986) in the context of the solar system. Such a model
2433: for the short-period extrasolar planets had to explain why the
2434: migration stopped at $\sim0.05$ AU and did not push the planet into
2435: the surface of the star. The idea of Lin \etal\ was that this was done
2436: by the tidal interaction between the planet and its parent star.
2437:
2438: This 'stopping mechanism' can work only if the star rotates on its spin
2439: axis faster than the orbital frequency of its planet. In such
2440: a case, the tidal interaction between the planet and the star tends to
2441: slow down the stellar rotation, taking angular momentum from the star
2442: into the planetary orbital motion. The additional orbital angular
2443: momentum forces the planetary orbital radius to {\it increase}. This
2444: result is counterintuitive, because the tidal interaction eventually
2445: {\it slows} both the planetary orbital motion and the stellar
2446: rotation. Nevertheless, it relies on the conservation of angular
2447: momentum and Kepler's laws, and must therefore be correct.
2448:
2449: According to this scenario, the planet can migrate inwards as long as the
2450: interaction with the disk is stronger than the stopping force of the
2451: tidal interaction with the star. The planet stops its migration when
2452: the two forces cancel each other out. The timescale of the
2453: radial drift caused by the interaction with the spinning star was
2454: estimated by Lin \etal\ (1996) to be:
2455: %
2456: \begin{equation}
2457: \tau_{radial}=\frac{a_p}{\mid\dot{a}_p\mid}=\frac{9}{2n_p} q^{-1}
2458: \left(\frac{a_p}{R_*}\right)^5 Q_*
2459: \propto \frac{M_*^{1/2}Q_*}{R_*^5} \, \frac{a_p^{13/2}}{m_p}
2460: \label{radial_timescale}
2461: \end{equation}
2462: %
2463: where $n_p$ is the Keplerian angular velocity of the planet,
2464: $q=m_p/M_*$ is the mass ratio between the planetary mass, $m_p$, and
2465: the stellar mass, $M_*$, $R_*$ is the stellar radius, $a_p$ is the
2466: planetary orbital radius and $Q_*$ is the stellar tidal dissipation
2467: factor (See Sasselov \etal\ 2003 for a different expression of
2468: $\tau_{radial}$). Note that Trilling \etal\
2469: \nocite{trilling1998}(1998) pointed out that the numerical coefficient
2470: in this equation is only true for bodies of uniform density, and
2471: should be corrected for the actual stellar mass profiles.
2472:
2473: The problem with this equation is the unknown factor $Q_*$, which
2474: describes how efficiently rotational energy is dissipated by friction
2475: within the star. Lin \etal\ (1996) adopted a value of $1.5\times 10^5$
2476: ``for a main-sequence star, based on the observations that the orbits
2477: of short-period pre-main-sequence binary stars and the main-sequence
2478: binary stars in the Pleiades cluster are circularized for periods up
2479: to 5 and 7 days, respectively''. We note, however, that the error on
2480: this derivation of the value of $Q_*$ from the transition period of
2481: the Pleiades could be substantial for two reasons. First, the binaries
2482: in the Pleiades cluster might have been circularized during their
2483: pre-main-sequence phase. In such a case, we can not use the transition
2484: period of the Pleiades at all. Second, we can not estimate the exact
2485: transition period, nor do we know the exact mass of the secondaries in
2486: the binaries of those clusters, as was discussed in
2487: Section~\ref{subsub_transition} in detail. The factor $Q_*$ is also a
2488: strong function of the star itself, depending on the depth of the
2489: convective envelope and the stellar radius at the time of the inward
2490: migration (e.g., Sasselov 2003).
2491:
2492: As we can see from Equation~\ref{radial_timescale}, the stopping
2493: mechanism is a strong function of the planetary orbital radius, and
2494: the radial-drift timescale drops dramatically for small distances. The
2495: equilibrium between the interaction with the disk and the tidal
2496: interaction with the spinning star depends on the disk mass and
2497: density profile on the one hand, and the stellar structure and the
2498: intensity of the stellar interaction on the other hand. According to
2499: this scenario, when the protoplanetary disk disappears, the planet will
2500: drift outward until its radial-drift timescale gets too long.
2501:
2502: \begin{figure}
2503: \includegraphics[width=13cm]{extrasolar_per_dist.eps}
2504: \caption{The period distribution of the extrasolar planets.}
2505: \label{extrasolar_period}
2506: \end{figure}
2507:
2508: Some support for the stopping mechanism scenario can be found in the
2509: period distribution of the known planets, which is plotted in
2510: Figure~\ref{extrasolar_period} (Data were taken from the Extrasolar
2511: Planets Encyclopaedia\footnote{{\tt
2512: http://vo.obspm.fr/exoplanetes/encyclo/catalog-RV.php}}). One
2513: possible interpretation of the figure is that between 2 and 200 days
2514: the distribution is flat in log period, with an excess of about 20
2515: planets around the period of 3 days. According to the migration
2516: scenario and its stopping extension, at least half of the planets with
2517: periods of around 3 days were stopped by the tidal interaction during
2518: their inward migration. The excess of planets with periods around 3
2519: days is still there even if we take out the transiting planets, for
2520: which there is a strong selection effect at short periods.
2521:
2522: This interpretation of the observed distribution implies that many of
2523: the parent stars of the planets with orbital periods of 3 days have,
2524: or at least had in the past, rotational periods shorter than 3 days.
2525: We can try to corroborate this conjecture by deriving the present
2526: stellar rotational periods of these systems. One approach to deriving
2527: the rotation period is based on the rotational broadening of the
2528: stellar profiles, relying on the stellar radius and
2529: inclination. However, even if we know the inclination, this approach
2530: is quite inaccurate, as discussed in Section~\ref{section_synchro}.
2531:
2532: Another approach is to follow the photometric variation of the parent
2533: stars that can reveal their rotational periods. Consider, for example,
2534: the M-type star Gliese 436, whose planet was discovered by Butler \etal\
2535: (\nocite{butler2004}2004), and its transiting nature was revealed only
2536: recently by Gillon \etal\ (\nocite{gillon2007}2007). Demory \etal\
2537: (\nocite{demory2007}2007) found some evidence for a photometric
2538: modulation with a period of about 45 days, which could be the
2539: rotational period of the parent star. If this is true, then the
2540: rotational period of the M star is much longer than the 2.6 day
2541: orbital period of the planet. Bonfils \etal\
2542: (\nocite{bonfils2007}2007) found a similar situation for Gliese 674,
2543: for which the orbital period is 4.7 days and the rotation period of
2544: the M star is about 35 days. However, the assumed long rotational
2545: periods of these two stars could be the result of spinning down of
2546: the stellar rotation, a commonly observed phenomenon, in M-type stars in
2547: particular (e.g., Cardini \& Cassatella \nocite{cardini2007}2007).
