1: \documentclass[12pt,preprint]{aastex}
2:
3: \shortauthors{Tsukamoto and Makino}
4:
5: \title{Formation of protoplanets from massive planetesimals in binary systems}
6: \author{YUSUKE TSUKAMOTO}
7:
8: \affil{Department of Earth and Planetary science,
9: University of Tokyo,7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan; tukamoto@margaux.astron.s.u-tokyo.ac.jp}
10:
11:
12:
13: \and
14: \author{ JUNICHIRO MAKINO}
15:
16: \affil{Division of Theoretical Astronomy, National
17: Astronomical Observatory, 2-2-1 Osawa, Mitaka, Tokyo 181-8588, Japan}
18:
19: \begin{abstract}
20:
21: More than half of stars reside in binary or multiple star systems and
22: many planets have been found in binary systems. From theoretical point of
23: view, however, whether or not the planetary formation proceeds in a
24: binary system is a very complex problem, because secular perturbation
25: from the companion star can easily stir up the eccentricity of the
26: planetesimals and cause high-velocity, destructive collisions between
27: planetesimals. Early stage of planetary formation process in binary
28: systems has been studied by restricted three-body approach with gas
29: drag and it is commonly accepted that accretion of planetesimals can
30: proceed due to orbital phasing by gas drag. However, the gas drag
31: becomes less effective as the planetesimals become massive. Therefore
32: it is still uncertain whether the collision velocity remains small and
33: planetary accretion can proceed, once the planetesimals become
34: massive. We performed {\it N}-body simulations of planetary formation
35: in binary systems starting from massive planetesimals whose size is
36: about 100-500 km. We found that the eccentricity vectors of
37: planetesimals quickly converge to the forced eccentricity due to the
38: coupling of the perturbation of the companion and the mutual
39: interaction of planetesimals if the initial disk model is sufficiently
40: wide in radial distribution. This convergence decreases the collision
41: velocity and as a result accretion can proceed much in the same way as
42: in isolated systems. The basic processes of the planetary formation,
43: such as runaway growth and oligarchic growth and final configuration
44: of the protoplanets are essentially the same in binary systems and
45: single star systems, at least in the late stage where the effect
46: of gas drag is small.
47:
48: \end{abstract}
49:
50: \keywords{binaries: close --- planetary systems: formation ---
51: methods: {\it n}-body simulations}
52:
53: \begin{document}
54: \maketitle
55:
56: \section{Introduction}
57: As of March 2007, 215 candidates of extra-solar planets have been
58: found and at least 30 of them are in binary or multiple star systems
59: (Raghavan {\it et al.} 2006). Table \ref{candidates} shows the
60: candidates of the close binary or multiple star system which has the
61: planets. In this table, $\gamma$ Cephei is an example of
62: close binary systems in which we are interested. According to Hatzes
63: {\it et al.} (2003), the semi-major axis and eccentricity of the
64: companion star of $\gamma$ Cephei are $18.5$ AU, and 0.36,
65: respectively. Planetary formation process in such a close binary
66: system is the main target of this paper. The frequency of planets in
67: binary systems is not significantly different from that for single star
68: systems (Desidera and Barbieri 2007). It is very important to
69: investigate the formation process in binary system because more than
70: half of stars reside in binary or multiple star systems.
71:
72: Many authors have investigated planetary accretion process using {\it
73: N}-body simulation (e.g., Aarseth, Lin, and Palmer 1993; Kokubo and
74: Ida 1996, 1998, 2000, 2002). In all of these simulations, the
75: evolution of protoplanetary disks around isolated stars was studied.
76: On the other hand, Quintana {\it et al.} (2002, 2006) have
77: investigated the planet formation in binary system from protoplanets
78: by {\it N}-body simulations. They studied the late stage of the
79: planet formation process (from protoplanets to planets).
80:
81: Marzari and Scholl (2000) and Th\'ebault {\it et al.} (2004, 2006)
82: investigated the distribution of the collision velocity between
83: planetesimals with and without gas drag in binary systems by
84: restricted three-body approach. They found that orbital phasing
85: occurred and collision velocity remained small due to the coupling of
86: gas drag effect and secular perturbation, and they concluded that
87: planetary accretion could proceed. As the planetesimals become
88: massive, however, the gas drag becomes ineffective and mutual
89: gravitational effect between planetesimals becomes important. In such
90: a condition, whether the collision velocity remains small is not
91: clear. As stated above, the final stage, from protoplanets to planets,
92: has been studied by $N$-body simulations, but there has been no $N$-body
93: work on the intermediate stage from massive planetesimals whose size
94: is 100 km - 500 km to protoplanets. In this paper, we focus on this
95: intermediate stage.
96:
97: In a binary system, the orbits of planetesimals change due to
98: perturbation from the companion. The most important term of the
99: perturbation in our simulation region is secular perturbation, if the
100: orbit of the companion is eccentric. The effect of secular perturbation is
101: that eccentricity vector moves on the "perturbation circle". Thus, if
102: interaction between planetesimals and gas drag are neglected,
103: planetesimals with different semi-major axis move on the perturbation
104: circles on different frequencies and phases, and therefore gain high
105: relative velocity. This is the reason why the collision velocity becomes high
106: in previous works without gas drag or self gravity (Marzari and Scholl 2000;
107: Th\'ebault {\it et al.} 2004, 2006). If the collision velocity becomes
108: high when the planetesimals become massive and thus, the effect of gas drag
109: becomes small, accretion process might halt since collisions might be
110: destructive.