2548:
2549: Other mechanisms can be invoked to explain the peak of the period
2550: distribution of the extrasolar planets at around 3 days. For example,
2551: one can assume that the protoplanetary disks have been evaporated when
2552: the planets arrived into their present radii, which correspond to
2553: periods around 3 days. After all, we know that the disks disappear
2554: after a period of the order of 10 Myrs, either by stellar wind and
2555: photoevaporation or by viscous processes in the disk itself (e.g.,
2556: Haisch \etal\ \nocite{haisch2001}2001). Another version of this
2557: stopping mechanism was suggested by Kuchner \& Lecar
2558: (\nocite{kuchner2002}2002), who assumed a central 'hole' in the
2559: protoplanetary disks, with an edge corresponding to an orbital period
2560: of 6 days. The central hole is formed by turbulent accretion due to
2561: magnetorotational instability that acts on the material in the central
2562: part of the disk, where small grains reached sublimation
2563: temperature. The planetary migration halts when the exterior 2:1
2564: Lindblad resonance reaches the external radius of the hole. Another
2565: stopping mechanism (e.g., Trilling \etal\ 1998) is based on the
2566: assumption that the short-period planets got so close to their parent
2567: stars that their Roche Lobe became smaller than their planetary
2568: radii. This caused mass transfer from the planets onto the stars,
2569: which resulted in an outward motion of the planet in order to conserve
2570: angular momentum.
2571:
2572: A different idea was raised recently by Burkert \& Ida
2573: (\nocite{burkert2007}2007), who suggested that the period distribution
2574: peak was in fact a pronounced dip at the range of 10--100
2575: days. Furthermore, their analysis indicated that the dip appeared only
2576: in planets around F-type stars, with masses larger than $1.2
2577: M_{\odot}$, while the distribution of planets around G-type stars showed
2578: log-flat distribution. To account for the different period
2579: distributions, Burkert \& Ida assumed that the sizes of the
2580: protoplanetary disks around F and G stars were different, resulting in
2581: different patterns of migration. Eventually, this
2582: caused a gap in the period distribution of planets around F
2583: stars, but not around G stars. It seemed that their simulations could
2584: produce the period distribution of the planets and did not necessitate
2585: an extra stopping mechanism.
2586:
2587: We note that the distinction between the period distribution of
2588: planets around F and G stars is an interesting feature of the planet
2589: population, and might be accounted for by different tidal interactions
2590: with the parent stars, instead by the different masses of the
2591: disks. This is so because tidal interaction in F stars is
2592: substantially weaker than in G stars, due to the fact that F stars do
2593: not have convective envelopes. Thus, F stars can stop the inward
2594: migration only when the planet is substantially closer to the parent
2595: star than the stopping radius of G stars. Consequently, if tidal
2596: interaction with their parent stars plays a role in the final orbits
2597: of planets, the difference in the period distribution of planets
2598: around G and F stars {\it could} have been caused by internal stellar
2599: structure differences.
2600:
2601: The present discussion clearly demonstrates that the extra planets at
2602: an orbital period of around 3 days are still not acceptable evidence
2603: for the action of tidal interaction between short-period planets and
2604: their parent stars, and that further observations and theoretical work
2605: are still necessary.
2606:
2607: \subsubsection{Inward drift}
2608: %---------------------------
2609:
2610: Tidal interaction can also push the planet inwards, if the planet
2611: orbital frequency is faster than the stellar spin. In such a case, the
2612: planet is dragged back as a result of tidal interaction with the star, an
2613: interaction that takes angular momentum from the planetary orbital
2614: motion, which results in a decrease in the orbital radius of the
2615: star. The timescale for the inward drift is given by
2616: Eq.~\ref{radial_timescale} that describes the outward drift, but now
2617: $\dot{a}_p$ is negative. However, unlike the outward drift, the inward
2618: drift is a runaway process, in which $a_p$ gets smaller and therefore
2619: the drift timescale, $\tau_{radial}$, gets shorter, which accelerates the
2620: drift further (e.g., Rasio \& Ford \nocite{rasio1996b}1996).
2621:
2622: It is therefore convenient to define the corotation radius, in which
2623: the planetary orbital period equals the stellar rotational
2624: period. Outside this radius the planet is pushed outward, while within
2625: this radius the planet is pushed inward. If interaction with the disk
2626: succeeds in pushing the planet inside the corotation radius, then
2627: tidal interaction with the star joins forces with the disk interaction
2628: to push the planet inward. The planet can survive inside its
2629: corotation radius only if the disk disappears {\it and} the drift
2630: timescale is long enough. Rasio \& Ford concluded that for 51 Peg the
2631: timescale for the inward drift is shorter than the stellar lifetime
2632: only if the orbital period is shorter than about 10 hours. This means
2633: that we would not be able to find planets around stars similar to 51
2634: Peg with orbital periods shorter than 10 hours, which indeed is the
2635: case.
2636:
2637: We note that the stellar radii and rotational periods do change during
2638: the stellar lifetime, and both the corotation radius and the
2639: radial-drift timescale necessarily follow these variations. As the
2640: stellar rotational period gets longer and the corotation radius
2641: increases, the planet can find itself inside the corotation radius,
2642: being pushed inwards by the tidal interaction with its parent
2643: star. Such a planet can survive only if its orbital radius is large
2644: enough so the drift timescale is longer than the stellar
2645: lifetime.
2646:
2647: As of November 2007, eleven planets were detected with periods shorter
2648: than 2 days, the shortest of which is OGLE-TR-56b, with a period of
2649: 1.2 days (Konacki \etal\ \nocite{konacki2003}2003). The discussion
2650: above highlights the need to carefully consider the stability of the
2651: orbits of these planets against inward drift. Sasselov (2003), for
2652: example, tried to exclude some versions of the tidal theories that
2653: predict too fast radial drift for OGLE-TR-56b. He suggested that for
2654: some theories it would be possible to detect in the near future
2655: changes in the orbit of OGLE-TR-56b by precise timing of its transit.
2656: Such a measurement would be a landmark of the observational evidence
2657: for tidal interaction in main-sequence stars, enabling us to follow
2658: the results of the interaction in real time, and not only collect
2659: indirect proofs for its action in the past.
2660:
2661:
2662: %-----------------------------------------------------------
2663: \subsection{Circularization of extrasolar-planet orbits}
2664: %-----------------------------------------------------------
2665:
2666: In order to study the tidal circularization of extrasolar planets, we
2667: plotted in Figure~\ref{extrasolar_ecc} the eccentricities versus
2668: periods of all known planets, like we did in Figure~\ref{e_log_p} for
2669: all the known spectroscopic binaries. The main two features of
2670: Figure~\ref{e_log_p} can also be found in Figure~\ref{extrasolar_ecc}
2671: (see Halbwachs \etal\ \nocite{halbwachs2005}2005 for a detailed
2672: comparison). All planets with periods shorter than a 'cutoff' period
2673: are circular, and all the extrasolar planets with longer periods have
2674: eccentricities below an 'upper envelope', which starts at eccentricity
2675: zero and climbs up asymptotically to a certain high value. The values
2676: of the planetary upper envelope parameters are different from those of
2677: the spectroscopic binaries. The upper envelope starts to rise at a
2678: period of 2.5 days and rises to an eccentricity of 0.8 when moving
2679: towards 25 days.