111:
112: It is a very important question whether destructive collisions occur when
113: the gravitational interaction between planetesimals is taken into account
114: because it determines whether the accretion process can continue after
115: gas drag becomes ineffective. Ito and Tanikawa (2001) studied the
116: evolution of the orbital elements of protoplanets under the
117: perturbation of Jupiter and found that the eccentricities of the
118: protoplanets aligned with each other, resulting in the evolution was
119: very similar to that without the influence of Jupiter. This alignment
120: is due to gravitational interaction between the protoplanets. If this
121: kind of alignment also occur for planetesimals in binary system,
122: the collision velocity can become smaller than the value predicted by
123: Marzari and Scholl (2000).
124:
125: In this paper, we report the results of {\it N}-body simulations of
126: protoplanet formation from massive ($3 \times 10^{23} - 1.4 \times
127: 10^{24}$ g) planetesimals. We start simulations with the
128: planetesimal disk in which all planetesimals have the same mass. The
129: gravitational interaction between planetesimals is included and
130: gas drag is neglected. We summarize theory of secular perturbation
131: and discuss what initial distribution should be used in section 2.
132: The numerical scheme and initial conditions are described in section
133: 3. We also discuss the validity of the "same mass" assumption of
134: planetesimals in section 3. The results of {\it N}-body simulations
135: are presented in section 4. In section 5, we sum up.
136:
137: \begin{table}[hbt]
138: \
139: \caption{Known candidates of close systems}
140: \label{candidates}
141: \begin{center}
142: \begin{tabular}{cccccccc}
143: \hline\hline
144: Name
145: & Component
146: & \begin{tabular}{c} $ a \sin i $ \\(AU) \end{tabular}
147: & \begin{tabular}{c} Projected \\ Separation \\ (AU) \end{tabular}
148: & \begin{tabular}{c} $M \sin i$ \\ ($M_{J}$)\end{tabular}
149: & eccentricity
150: & References \\
151: \hline
152:
153: HD 041004 & B & ... & 22 & ... & ... & 1, 2, 3, 4 \\
154: & C & 0.016 & ... & 18.4 & 0.08 & 3, 4 \\
155: & b & 1.31 & ... & 2.3 & 0.39 & \\
156: \hline
157:
158: $\gamma$ Cephei & B & 20.3 & ... & ... & 0.39 & 1, 5, 6, 7, 8 \\
159: & b & 2.03 & ... & 1.59 & 0.2 & \\
160: \hline
161:
162: GJ 893 & B & ... & 2248 & ... & ... & 9, 10 \\
163: & C & ... & 18 & ... & ... & \\
164: & b & 0.3 & ... & 2.9 & ... & \\
165: \hline\hline
166:
167: \end{tabular}
168: \end{center}
169:
170: \tablecomments{These parameters are from Raghavan {\it et al.} (2006) and
171: its references. Column component lists companions (B, C,
172: D,...) and planets (b, c, d, ...).}
173: \tablerefs{(1)Eggenberger {\it et al.} 2004;
174: (2)See 1896; (3) Zucker {\it et al.} 2003; (4) Zucker {\it et al.}
175: 2004; (5) Mason {\it et al.} 2001; (6) Campbell {\it et al.}
176: 1988; (7) Griffin {\it et al.} 2002; (8) Hatzes {\it et al.}
177: 2003; (9) Zacharias {\it et al.} 2004; (10) Wilson 1953}
178: \end{table}
179:
180:
181: \section{Theoretical Preparation }
182: \subsection{Brief description of secular perturbation}
183: If the mutual interaction of planetesimals and gas drag are
184: negligible, the orbital evolution of a planetesimal in a binary system
185: can be described by restricted three-body approximation, and the time
186: evolution of eccentricity and longitude of pericenter of the planetesimal
187: are given by secular perturbation theory. The secular evolution of
188: the {\it h} and {\it k} variables of the planetesimal are expressed as
189: (Heppenheimer 1978; Whitmire {\it et al.} 1998; Th\'ebault {\it et
190: al.} 2006)
191: \begin{eqnarray}
192: h(t)& = &e_{p} \sin (At+\varpi_{0})~,\\
193: k(t)&= &e_{p} \cos (At+\varpi_{0}) + e_{f} ~,
194: \end{eqnarray}
195: where
196: \begin{eqnarray}
197: h(t)& = &e \sin ( \varpi)~,\\
198: k(t)&= &e \cos ( \varpi) ~,
199: \end{eqnarray}
200: and $e$ and $ \varpi$ are the eccentricity and the longitude of
201: pericenter of the planetesimal. Here, {\it A}, $e_{p}$, $ \varpi_{0}$, and
202: $e_{f}$ are constants which depend on the strength of the secular
203: perturbation. We take the {\it k} axis as direction of the
204: eccentricity vector of the companion. In other words, the eccentricity
205: vector of the companion is $( e_{B}, 0 )$. The semi-major axis of the
206: companion is $a_{B}$ and its mass is $M_{B}$. According to the
207: secular perturbation theory, $e_{f}$ is given by
208: \begin{eqnarray}
209: e_{f} = \frac{5}{4} \frac{a}{a_{B}} \frac{e_{B}}{(1-e^{2}_{B})} ~.
210: \end{eqnarray}
211: This $e_{f}$ is forced eccentricity induced by the companion
212: star. Here,
213: {\it a} is the semi-major axis of the planetesimal. The angular
214: velocity of rotation on k-h plane, {\it A}, is given by
215: \begin{eqnarray}
216: A = \frac{3}{2} \pi \frac{1}{(1-e^{2}_{B})^{3/2}}m_{B} \frac{a^{3/2}}{a_{B}^{3}} ~.
217: \end{eqnarray}
218: We use system of units in which the solar mass, 1 AU and 1 year
219: are all unity. Remaining two constants, $\varpi_{0}$ and $e_{p}$, are
220: determined from the initial orbital elements of the planetesimal. We
221: call the vectors $ {\bf e } = (k,h),~ {\bf e}_{f} = (e_{f} , 0) $ the
222: eccentricity vector and the forced eccentricity vector,
223: respectively. On k-h plane, the eccentricity vector of a planetesimal
224: moves on a circle centered at ${\bf e}_{f}$.
225:
226: \subsection{Initial eccentricity of planetesimals}
227: We consider two types of initial conditions for planetesimals. In the
228: first one, the initial distribution of eccentricity vectors of
229: planetesimals is centered at the forced eccentricity vector. The
230: distributions of ${\bf e} -{\bf e}_{f}$ and inclination {\it i} are
231: both given by Rayleigh distribution with dispersions $\langle |{\bf
232: e}- {\bf e}_{f}|^{2} \rangle ^{1/2}=2 \langle i^{2} \rangle
233: ^{1/2}=0.02$. We call this model "forced" model. In the second model,
234: the distribution of eccentricity vector is centered at the origin of
235: k-h plane. The distributions of eccentricity {\it e} and inclination
236: {\it i} are also given by Rayleigh distribution with dispersions
237: $\langle e^{2} \rangle ^{1/2}=2 \langle i^{2} \rangle
238: ^{1/2}=0.02$. This is the same distribution as those used in previous works
239: for {\it N}-body simulation of planetary formation in isolated systems
240: (Kokubo and Ida 2002). We call this model "circular" model.
241:
242: We argue that the "forced" model is more suitable for simulation
243: in binary system than the circular model for the following
244: reasons. Consider a narrow region of a planetesimal disk such as the
245: region of 0.95 AU $< a <$ 1.05 AU. If the eccentricity vectors of
246: planetesimals in this region do not distribute around the forced
247: eccentricity vector, they will rotate around the position of the
248: forced eccentricity vector. The planetesimals in a narrow region
249: would rotate together because the angular velocity {\it A} is similar
250: for planetesimals with similar values of the semi-major axis {\it
251: a}. In addition, they align due to secular interactions between
252: planetesimals (see, e.g., Ito and Tanikawa 2001), Because of this
253: collective motion, the relative velocity between planetesimals is kept
254: small in this narrow "ring".
255:
256: However, if we consider a wider region, it would behave as collection
257: of many narrow rings. If we ignore interactions between the rings, they
258: would rotate on their own angular velocities {\it A}. Since
259: neighboring rings have slightly different values of {\it A}, they
260: would soon physically collide with each other, resulting in
261: damping of "free" eccentricity. Gravitational interaction between
262: rings would also dump relative difference of eccentricity vectors.
263: In other words, planetesimals in the
264: circular model would first relax to the forced model. We argue that
265: the equilibrium state is more suitable for the initial conditions.
266: This is why we consider "forced" model.
267:
268: As we mentioned in introduction, we study the late stage of the
269: formation process of protoplanets. The earlier phase have been studied
270: by Marzari and Scholl (2000) and others. Thus, it might seem reasonable
271: to set the initial
272: eccentricity and longitude of pericenter of planetesimals to the
273: equilibrium value of the orbital phasing in Marzari and Scholl
274: (2000) instead of the circular model.
275: However, We found it is a bit difficult to use
276: their results, because they adopt circular gas disk model. This circular gas
277: disk in a binary system with eccentric companion seems a bit
278: unnatural. Due to the perturbation
279: from the companion star, the gas disk may be twisted. We could not
280: find previous works which directly studied the equilibrium state of
281: gas disk under the secular perturbation of companion, but recent work
282: by Papaloizou (2005) seems to imply that eccentric disk can be
283: long-lived even if there is no companion. So it seems likely that gas
284: disk is not circular when eccentric companion exists. This may change
285: gas drag effect to the planetesimals and might affects the
286: direction of the orbital phasing. This is the reason why we do not
287: use the Marzari's results. Even if the orbital elements of them is
288: correct, our circular model is very close to their results and it would
289: soon relax to the "forced" model. So we expect that it makes no significant
290: change to our results. Clearly, more studies on the dynamics of gas
291: disk is necessary.