2680:
2681: \begin{figure}
2682: \includegraphics[width=13cm]{extrasolarEvsP.eps}
2683: \caption{The eccentricity of the extrasolar planets as a function of
2684: the orbital period. The dashed line is the upper envelope, with an
2685: equation $f=E-A \exp(-(p B)^c)$, where $E=0.8$, $A=8$, $B=6$ and
2686: $C=0.35$.}
2687: \label{extrasolar_ecc}
2688: \end{figure}
2689:
2690: Figure~\ref{extrasolar_ecc} displays only three exceptions to these two
2691: features. Two of them are the very highly eccentric planets HD
2692: 80606b (Naef \etal\ \nocite{naef2001}2001) and HD 20782b (Jones \etal\
2693: \nocite{jones2006}2006), both with an eccentricity of 0.92, and the
2694: third one is HD 41004Bb --- a planet with one of the shortest periods,
2695: 1.3 days, with a small but significant eccentricity of $0.081\pm0.012$
2696: (Zucker \etal\ \nocite{zucker2003}2003; \nocite{zucker2004}2004). Note
2697: that the three eccentricities would not be considered exceptions
2698: for spectroscopic binaries, because the upper envelope for
2699: binaries reaches a value of 0.98, and the 'cutoff' period is as short
2700: as 0.35 days.
2701:
2702: HD 41004 is an interesting system which is composed of a double visual
2703: star, HD 41004 A and B, separated by about 20 AU. Each of the two
2704: stars has a low-mass companion. Component A has a planet with a
2705: period of about 1000 days, with a minimum mass of 2.5 Jupiter mass
2706: (=$M_J$), whereas component B has a companion with a short period and
2707: a small but significant eccentricity, with a minimum mass of 18 $M_J$
2708: (Zucker \etal\ 2004). We note that the general consensus puts the
2709: upper-mass limit for planets around 10--15 $M_J$, and therefore it is
2710: not clear if HD 41004Bb is a planet, or rather a low-mass companion
2711: which should be referred to as a brown-dwarf secondary. If HD 41004Bb
2712: is a brown-dwarf companion, it should be moved from
2713: Figure~\ref{extrasolar_ecc} to Figure~\ref{e_log_p}, where its
2714: location is below the upper envelope. Another possibility is that the
2715: visual companion, HD 41004A, pumped eccentricity into the orbit of Bb,
2716: as is discussed regarding triple systems in
2717: Section~\ref{subsection_triple_ecc}. A similar idea was already
2718: proposed for another high eccentricity planet, 16 Cyg B (Mazeh \etal\
2719: \nocite{mazeh1997}1997; Holman \etal\ \nocite{holman1997}1997). For a
2720: review of this idea see Mazeh (\nocite{mazeh2007}2007).
2721:
2722: The two prevailing features of Figure~\ref{extrasolar_ecc} might be
2723: attributed to tidal circularization acting on the extrasolar planets,
2724: exactly as the general consensus interprets Figure 1. However,
2725: concealed behind the model for the eccentricity of the spectroscopic
2726: binaries is the assumption that all binaries, without any period
2727: difference, have been formed with the same eccentricity distribution,
2728: including the highly eccentric binaries. Therefore, the fact that
2729: short-period binaries are all circular must be the result of some
2730: circularization processes occurring {\it after} the binary
2731: formation. On the other hand, the prevailing assumption for planets
2732: was, until just a few years ago, that all planets should have circular
2733: or at least almost circular orbits, like all the planets in our solar
2734: system. If planets are formed in an accretion disk, then the 'natural'
2735: outcome is planets with circular orbits, as the interaction of the
2736: planetesimal with the gas and dust in the disk should dissipate any
2737: initial eccentricity. This is one of the reasons why HD114762b,
2738: detected already in 1989 (Latham \etal\ 1989) \nocite{latham1989} with
2739: an eccentricity of 0.4, was not considered as a planet candidate at
2740: that time. Therefore, the discovery of planets with large
2741: eccentricities was a surprise to the astronomical community. Thus,
2742: theoretical effort was {\it not} devoted to explaining the circular
2743: short-period planets, but instead to building a reasonable scenario
2744: for the unexpected eccentricities of the long-period planets.
2745:
2746: Some studies associated the observed eccentricities of extrasolar
2747: planets with another surprising feature of the planets --- their small
2748: semi-major orbital axis. As discussed in the previous subsection, the
2749: assumed paradigm to explain the short distances of the short-period
2750: extrasolar planets is an inward migration driven by interaction with the
2751: disk. The effect of this interaction on
2752: the planetary orbital eccentricity (e.g., Goldreich \& Sari
2753: \nocite{sari2003}2003) is still a matter of controversy (see
2754: discussion and short review by Moorhead \& Adams
2755: \nocite{moorhead2007}2007). Other studies tried to explain the
2756: existence of short-period planets and the planetary eccentricities as
2757: a consequence of interaction with other, still undetected, planets
2758: (e.g., Weidenschilling \& Marzari \nocite{marzari1996}1996; Rasio \&
2759: Ford \nocite{rasio1996}1996; Zhou \& Lin \nocite{zhou2007}2007; Juric
2760: \& Tremaine \nocite{juric2007}2007). It is not clear if these
2761: scenarios could explain the distribution of eccentricities in
2762: long-period planets and the circular orbits of short-period planets at
2763: the same time. If this is not possible, and the assumed mechanism that
2764: pumped eccentricities into the long-period orbits also necessarily
2765: built high eccentricity in some of the short-period planets, then we
2766: might need to invoke tidal circularization to account for the circular
2767: orbits of the short-period planets.
2768:
2769: The circularization of a planetary orbit can come from
2770: dissipative processes that take place either in the star or in the planet.
2771: The circularization timescale due to processes in the {\it planet} is
2772: usually written (e.g., Mardling 2007) as
2773:
2774: \begin{equation}
2775: \tau_{circ,p}=\frac{2}{21n_p} \left(\frac{Q_p}{k_p}\right)
2776: q \left(\frac{a_p}{r_p}\right)^5
2777: \ ,
2778: \end{equation}
2779: %
2780: where $Q_p$ is the planetary dissipation parameter, $k_p$ is the
2781: planetary tidal Love number (Goldreich \& Soter \nocite{gold1966}1966)
2782: and $r_p$ is the planetary radius (See Mardling
2783: \nocite{mardling2007}2007 for an essential discussion of the slightly
2784: different definitions of the Q parameters by the different studies).
2785: For the corresponding timescale due to processes in the star,
2786: $\tau_{circ,*}$, one should replace $Q_p$, $k_p$, $r_p$ by the
2787: corresponding values of the star, $Q_*$, $k_*$, $R_*$, and switch
2788: between $m_p$ and $M_*$ (e.g., Carone \& P{\"a}tzold
2789: \nocite{carone2007}2007). Unfortunately, as was already noted in the
2790: previous subsection, the theoretical estimate of
2791: the ratios $Q_p/k_p$ or $Q_*/k_*$ varies over one or two orders of
2792: magnitude (e.g., Lin \etal\ 1996; Trilling \etal\
2793: \nocite{trilling1998}1998; Trilling \nocite{trilling2000}2000). For
2794: example, Trilling (2000) used $Q_* \geq 10^6$ for main-sequence
2795: stars and $1.5\times 10^4$ -- $1.5\times 10^4$ for PMS
2796: stars. Therefore, the theory is in a 'safe' situation, in which it can
2797: not be confronted with the observations.