292:
293:
294: \subsection{Collision velocity}
295: As noted in section 2.1, we adopt the Rayleigh distribution with
296: dispersions $\langle e^{2} \rangle ^{1/2}=2 \langle i^{2} \rangle
297: ^{1/2}=0.02$ for the "circular" model (Ida and Makino 1992), and $\langle
298: |{\bf e}- {\bf e}_{f}|^{2} \rangle ^{1/2}=2 \langle i^{2} \rangle
299: ^{1/2}=0.02$ for the "forced" model (see Table 2 ). These conditions imply
300: collision velocity is initially $v_{col} \simeq 500-1000~
301: \mathrm{m~s^{-1}}$ at 1 AU for the initial distribution. When the
302: effect of gas drag is taken into account, the equilibrium eccentricity
303: and inclination are estimated as $ \langle e^{2} \rangle ^{1/2}=2
304: \langle i^{2} \rangle ^{1/2}=0.0042$ (Kokubo and Ida 2000). This
305: value is probably more reasonable.
306: In this paper, however, we use the value 0.02
307: because we neglect gas drag force in our calculation and the value of
308: the eccentricity and inclination will soon relax to the value of gas-free
309: case. Our focus is not on the quantitative analysis of the collision
310: velocity but on the qualitative analysis of the coupling of
311: secular perturbation and gravitational interactions between
312: planetesimals. To determine the realistic value of the collision
313: velocity, we should include gas drag force in our calculation.
314:
315: \section{ Method of Calculation}
316: \subsection{Initial conditions}
317: The model parameters for the companion star are summarized in table
318: \ref{p of c}. Mass of the primary star is $ 1~ M_{ \odot } $ for all
319: models. We adopt three models for companion. Model alpha corresponds to
320: $\alpha$ Cen system. The mass ratio is chosen to be the same as that
321: of $\alpha$ Cen system, (1.1:0.9). For the other two models, we use a
322: somewhat smaller semi-major axis than those of observed binary systems in
323: which the planets are found. We choose this value to study the case
324: in which secular perturbation is strong. For circular models, we use
325: an axisymmetric surface mass density distribution for planetesimal
326: disk whose surface mass density is given by
327: \begin{eqnarray}
328: \Sigma_{solid} = \Sigma_{0}(\frac{a}{ \mathrm {1~AU} } )^{-3/2}
329: ~\mathrm{ g~ cm^{-2} } ~,
330: \end{eqnarray}
331: where {\it a} is distance from the primary star, and $ \Sigma_{0}$
332: is the reference surface density at 1 AU (We adopt $ \Sigma_{0} = 10
333: ~\mathrm{g ~cm^{-2}}$). Table \ref { inicond } shows model
334: parameters. Here, {\it N} is the number of planetesimals. For
335: radial distribution, we use two disk models, "wide disk" models (models
336: 0 - 5) and "narrow disk" models (models 6 - 8). In the "wide disk"
337: models, we set inner and outer cutoff radii to be 0.5 AU and 1.5 AU,
338: respectively. In the "narrow disk" models, inner and outer cutoff radii
339: are 0.95 AU and 1.05 AU, respectively. Planetesimals have equal mass
340: in all models. The number of planetesimals is 10000 in the wide disk
341: models. In two narrow disk models (model 6, 7), Number of
342: planetesimals is 5000. In one of narrow disk models (model 8), the
343: number of planetesimals is 975. This is a "cutoff" model which has the
344: planetesimals of the same mass as in the wide disk models. The density of
345: planetesimals is 2 $\mathrm{g~cm^{-3}}$. We increase their radii by a
346: factor 5 to accelerate accretion process (Kokubo and Ida 1996). At
347: first, all planetesimals have same mass ($ \simeq 1.44 \times 10^{24}$
348: g in models 0 - 5 and 8 and $2.88 \times 10^{23}$ g in model 6 and 7).
349:
350: One critical question is if our "same mass" setup can be really
351: regarded as description of the later stage of planetary
352: formation. Recent theoretical and numerical works of planetary
353: accretion (for example Inaba {\it et al.} 2001, Rafikov 2003) seem to
354: suggest that the "orderly" phase of growth does not exist, and runaway
355: growth starts from much smaller mass than that of our setup. Our
356: setup, therefore, is not realistic. However, statistical calculation by
357: Inaba {\it et al.} (2001) has shown that once massive
358: planetesimals ($ M \simeq 10^{23}$ g) formed, planetesimals with mass
359: less than $10^{21}$ g become dynamically unimportant. So even though
360: our initial condition is oversimplified, it might still give
361: qualitatively valid description of the later phase.
362:
363: Two planetesimals are considered to collide when their distance
364: becomes less than the sum of their radii. We assume the perfect
365: accretion under which planetesimals always accrete when they
366: collide. Whether or not this assumption is good depends on the
367: distribution of the collision velocities. The escape velocity of our
368: planetesimals is $200-600 ~\mathrm{m~s^{-1}}$ and is of the same order
369: as the initial collision velocity. Furthermore, if gas effect is
370: included, the eccentricity
371: dispersion might be less than our estimate and the collision velocity
372: becomes smaller. So we believe the assumption of perfect accretion is
373: valid.
374:
375: As stated above, we increase the radii of planetesimals {\it f}-fold
376: (here, $f=5$) to save calculation time. This acceleration of
377: accretion may affect time evolution of planetesimals, especially
378: the time evolution of orbital elements on k-h plane because the
379: increase of radii changes timescale of accretion but does not
380: affect secular perturbation of companion star.