2798:
2799:
2800: \subsubsection{Small eccentricity induced by a second planet}
2801: \label{subsection_small_e}
2802: %-------------------------------------------------------------
2803:
2804: In Section~\ref{subsection_MS} we discussed the Mazeh \& Shaham (1979)
2805: effect, in which a third distant companion can induce eccentricity
2806: in the binary orbit, even for a short-period orbit for which the
2807: tidal circularization had reduced the binary eccentricity. It was
2808: emphasized that the Mazeh \& Shaham effect can work even if the binary
2809: eccentricity starts at exactly zero value. This effect can have
2810: an impact on the evolution of close binaries, as the injected binary
2811: eccentricity invokes frictional forces inside the two stars that
2812: dissipate rotational energy and transfer angular momentum between the
2813: stellar rotation and the binary orbital motion.
2814:
2815: Similar ideas have been proposed with regard to extrasolar planets. It
2816: was suggested that another, as-of-yet undetected planet pumps small
2817: eccentricity into the orbit of some known short-period planets. One
2818: such example is HD 209458b (Mazeh \etal\ \nocite{mazeh2000}2000), the
2819: first transiting planet (Charbonneau \etal\ \nocite{charbonneau2000}2000;
2820: Henry \nocite{henry2000}2000) that was found to have a somewhat
2821: surprisingly large radius (e.g., Burrows \etal\
2822: \nocite{burrows2000}2000). The conjecture was that a small
2823: eccentricity of HD 209458b orbit (e.g., Laughlin \etal\
2824: \nocite{laughlin2005}2005) causes tidal dissipation inside the planet,
2825: which serves as another source of energy that inflates the planet
2826: (e.g., Bodenheimer \etal\ \nocite{bodenheimer2001}2001; Bedenheimer
2827: \etal\ \nocite{bodenheimer2003}2003; Gu \etal\ \nocite{gu2004}2004;
2828: Mardling 2007). Two other examples are WASP-1b (Cameron \etal\
2829: \nocite{cameron2007}2007) and HAT-P-1b (Bakos \etal\
2830: \nocite{bakos2007b}2007b).
2831:
2832: The idea of a planetary eccentricity pumped by another as-of-yet
2833: undetected planet can be verified or refuted by further
2834: observations. Additional accurate radial-velocity observations can
2835: constrain the orbital eccentricity of the known short-period planets
2836: as well as detect unknown additional planets. Another avenue for
2837: accurate derivation of orbital eccentricity of the transiting
2838: planets is to detect the secondary eclipse, an approach recently
2839: carried out in the IR by the {\it Spitzer} satellite (e.g., Deming
2840: \etal\ \nocite{deming2005}2005). When the orbital parameters of those
2841: planets are determined, we will be able to assess the validity of the
2842: idea of the pumping eccentricity, particularly for transiting planets
2843: for which we can determine the radius and eccentricity.
2844:
2845: %-------------------------------------------------
2846: \subsection{Synchronization of planetary rotation}
2847: %-------------------------------------------------
2848:
2849: Rotational synchronization of stars with short-period planets was not
2850: considered by the astrophysical community, as it was assumed that the
2851: masses of the planets are too small to synchronize the stellar rotation
2852: with the planetary orbital frequency. See, for example, the discussion
2853: of the stopping mechanism (Section~\ref{subsubsection_stopping}),
2854: where we reviewed two cases with stellar rotational periods much longer
2855: than the orbital periods. At the same time, it was taken for granted
2856: that tidal dissipation inside the short-period planets has synchronized
2857: their own rotation with their orbital frequency (e.g., Rasio \etal\ 1996;
2858: Ogilvie \& Lin \nocite{ogilvie2004}2004).
2859:
2860: Lubow \etal\ (\nocite{lubow1997}1997) claimed that even a planet with
2861: a period of 200 days should be synchronized in 10 Gyrs (see also
2862: Ivanov \& Papaloizou 2007 and Ogilvie \& Lin
2863: \nocite{ogilvie2007}2007). As is commonly known, we find many planets
2864: with eccentric orbits at that period (see, for example,
2865: Figure~\ref{extrasolar_ecc}). Therefore, we expect to have planets
2866: with orbital periods shorter than 200 days with
2867: pseudo-synchronized rotational periods (see detailed derivation of the
2868: pseudo-synchronization period by Ivanov \& Papaloizou
2869: \nocite{ivanov2007}2007).
2870:
2871: How can we observe planetary synchronization or
2872: pseudo-synchronization? Planets that are rotationally locked in a
2873: circular orbit have one hemisphere exposed to intense stellar
2874: insolation all the time, causing a significant difference in
2875: temperature between their day and night sides. The infrared emission
2876: of the day side was already detected by {\it Spitzer} through the
2877: secondary eclipse of HD 209458b (Deming \etal\
2878: \nocite{deming2005}2005; 2007\nocite{deming2007}), TrES-1 (Charbonneau
2879: \etal\ \nocite{charbonneau2005}2005), 189733b (Deming \etal\
2880: \nocite{deming2006}2006), and Gliese 436b (Deming \etal\
2881: \nocite{deming_436_2007}2007; Demory \etal\ 2007), and even by the
2882: phase modulation of the non-transiting planet around $\upsilon$ And
2883: (Harrington \etal\ \nocite{harrington2006}2006). The exact
2884: periodicity and the phase of the IR modulation can confirm the
2885: synchronization assumption (e.g., Seager \etal\
2886: \nocite{seager2005}2005). Furthermore, IR modulation can reveal
2887: a signature of pseudo-synchronized planetary rotation.
2888:
2889: As-of-yet, {\it Spitzer} detected IR modulations only of planets with
2890: circular or almost circular orbits with short periods, of the order
2891: of 3 days. Very recently the transit of a rather long-period
2892: eccentric planet, HD 17156b, has been discovered (Barbieri \etal\
2893: \nocite{barbieri2007}2007). It would be quite interesting to follow
2894: the IR modulation of this planet and find out if the planetary
2895: rotation is pseudo-synchronized with the orbital period of 21.2 days
2896: and with an eccentricity of 0.67.
2897:
2898: %---------------------------------------------------------------
2899: \subsection{Alignment of planetary motion with stellar rotation}
2900: %---------------------------------------------------------------
2901:
2902: The {\it a priori} approach to extrasolar planetary alignment was
2903: similar to the early attitude to planetary eccentricity. Based on the
2904: solar system features, it was expected that the planetary orbital
2905: plane of motion would be aligned with the stellar rotational axis. The
2906: theory predicted that the dissipative processes in the protoplanetary
2907: disk tend to damp the initial inclinations of the planetesimals (e.g.,
2908: Ward \& Hahn \nocite{ward1994}1994; see also a later study by
2909: Cresswell \etal\ \nocite{cresswell2007}2007). However, in the spirit
2910: of the new approach to the formation of planetary eccentricities by
2911: the interaction with the disk, Thommes \& Lissauer
2912: (\nocite{thommes2003}2003), for example, suggested that when more than
2913: one planet is present in the system, the interaction with the material
2914: of the disk and the resonant interaction between the planets can turn
2915: the system noncoplanar. Interestingly, spin-orbit misalignment might
2916: cause tidal heating of the planet and contribute to its inflated
2917: radius, as was suggested by Winn \& Holman (\nocite{winn_holman2005}2005)
2918: when trying to account for the large radius of HD 209458b (see
2919: Section~\ref{subsection_small_e}). However, this idea was challenged
2920: by Levrard \etal\ (\nocite{levrard2007}2007) and Fabrycky \etal\
2921: (\nocite{fabrycky2007}2007).
2922:
2923: To observe the relative inclination of the planetary spin axis, we can
2924: use the RM effect for transiting planets (see Gaudi \& Winn
2925: \nocite{gaudi2007}2007 for a thorough discussion of the effect and
2926: Section~\ref{rm_effect} for a discussion of the same effect observed
2927: in eclipsing binaries). The amplitude of the radial-velocity
2928: modulation of the RM effect is
2929: \begin{equation}
2930: \Delta_{RM} \sim V_*\sin i_* \left(\frac{r_p}{R_*}\right)^2 \ ,
2931: \end{equation}
2932: %
2933: where $ V_*\sin i_*$ is the rotational broadening of the stellar
2934: spectral lines. The RM amplitude of transiting planets is necessarily
2935: small, as $r_p/R_*\sim0.1$. It can be as small as $\sim 10$ m
2936: s$^{-1}$, if $V_*\sin i_*$ is 1 km s$^{-1}$. Nevertheless, the effect
2937: can have an amplitude as large as $\sim 50$ m s$^{-1}$, if $V_*\sin
2938: i_*$ is 5 km s$^{-1}$. Such amplitude can be easily observed with
2939: the typical precision achieved for the study of extrasolar planets.
2940:
2941: Recent observations (e.g., Queloz \etal\ \nocite{queloz2000}2000; Winn
2942: \etal\ \nocite{winn2006}2006b; Narita \nocite{narita2007}2007) that
2943: aimed specifically to follow the RM effect in a few transiting planets
2944: could not find any evidence for spin-orbit misalignment. This is true
2945: even for 147506b (HAT-P-2b; Winn \etal\ \nocite{winn2007}2007;
2946: Loeillet \etal\ \nocite{loeillet2007}2007) for which a substantial
2947: eccentricity has been detected (Bakos \etal\
2948: \nocite{bakos2007}2007a). Unfortunately, as discussed above, the
2949: theoretical interpretation of these findings is not clear. It could be
2950: attributed to the formation processes --- planets are formed in
2951: aligned orbits, or it could be the outcome of some tidal dissipation
2952: between the spinning star and the planetary motion. We are still
2953: awaiting the discoveries of misaligned planets, which will help to
2954: clarify the broad scope of planetary alignment.
2955:
2956: %=====================================================
2957: %
2958: \section{In the era of large-scaled photometric surveys}
2959: %
2960: %=====================================================
2961:
2962: The advent of sensitive, large CCDs in the service of the astronomical
2963: community in the last decade or so has made large-scaled photometric
2964: surveys possible, yielding hundreds of accurate photometric
2965: measurements of many thousands of stars. Although the drive for these
2966: surveys was not always the study of binary stars, and eclipsing
2967: binaries in particular (e.g., MACHO, see Alcock \etal\ 1977), hundreds
2968: of eclipsing binaries have been discovered as a result of these
2969: surveys (e.g., Devor 2005\nocite{devor2006}).
2970:
2971: In the past, even when a star was discovered to be an eclipsing
2972: binary, a special, additional observational effort was needed in order
2973: to acquire its lightcurve in order to derive its orbital
2974: parameters. In contrast, for eclipsing binaries discovered by large
2975: surveys we get, without any additional observational effort,
2976: lightcurves of unprecedented precision that are spread over a long
2977: period of time. Therefore, the present large-scaled photometric
2978: surveys are revolutionizing the study of eclipsing binaries, yielding
2979: new tools for the study of tidal interaction in particular. This
2980: section reviews two studies that have used two large surveys, MACHO
2981: and OGLE, which performed an intensive monitoring of the LMC for a few
2982: years. Both MACHO (Alcock \etal\ \nocite{alcock1997}1997; Derekas
2983: \etal\ \nocite{derekas2007}2007)) and OGLE (Udalski \etal\
2984: \nocite{udalski1998}1998; Wyrzykowski \etal\ \nocite{wyr2004}2004)
2985: data sets were used to identify the largest sample of eclipsing
2986: binaries ever studied, with direct implications for the study of their
2987: tidal interaction. The goal of this section is to demonstrate the
2988: far-reaching potential of these surveys.
2989:
2990: \subsection{Apsidal motion of eccentric binaries in the LMC}
2991: %----------------------------------------------------------------
2992:
2993: Michalska \& Pigulski \nocite{michalska2005} (2005) studied the
2994: discovered binaries in the LMC by the OGLE survey, and searched for
2995: binaries that are suitable for distance
2996: determination. Serendipitously, they found 14 binaries with
2997: significant apsidal precession. The LMC data spanned 2000 days,
2998: and therefore they could detect slow variation of the longitude of the
2999: periastron and derive long apsidal periods, of the order of thousands
3000: of years. In a follow-up paper, Michalska (\nocite{michalska2007}2007)
3001: found evidence for apsidal motion in the data of an additional 11
3002: binaries.
3003:
3004: \begin{figure}
3005: \includegraphics[width=13cm]{LMC_apsidal.eps}
3006: \caption{The apsidal precession period as a function of the binary
3007: period for the LMC binaries. Binaries discussed by Michalska \&
3008: Pigulski (2005) are denoted by filled circles, and those discussed by
3009: Michalska (2007) by $\ast$. A dashed line connects the binaries sorted
3010: by period. Two straight lines were fitted, one for
3011: the first 14 short-period binaries, up to a period of 3.2 days, and
3012: the other for the binaries with longer periods. The expected GR
3013: precession is also plotted for binaries with total masses of 10
3014: $M_{\odot}$ and for almost circular orbits.}
3015: \label{apsidal_motion_LMC}
3016: \end{figure}
3017:
3018: Unfortunately, the radii and masses of the components of the binaries
3019: of Michalska \& Pigulski and Michalska are not known, and therefore no
3020: comparison between the theory of apsidal motion induced by the tidal
3021: interaction and the observations is possible. However, we can try to
3022: take advantage of the fact that for the OGLE LMC data, most stars are
3023: in a limited magnitude range, and therefore their radii and masses are
3024: not very different from one another. Therefore, the apsidal precession
3025: period should depend mostly on the binary period. To check this
3026: hypothesis we plotted in Figure~\ref{apsidal_motion_LMC} the derived
3027: apsidal period as a function of the binary period. The period range is
3028: between about $2$ and $7$ days, and the apsidal period range is
3029: between $50$ and $2000$ years. For the binaries with periods below $3$
3030: days, the apsidal period shows a linear dependence on the binary
3031: period, as if the ratio $R_1/a$ in Eq. (3.1) is constant, where $R_1$
3032: is the radius of the primary and $a$ is the semi-major axis. For the
3033: binaries with periods longer than $3$ days the apsidal period is
3034: proportional to $ P^{11/3}$, where $P$ is the binary period. Within the
3035: errors this slope is consistent with the exponent of Equation (3.4).