381:
382:
383: \begin{table}[hbt]
384: \
385: \caption{Model parameters for the companion star}
386: \label{p of c}
387: \begin{center}
388: \begin{tabular}{cccc}
389: \hline\hline
390: Name & Mass & \begin{tabular}{c}semi-major axis \\(AU) \end{tabular} &
391: eccentricity \\
392: \hline
393:
394: nocomp & 0 & .... & .... \\
395: e25 & $ 1 M_{ \odot }$ & 16 & 0.25 \\
396: e50 & $ 1 M_{ \odot }$ & 16 & 0.5 \\
397: alpha & $ 0.82 M_{ \odot } $ & 23.4 & 0.52 \\
398: \hline
399:
400: \end{tabular}
401: \end{center}
402: \end{table}
403:
404: \begin{table}[hbt]
405:
406: \caption{Initial conditions}
407: \label{ inicond }
408: \begin{center}
409: \begin{tabular}{ccccccc}
410: \hline\hline
411: Model & {\it N} &
412: \begin{tabular}{c} Width of disk \\ (AU) \end{tabular}
413: & companion & initial orbit
414: \\
415: \hline
416:
417: 0 & 10000 & 0.5-1.5 & nocomp & ... \\
418: 1 & 10000 & 0.5-1.5 & e25 & forced \\
419: 2 & 10000 & 0.5-1.5 & e50 & forced \\
420: 3 & 10000 & 0.5-1.5 & e25 & circular \\
421: 4 & 10000 & 0.5-1.5 & e50 & circular \\
422: 5 & 10000 & 0.5-1.5 & alpha & circular \\
423: 6 & 5000 & 0.95-1.05 & e25 & circular \\
424: 7 & 5000 & 0.95-1.05 & alpha & circular \\
425: 8 & 975 & 0.95-1.05 & e25 & circular \\
426:
427: \hline
428:
429:
430: \end{tabular}
431: \end{center}
432: \end{table}
433:
434: \subsection{Integration scheme}
435: We use the fourth-order Hermite scheme (Makino and Aarseth 1992) with
436: hierarchical timesteps (Makino 1991) improved for planetary systems
437: (Kokubo {\it et al.} 1998 ) for numerical integration of planetesimals
438: and companion star. The equation of motion for planetesimals is
439: given by
440: \begin{eqnarray}
441: {\bf a}_{i} = - \sum_{i \neq j} G m_{j} \frac{ {\bf r_{i j}} }{r_{i j}^{3}}
442: - G(m_{i} +M_{p})\frac { {\bf r_{i p}} } {r_{i p}^{3}} -G M_{c}
443: ( \frac { {\bf r_{i c}} }{ r_{i c}^{3} } + \frac { {\bf r_{p c}}
444: }{r_{p c}^{3}} ) ~,
445: \end{eqnarray}
446: where $M_{p},M_{c}, r_{ip}$, and $r_{i c}$ are mass of the primary
447: star, mass of companion star, position of planetesimal relative
448: to the primary star and that relative to the companion star. We use
449: position of the primary star as origin of the coordinate for
450: planetesimals. The motion of the primary star due to the gravitational
451: forces of planetesimals is neglected.
452:
453: Most expensive part of the numerical integration is calculation of
454: mutual gravitational interaction between planetesimals. We use
455: GRAPE-6 (Makino {\it et al.} 2003) and GRAPE-6A (Fukushige {\it et
456: al.} 2005) to calculate the gravitational interaction between
457: planetesimals. We also integrate orbit of the companion star
458: using the fourth-order Hermite scheme. For both the planetesimals and
459: the companion stars, we use the standard timestep criterion (Aarseth 1985)
460: \begin{eqnarray}
461: \Delta t =\sqrt{ \eta \frac{|a||a^{(2)}|+ |\dot{a}|^{2}}
462: {|\dot{a}||a^{(3)}|+ |a^{(2)}|^{2}}} ~.
463: \end{eqnarray}
464: Since the orbital period of the companion is long, the accuracy of its
465: orbit is more than enough.
466:
467: \section{Results}
468: \subsection{Planetary accretion in binary systems}
469: Figure \ref {snapshots} shows the time evolution of planetesimals of
470: model 1 (left) and model 0 (right) on the a-e plane. We integrate the
471: system for $5 \times 10^{5}$ years. We can see that protoplanets grow
472: in a very similar way in these two models. This behavior is essentially
473: the same for all models.
474:
475: Figure \ref {mass_number} shows cumulative mass distribution of
476: planetesimals of the region $0.9 {\rm AU}< a <1.1 {\rm AU}$. The time
477: evolutions of all of these models are very similar. In all cases, the mass
478: distributions first relax to the power-law distribution with power
479: index $d \log n_{c}/d \log m \simeq -1.5$ (top panels) where $n_{c} $
480: is the cumulative number of planetesimals. This power-law distribution
481: is characteristic of the runaway growth (Makino {\it et al.} 1998,
482: Kokubo and Ida 2000).
483:
484: As time goes on, the mass distributions become bimodal: planetesimals
485: with mass $ M \simeq 1-10 \times 10^{24}$ g and large protoplanets
486: with mass $ M \simeq 10^{27}$ g. This bimodal mass distribution is the
487: result of the runaway and the oligarchic growth of protoplanets
488: (Kokubo and Ida 1996,1998).
489: The time evolution of the mass distribution of binary systems is
490: almost the same as that in single star systems. It means that
491: secular perturbation does not change the process of the protoplanet
492: formation. Runaway growth and oligarchic growth also occur in
493: binary systems in a similar way as in the single star system.
494:
495: In figure \ref {time_max_e25} - \ref {time_ave_e50}, we show the
496: evolution of the mass of most massive planetesimals (\ref
497: {time_max_e25} and \ref {time_max_e50}) and average mass of
498: planetesimals (\ref {time_ave_e25} and \ref {time_ave_e50}). We can
499: see that the evolution is rather similar and there is no systematic
500: tendency due to the presence of the companion.
501:
502: The evolution of the surface mass density of model 1 ($e=0.25$,
503: forced) and that of model 0 (no companion) are shown in Figure
504: \ref{surface}. We found that density profile of planetesimals in a
505: binary system is virtually the same as that in a single star system.
506:
507: From secular perturbation theory, time evolution of
508: semi-major axis of a planetesimal is given by
509: \begin{eqnarray}
510: \frac{da}{dt} = \frac{2}{na} \frac{ \partial R}{ \partial \epsilon}~,
511: \end{eqnarray}
512: where $ \epsilon$ and $R$ are mean longitude at epoch and
513: disturbing function, respectively. In practice, variation of
514: $\epsilon$ can usually be neglected since it is a small effect. Thus,
515: mass migration does not occur since secular perturbation hardly
516: change the semi-major axis of planetesimals. Our results are
517: consistent with this theoretical expectation.
518:
519: \begin{figure}
520: \plotone{f1.eps}
521:
522: \caption{Distribution of planetesimals on the {\it a-e} plane for model 1
523: (left)
524: and model 0 (right). The circles represent planetesimals. The filled circle represent
525: protoplanets with mass more than 100 times the initial mass.}
526: \label{snapshots}
527: \end{figure}
528:
529:
530: \begin{figure}
531: \plotone{f2.eps}
532:
533: \caption{ Cumulative mass distribution of planetesimals. Left panels
534: show the results of the forced models (model 1, 2) and right panels
535: show the results of the circular models (model 3, 4). In all panels,
536: the results of isolated model (model 0) are also shown. Times are $
537: 5\times 10^{4},~1\times 10^{5},~ 3\times 10^{5},~5 \times 10^{5}$
538: years from top to bottom. The dotted lines at the top panels indicate
539: the slope of -1.5.}
540: \label{mass_number}
541: \end{figure}
542:
543:
544: \begin{figure}
545: \plotone{f3.eps}
546:
547: \caption{The time evolution of maximum mass of planetesimals
548: between $0.9{\rm AU}<a<1.1{\rm AU}$. Solid, dashed, and dotted
549: curves show the result of the model 0, 1, and 3, respectively. }
550: \label{time_max_e25}
551: \end{figure}
552:
553: \begin{figure}
554: \plotone{f4.eps}
555:
556: \caption{Same as figure \ref {time_max_e25} but for models 0, 2, and 4. }
557: \label{time_max_e50}
558: \end{figure}
559:
560: \begin{figure}
561: \plotone{f5.eps}
562:
563: \caption{Same as figure \ref {time_max_e25} but the average mass is shown. }
564: \label{time_ave_e25}
565: \end{figure}
566:
567: \begin{figure}
568: \plotone{f6.eps}
569:
570: \caption{ Same as figure \ref {time_ave_e25} but for models 0, 2, 4. }
571: \label{time_ave_e50}
572: \end{figure}
573:
574: \begin{figure}
575: \epsscale{0.8}
576: \plotone{f7.eps}
577: \caption{The surface mass density of planetesimals for model 1
578: (solid) and model 0 (dotted) at $t=1 \times 10^{2},1
579: \times 10^{5},2 \times 10^{5},4 \times 10^{5}$ years, from top to bottom. }
580: \label{surface}
581: \end{figure}
582:
583: \subsection{ Time evolution of eccentricity}
584: The time evolutions of the distribution of the planetesimals on k-h
585: plane of model 1 (forced model of $e=0.25$), 3 (circular model of
586: $e=0.25$), 8 (cutoff model of $e=0.25$), 4, and 5 (circular models)
587: are shown in Figure \ref {snapshot_hk}.
588:
589: The top three panels show models 1, 3, 8, all with an e25 companion. In
590: model 1 (top panel), planetesimals are initially distributed around
591: the forced eccentricity, and this does not change in time. In model 3,
592: planetesimals are initially distributed around zero eccentricity, but
593: this initial distribution is replaced by a distribution centered at
594: the forced value. The behavior of planetesimals in model 8 is very
595: different. The distribution after 30,000 years is not centered at the
596: forced value and keeps rotating around the forced eccentricity vector
597: (see figure \ref {time_e_e25_alpha_multi} and description below).