3036:
3037: We note that one system, with a period of 6.33 days, has a substantially
3038: shorter precession period than the straight-line fit. This is OGLE LMC
3039: 05065201-6825466, the brightest star in the sample of 98 binaries
3040: analysed by Michalska \& Pigulski. This star, with a brightness of $V=
3041: 14.04$, probably has a substantially larger radius than the other
3042: stars in Michalska \& Pigulski and Michalska samples, and therefore
3043: might have a shorter apsidal period.
3044:
3045: The Michalska \& Pigulski and Michalska studies comprise the first
3046: systematic attempt to consider the apsidal motion in binaries detected
3047: by large-scaled systematic photometric surveys. We anticipate many more
3048: similar studies in the future.
3049:
3050: %-------------------------------------------------------------
3051: \subsection{Circularization of eclipsing binaries of the
3052: SMC and LMC}
3053: \label{subsection_circ_LMC}
3054: %-------------------------------------------------------------
3055:
3056: A seminal work that used the new sets of binaries to study tidal
3057: interaction was the study of North \& Zahn (\nocite{north2003}2003),
3058: which analysed the detached eclipsing binaries cataloged by Udalski
3059: \etal\ (1998) in the SMC and by Alcock \etal\ (1997) in the LMC.
3060: Because of the magnitude range of the two surveys, all the binaries in
3061: these two samples have early-type primaries, mostly B-type stars
3062: (North \& Zahn 2003). Following Giuricin \etal\ (1984; see a review of
3063: their work in Section~\ref{subsection_early_type}), North \& Zahn
3064: plotted the orbital eccentricity as a function of the fractional radii
3065: of the eclipsing binaries in the SMC and LMC. Like Giuricin \etal\
3066: they tried to find out the cutoff fractional radius, above which all
3067: binaries are circularized. They were also trying to find out whether
3068: there is any difference between the circularization processes in
3069: early-type stars in our Galaxy and those in action in binaries in the
3070: SMC and LMC.
3071:
3072: North \& Zahn pointed out that with the precision of the two surveys
3073: it was difficult to derive the orbital eccentricity from the lightcurve
3074: of an eclipsing binary. The phase of the secondary eclipse, which could
3075: be accurately determined, yielded only $e\cos\omega$, where $e$ is the
3076: eccentricity and $\omega$ is the longitude of the periastron. North \&
3077: Zahn overcame this problem by plotting the absolute value of
3078: $e\cos\omega$ instead of the eccentricity itself. As they were
3079: interested only in the statistical features of the sample, and in the
3080: cutoff fractional radius in particular, the fact that only
3081: $e\cos\omega$ was available to the analysis did not hamper their study.
3082:
3083: North \& Zahn faced another problem. The given precision of the
3084: lightcurves of the two surveys did not really allow them to derive the
3085: radii of the two components of each binary. One could only get the sum
3086: of the two radii (in terms of the orbital separation), while the ratio
3087: of the two radii was not well determined. North \& Zahn had therefore
3088: to {\it assume} that the two stars have equal radii. Only with the aid of this
3089: assumption could they estimate the primary radius, which is, in fact,
3090: the average of the two radii. With these two modifications, the
3091: analysis of North \& Zahn (2003) yielded results similar to Giuricin
3092: \etal\ (1984). The cutoff fractional radius was about 0.25, with no
3093: statistical difference between the early-type binaries of the Galaxy
3094: and those of SMC and LMC.
3095:
3096: In a follow-up paper, North \& Zahn (\nocite{north2004}2004) analysed
3097: many additional eclipsing-binary lightcurves, so that their samples
3098: included 148 binaries in the SMC and 353 binaries in the LMC. They
3099: then found, again, that the critical fractional radius was
3100: in the range $0.24-0.26$, irrespective of mass, surface gravity and
3101: metallicity. This value of the critical radius was consistent with the
3102: previous studies, and is compatible with the theory of Zahn (1975).
3103:
3104: North \& Zahn (2003; 2004) did not use all the binaries detected in
3105: the OGLE LMC data. In fact, Wyrzykowski \etal\ (2004) identified 2580
3106: eclipsing binaries in the LMC OGLE II photometric data alone. However,
3107: it is impractical to manually analyse such a large number of
3108: binaries. One would need an automated analysis to handle such a large
3109: set of binaries. Such algorithms were indeed constructed recently
3110: (e.g., Wyithe \& Wilson \nocite{wyithe2001}2001; Devor \& Charbonneau
3111: \nocite{devor2005}2006). Although the automated codes can err in some
3112: cases, for most of the binaries the algorithms yield the right
3113: solutions, rendering them powerful tools in the
3114: study of the statistical features of the close binary population
3115: (e.g., Mazeh \etal\ 2006).
3116:
3117: Following this trend, Tamuz \etal\ (\nocite{tamuz2006}2006)
3118: constructed an automated algorithm, EBAS, to analyze eclipsing binary
3119: lightcurves, an algorithm which is based on the EBOP code (Popper \&
3120: Etzel \nocite{popper1981}1981; Etzel \nocite{etzel1981}1981). The
3121: algorithm was then used by Mazeh \etal\ (\nocite{mazeh2006}2006) to
3122: analyze most of the eclipsing binaries detected by Wyrzykowski \etal\
3123: (2004) in the OGLE LMC data. Mazeh \etal\ then plotted the
3124: eccentricity as a function of the fractional radius, with two minor
3125: changes relative to the plot of North \& Zahn. First, instead of the
3126: primary fractional radius, they plotted the total fractional radii
3127: $R_{total}=R_1+R_2$ --- where $R_1$ and $R_2$ are the fractional radii
3128: of the primary and secondary, respectively. Second, they specifically
3129: plotted $e\cos\omega$ instead of its absolute value. They found that
3130: the critical total fractional radius is 0.5, consistent with the North \&
3131: Zahn result.