598: This difference between narrow disk models and wide disk models will be
599: discussed in more details in section 4.2.1. The bottom two panels of
600: figure \ref {snapshot_hk} show that the eccentricity vectors of the
601: planetesimals of other circular models with wide distribution also
602: converge to the forced eccentricity vector. We will discuss the time
603: evolution of proper eccentricity in section 4.2.2.
604:
605: \begin{figure}
606: \epsscale{0.6}
607: \plotone{f8.eps}
608: \caption{The distribution of planetesimals of model 1, 3, 8, 4, 5 (top
609: to bottom) on k-h plane. Left panels show initial distributions and
610: right panels show evolved state ($3 \times 10^{4}$ years, for models 1,
611: 3, and 8 and $2 \times 10^{4}$ and $8 \times 10^{4}$ years for models 4
612: and 5)
613: We plotted planetesimals between $0.95 ~\mathrm{AU} < a < 1.05~\mathrm{AU}$. The center of
614: circles is the forced eccentricity for $a= 1$ AU and radii of circles are 0.01,
615: 0.03, 0.05 respectively.
616: The open and filled circles represent the planetesimals with mass 10
617: times and 100 times the initial value.}
618: \label{snapshot_hk}
619: \end{figure}
620:
621: \subsubsection{Difference between wide and narrow models}
622: We performed simulations of the narrow disk models (models 6, 7, 8) to
623: see the difference between the wide and narrow distributions. We use two
624: types of narrow disk models. One has the planetesimals of the same
625: mass as in wide disk models, and thus the number of planetesimals is small
626: ($N=975$, model 8). The other has a large number of smaller
627: planetesimals (see Table 2).
628:
629: Figure \ref {time_e_e25_alpha_multi} shows the time evolution of
630: eccentricity. In wide disk models, the oscillation of eccentricity
631: is damped quickly due to the convergence to the forced eccentricity
632: vector as shown in figure \ref {snapshot_hk}. In narrow disk models,
633: however, the eccentricity liberates around the forced eccentricity and
634: does not converge to the forced eccentricity.
635:
636: This difference can be understood as follows. As we discussed in
637: section 2, the planetesimals rotate around the forced eccentricity
638: vector on k-h plane and the angular velocity {\it A} is a function of
639: semi-major axis of planetesimals. In the case of narrow disk models,
640: the range of {\it A} is small, and it is possible that all
641: planetesimals synchronize due to mutual gravitational
642: interaction. Thus in narrow disk models planetesimals move
643: collectively on k-h plane. In the case of the wide disk models,
644: however, such collective motion is not allowed since the range of {\it
645: A} is too large. If planetesimals with different values of {\it a}
646: (therefore, {\it A}) rotate around the its own values of forced
647: eccentricity on their own timescales, collision is enhanced and
648: eccentricity will be damped to forced values. Thus,
649: oscillation of eccentricity damps quickly in wide models.
650:
651: We performed a simulation of narrow disk models to see how this behavior is
652: affected by the mass of planetesimals. The bottom panel of Figure \ref
653: {time_e_e25_alpha_multi} shows the time evolution of eccentricity of
654: model 8. The behavior of model 8 is almost the same as that of model
655: 5. Thus, the number of planetesimals or mass of each planetesimal do
656: not change the result. In narrow disk models, the convergence to
657: the forced eccentricity vector never occur.
658:
659: \begin{figure}
660: \epsscale{1.0}
661: \plotone{f9.eps}
662:
663: \caption{The time evolution of the average eccentricity in narrow disk
664: models (models 6 - 8) compared with those in wide disk models (models
665: 3, 5, and 3 from top to bottom). For wide disk models, the
666: eccentricity is averaged for planetesimals in the range of $0.95
667: ~\mathrm{AU} < a < 1.05 ~\mathrm{AU}$. The solid curves show results
668: of wide disk models and dashed curves show results of narrow disk models.}
669:
670: \label{time_e_e25_alpha_multi}
671:
672: \end{figure}
673:
674:
675: \subsubsection{Time evolution of proper eccentricity}
676: Time evolution of proper eccentricity, ({\it i.e.,} distance from the
677: forced eccentricity in k-h plane) is shown in figure \ref
678: {e_eforced_global}. The proper eccentricity is averaged for
679: planetesimal with semi-major axis between $0.95~\mathrm{AU}<a<
680: 1.05~\mathrm{AU} $. In circular models (model 3, 4, and 5), the
681: proper eccentricity quickly decreases through collisional damping and
682: gravitational relaxation. This means that the eccentricity vector of
683: the planetesimals converge to the forced eccentricity vector. In
684: forced models (model 1, 2) and single star model (model 0), on the
685: other hand, the proper eccentricity does not change significantly in early
686: stage of the simulation (before 40,000 years).
687: The convergence in model 5 ($\alpha$ Cen model) is slower than that
688: in the other two circular models and before convergence the planetary
689: accretion has almost finished. This is because of the small angular
690: velocity {\it A} of $\alpha$ Cen system. However, if we adopt real
691: radii, the convergence would probably occur before the accretion
692: completes. The time evolution of proper eccentricity after the
693: convergence is similar to that in the forced models and the single
694: star model. The proper eccentricity increases in the late stage due to
695: the viscous stirring.