3132:
3133: \begin{figure}
3134: \includegraphics[width=13cm]{LMC_EB.eps}
3135: \caption{The eccentricity, multiplied by $\cos\omega$, of the OGLE
3136: eclipsing binaries of the LMC as a function of total fractional
3137: radius. Included are 1335 binaries with $17\leq I \leq 19$. The
3138: dashed line represents the upper envelope (see text).}
3139: \label{LMC}
3140: \end{figure}
3141:
3142: Here we follow Mazeh \etal\ (2006) and plot in Figure~\ref{LMC} $e
3143: \cos \omega$ as a function of the total fractional radius for all 1335
3144: binaries with $19\leq I \leq 17$ analysed by Tamuz \etal\ (2005). We
3145: also plotted a straight line upper envelope, that goes like
3146:
3147: \begin{equation}
3148: \pm e\cos\omega=0.857-1.714R_{total} \ ,
3149: \end{equation}
3150: %
3151: which implies, if we ignore the $\cos\omega$ factor, that the periastron
3152: distance, $1-e$, in terms of the semi-major axis, is
3153:
3154: \begin{equation}
3155: 2(R_1+R_2) \stackrel{<}{\sim} 1-e
3156: \end{equation}
3157: %
3158: (Mazeh \etal\ 2006). This could mean that if the two stars got too
3159: close at periastron, to the extent that the distance between them was
3160: less than twice their total radius, then circularization processes went into
3161: action and the eccentricity was reduced until the two stars became further
3162: apart. This is not the only interpretation. One can imagine that
3163: binaries were not formed too close together, lest they violate this
3164: relation. Obviously, the present distribution could be the
3165: caused by the two scenarios --- formation and tidal evolution.
3166:
3167: Note that similarly to Giuricin \etal\ (see
3168: Section~\ref{subsection_early_type}), we also find a few binaries with a
3169: primary radius larger than 0.3 which display small but significant
3170: eccentricities. We also tend to attribute these eccentricities to
3171: unseen third distant companions, an effect suggested by Mazeh \&
3172: Shaham (1979; see detailed discussion in Section~\ref{subsection_MS}).
3173:
3174: The North \& Zahn (2003; 2004) and Mazeh \etal\ (2006) studies were the
3175: first to use large-scaled photometric surveys to touch upon
3176: circularization interaction. We anticipate many more, similar studies
3177: in the future.
3178:
3179: %======================
3180: %
3181: \section{Discussion}
3182: %
3183: %======================
3184:
3185: This paper has attempted to review a large corpus of observational
3186: evidence for tidal interaction in close binary systems, including
3187: short-period extrasolar planets. We first discussed two tidal effects
3188: that can be observed directly --- the ellipsoidal variation and the
3189: apsidal motion. Unfortunately, these two effects have been detected in
3190: the past in only a small number of stars, because the detection
3191: required special photometric observations which involved a
3192: considerable effort. The small number of known binaries with detected
3193: ellipsoidal modulation or apsidal motion allowed for only a limited
3194: confrontation between the theory and the observations. To perform a
3195: comprehensive confrontation we need many more binaries with accurate
3196: measurements of these two effects.
3197:
3198: Ellipsoidal modulation was mostly used in the past to study known
3199: spectroscopic binaries (e.g., HD 149420 --- Fekel \& Henry
3200: 2006)\nocite{fekel2006}, or binaries with compact objects. This
3201: situation is being changed dramatically, as ground-based large-scaled
3202: photometric surveys have been operated in the last decade. We already
3203: discussed two such 1-m class exhaustive projects, MACHO (Alcock \etal\
3204: 1993\nocite{alcock1993}) and OGLE (e.g., Udalski \etal\
3205: 1997\nocite{udalski1997}). Furthermore, a new wave of observational
3206: effort to search for transiting planets by small, 10-cm class,
3207: telescopes is now at its full thrust (e.g., HATnet --- Bakos \etal\
3208: \nocite{bakos2004}2004; XO --- McCullough \etal\
3209: \nocite{mccullough2005}2005; WASP --- Pollacco \etal\
3210: \nocite{pollacco2006}2006; TrES --- Alonso \etal\
3211: \nocite{alonso2007}2007). These projects are collecting an
3212: unprecedented number of photometric datasets that harbor thousands of
3213: ellipsoidal binaries.
3214:
3215: The amplitude of the ellipsoidal modulation can be of the order of a
3216: few percent, and therefore should be easily detected in these large
3217: photometric datasets. Detecting ellipsoidal variables can become a
3218: tool for discovering new binaries. In such cases, one must distinguish
3219: between the ellipsoidal modulation and other possible variations,
3220: particularly stellar spots and pulsation. For that purpose, one must
3221: rely on the theoretical shape of the modulation and on observations in
3222: different passbands. A first step towards this goal has been performed
3223: already by Derekas \etal\ (2006\nocite{derekas2006}), who identified
3224: ellipsoidal variables in the sample of variable pulsating red giants
3225: within the MACHO data.
3226:
3227: The huge datasets are having another effect on the study of close
3228: binaries: they are already revealing a large number of new eclipsing
3229: binaries (e.g., Devor 2005). Therefore the number of known eclipsing
3230: binaries will increase in the next few years by one or two orders of
3231: magnitude. The tens of thousands of new eclipsing binaries will
3232: revolutionize our understanding of close binaries in general and their
3233: statistical features in particular. These datasets will reveal many
3234: other effects of eclipsing binaries, such as apsidal motion and third
3235: distant companions through the time-travel effect. We will then be
3236: well equipped to confront the theory of apsidal motion with observations,
3237: and learn more about the frequency and distribution of triple systems.
3238: We will be able to verify conjectures regarding the correlation
3239: between short-period binaries and distant companions.
3240:
3241: The influx of photometric data is about to increase, as more automated
3242: photometric surveys with 1-m class telescopes will set into
3243: action. This includes SkyMaper (Bayliss \& Sackett
3244: 2007\nocite{bayliss2007}), Las Cumbres Observatory Global Telescope
3245: (Brown \etal\ \nocite{brown2007}2007), Pan-STARRS (e.g., Holman \etal\
3246: \nocite{holman2007}2007), and finally LSST (e.g., Strauss \etal\
3247: 2006\nocite{strauss2006}). One such binary was discovered already by
3248: Blake \etal\ (\nocite{blake2007}2007), using the Sloan Digital Sky
3249: Survey II. This influx of photometric data will be complemented by
3250: extremely accurate datasets from space missions, which will have an
3251: immediate impact on binary studies. This trend has already been
3252: started by HIPPARCOS (e.g., Jorissen \etal\
3253: \nocite{jorrisen2004}2004), and will be joined by CoRoT (Maceroni and
3254: Ribas \nocite{maceroni2007}2007), Kepler (Koch \etal\
3255: \nocite{koch2007}2007) and GAIA (e.g., Niarchos
3256: 2006\nocite{niarchos2006}) missions. For the future of binary studies
3257: in the 'brave new era,' see a review by Guinan \& Engle
3258: (\nocite{guinan2006}2006).