696:
697: Figure \ref {e_eforced_mass} shows time evolution of the mass
698: weighted average of proper eccentricity for the same radial range as
699: in figure \ref {e_eforced_global}. In the early stage of the
700: simulation, its behavior is similar to that of the simple average in
701: figure \ref {e_eforced_global}. However, in the late stage, the
702: increase is slow because of dynamical friction to seeds of protoplanet.
703: In single star systems, the eccentricity of seeds of protoplanet
704: becomes small ({\it i.e.}, the eccentricity vectors converge to the origin of
705: k-h plane). On the other hand, in binary systems, the
706: eccentricity vectors converge to the forced eccentricity and the
707: evolution of proper eccentricity is similar to the evolution of
708: eccentricity in the single star system. This means that the orbit of
709: planetesimals does not become circular as in single star system
710: but become eccentric in binary systems and their pericenter aligns to
711: the pericenter of companion star.
712:
713: The distance between planetesimals on k-h plane determines the
714: collision velocity. Thus, the convergence reduces the collision
715: velocity between planetesimals. In model alpha, for example, the
716: maximum distance between planetesimals could becomes 0.08 by secular
717: oscillation if the mutual gravitational effect is neglected. It
718: corresponds to the collision velocity of 3000-4000 $\mathrm{m~s^{-1}}$
719: at 1 AU. Collision with this velocity is destructive even for the
720: mass of the planetesimal of
721: about $1.4 \times 10^{24}$ g (its escape velocity is about 600
722: $\mathrm{m~s^{-1}}$). On the other hand, The collision velocity is
723: reduced to about 500-1000 $\mathrm{m~s^{-1}}$ with this convergence.
724: As we mentioned above, gas drag
725: affects the eccentricity dispersion of the planetesimals ({\it i.e.,}
726: gas drag can not be negligible in this sense). The simulations in
727: which gas drag and mutual gravitational effects are taken
728: into account is required to determine the precise value of the
729: collision velocity.
730:
731: \begin{figure}
732:
733: \plotone{f10.eps}
734:
735:
736: \caption{
737: The time evolution of average of proper eccentricity. the proper
738: eccentricity is averaged for planetesimals with $0.95 ~\mathrm{AU} < a
739: < 1.05 ~\mathrm{AU}$.
740: Solid, long-dashed, short-dashed, dotted, and upper dot-dashed curves show the results of
741: models 1 - 5, respectively. Lower dot-dashed curve shows the result of
742: model 0.
743: }
744: \label{e_eforced_global}
745:
746: \end{figure}
747:
748: \begin{figure}
749:
750: \plotone{f11.eps}
751:
752:
753: \caption{
754: The same as figure \ref {e_eforced_global} but mass-weighted averages
755: are shown.
756: }
757:
758: \label{e_eforced_mass}
759:
760: \end{figure}
761:
762: \section{Conclusion and Discussion}
763: We have performed the simulations of formation of protoplanets from
764: massive planetesimals (with radii of 100km-500km) in close binary
765: systems, for which gas drag effect is negligible and the
766: coupling of the secular perturbation and gravitational interaction
767: is important.
768:
769: We found that eccentricity vectors of planetesimals quickly
770: converge to the forced eccentricity if the initial disk model is
771: sufficiently wide in radius and secular perturbation is
772: sufficiently strong. The eccentricity vectors of planetesimals which
773: have heavier mass than the other planetesimals converge
774: to the forced eccentricity more strongly. This convergence results in
775: orbital phasing and it reduces value of collision velocity to
776: less than the value predicted in studies with restricted three-body
777: approach.
778: Runaway growth and oligarchic growth also occur in binary system much
779: in the same way as isolated star system at least in the late stage
780: of the formation.
781: The final configuration of protoplanets is not different between
782: close binary systems and isolated star systems.
783:
784: Our simulations, however, have following limitations. (1) We
785: underestimated effect of secular perturbation. The {\it f}-fold
786: change of radius increases frequency of collision and accelerates
787: planetary accretion. This acceleration of the formation process causes
788: relative under-estimate of secular perturbation. We plan to
789: perform the simulation with real radii of planetesimals to see if this
790: effect would make any difference. (2) We investigated only three binary
791: systems. More simulations of various binary systems are
792: required to determine quantitative relationship between the
793: convergence of eccentricity vectors and strength of secular
794: perturbation. (3) We neglected effect of gas drag. This is okay for
795: the late stage which we studied. In earlier stage, however, gas drag
796: is clearly important. We plan to calculate the equilibrium state of
797: gas disk under the secular perturbations and perform the
798: simulations which include the gas effect in future works.
799:
800:
801: \section *{Acknowledgments}
802: We express our sincere gratitude to Masaki Iwasawa and Keigo Nitadori
803: for useful comments on our {\it N}-body code. We also thank Eiichiro
804: Kokubo and Satoshi Inaba for many useful suggestions. We thank the
805: anonymous referee whose critical comments helped us greatly to make
806: the paper clear. This research is partially supported by the Special
807: Coordination Fund for Promoting Science and Technology (GRAPE-DR
808: project), Ministry of Education, Culture, Sports, Science and
809: Technology, Japan.
810:
811: \begin{thebibliography}{}
812:
813: \bibitem[]{}
814: Aarseth, S.~j. 1985, in {\it Multiple