3259:
3260: We have discussed the long-term tidal processes of circularization,
3261: synchronization and spin alignment. As was reviewed, binary
3262: circularization has attracted many studies, particularly of the
3263: transition period between circular and eccentric
3264: binaries. Nevertheless, no consensus has been reached with regard to
3265: the theoretical timescale of the tidal processes, and even the
3266: dominant dissipation process behind the circularization is still being
3267: debated. Synchronization and alignment were studied much less. For
3268: example, the number of eclipsing binaries with observed RM effect by
3269: which we study the spin-orbit alignment is only six, an extremely
3270: small number. Modern spectrographs with efficient digital detectors
3271: can observe many more eclipsing binaries, and the development of tools
3272: to analyze the observations can enable us to accurately derive the
3273: spin-orbit inclination. The results may well be crucial for our
3274: understanding of binary formation, as this will help us estimate the
3275: primordial alignment of binary population. In case binaries are formed
3276: with some spread of spin-orbit inclinations, we should also expect
3277: coeval samples of binaries to exhibit a transition period between
3278: aligned and non-aligned binaries. Such an observed transition can
3279: serve as another confirmation of the theory of tidal interaction. It
3280: would be of extreme importance to compare the aligned/non-aligned
3281: transition period with the pseudo-synchronized/non-pseudo-synchronized
3282: and circularized/eccentric transition periods.
3283:
3284: An additional feature of the new large accurate photometric datasets
3285: is their long time span, of the order of a few years, that will enable
3286: us to detect minute variations which disclose long-term variations.
3287: The search of Michalska \& Pigulski (2005) and Michalska (2007) for
3288: evidence of apsidal motion discussed above is one example. The works
3289: of Kubiak \etal\ (\nocite{kubiak2006}2006) and Pilecki \etal\
3290: (\nocite{pilecki2007}2007) are two other examples of studies that
3291: utilized the long time span of the new data sets. Some of the
3292: yet-to-come studies will be able to find eventually direct evidence
3293: for long-term tidal processes in close binaries.
3294:
3295: Finally, we have discussed in brief some ideas about tidal evolution
3296: of short-period extrasolar planets. Many of the short-period planets
3297: are being discovered by photometry which detects how these planets
3298: eclipse their parent stars. The transiting planets hold a higher
3299: potential for the study of tidal effects, as we know so much more
3300: about them, including their masses and radii. They can be used as
3301: clean and unique laboratories to study tidal effects, as the
3302: complexity of their systems is reduced because of the small masses and
3303: radii of the planets. As of October 2007, 27 transiting planets are
3304: known, which is about half of the short-period extrasolar planets, the
3305: others have been discovered by radial-velocity measurements. We
3306: anticipate that the balance between the radial-velocity and transiting
3307: planets will change in the near future in favour of the transiting
3308: planets. The rapidly increasing number of transiting planets will
3309: enable us to study the possible correlation between presumed tidal
3310: effects and the stopping mechanism, the inflated radii of some
3311: planets, synchronization of the stars and planets and their spin-orbit
3312: alignment. A comprehension of these features will help us assess the
3313: role of tidal interaction in the orbital evolution of short-period
3314: extrasolar planets.
3315:
3316:
3317: %=============================
3318: %
3319: \section*{Acknowledgements}
3320: %
3321: %=============================
3322:
3323: Special thanks to Jean-Paul Zahn for inviting me to the Oleron summer
3324: school to give a series of lectures on the subject of this
3325: review. This paper emerged from those lectures. I am deeply thankful
3326: for his support and patience during the long period of writing the
3327: manuscript and for his enlightening comments. I am indebted to Soeren
3328: Meibom and Itzhak Goldman for many fruitful discussions and
3329: enlightening comments. Special thanks to Peter Eggleton, Dan Fabrycky,
3330: Scott Gaudi, Mike Lecar, Doug Lin, Andrei Tokovinin and Guillermo
3331: Torres who read an early version of the manuscript and send me
3332: thoughtful comments and suggestions. I wish to heartily thank my
3333: students Avi Shporer, Aviv Ofir and Yevgeny Tsodikovich for technical
3334: help with handling the data and with the plotting.
3335:
3336: This work was supported by the Israeli Science Foundation through
3337: grant no. 03/233. I thank the Swiss National Science
3338: Foundation and the Smithsonian Institution for supporting my stay
3339: at the Geneva Observatoire and at the CfA, during which I worked on
3340: this paper.
3341:
3342:
3343: \bibliographystyle{astron}
3344: \bibliography{57}
3345:
3346: \end{document}
3347:
3348:
3349: This paper reviews the rich corpus of observational evidence for tidal
3350: effects in binaries, mostly based on photometric and radial-velocity
3351: measurements.
3352:
3353: We review photometric evidence for ellipsoidal variability in systems
3354: whose binarity is known, with compact objects in particular. We
3355: discuss apsidal motion observed in eclipsing binaries, for which the
3356: timing of the secondary eclipse can determine accurately the longitude
3357: of the periastron. The tidal apsidal precession can be obtained in
3358: these systems only if we are able to subtract the other factors that
3359: can drive the apsidal precession.
3360:
3361: Among the long-term effects, circularization was studied the most, and
3362: a transition period between circular and eccentric orbits has been
3363: derived for eight coeval samples of binaries. The transition periods
3364: support the idea of tidal circularization and could, in principle,
3365: determine an absolute scaling of the efficiency of the long-term
3366: circularization processes. The study of synchronization and spin-orbit
3367: alignment is less advanced. As binaries are supposed to reach
3368: synchronization before circularization, one can expect finding
3369: eccentric binaries in pseudo-synchronization state, the evidence for
3370: which is reviewed. We also discuss synchronization in PMS and young
3371: stars, and compare the emerging timescale with the circularization
3372: timescale. The paper reviews the Rossiter-McLaughlin effect and its
3373: potential to study spin-orbit alignment.
3374:
3375: The paper discusses the tidal interaction in close binaries that are
3376: orbited by a third distant companion, and reviews the effect of
3377: pumping the binary eccentricity by the third star. We elaborate on the
3378: impact of the pumped eccentricity on the tidal evolution of close
3379: binaries residing in triple systems, and review the idea that such an
3380: interaction can shrink the binary separation.
3381:
3382: The paper discusses the extrasolar planets and the
3383: observational evidence for tidal interaction with their parent
3384: stars. This includes a mechanism that can induce radial drift of
3385: short-period planets, either inward or outward, depending on the
3386: planetary radial position relative to the corotation radius. This
3387: effect can stop the inward migration driven by the interaction with
3388: the protoplanetary disk, or can push the planet into the surface of
3389: the star. Another effect is the circularization of planetary orbits,
3390: the evidence for which can be found in eccentricity-versus-period plot
3391: of the planets already known.
3392:
3393: The paper discusses the revolution of the study of binaries that is
3394: currently taking place, which has been driven by large-scaled
3395: photometric surveys that are detecting many thousands of new binaries
3396: and tens of extrasolar planets. In particular, we review several
3397: studies that have been used already thousands of lightcurves of
3398: eclipsing binaries to study tidal circularization of early-type stars
3399: in the LMC.
3400:
3401: Whenever possible, the paper attempts to address the possible
3402: confrontation between theory and observations, and to point out
3403: noteworthy cases and observations that can be performed in the future
3404: and may shed some light on the key questions that remain open.
3405:
3406: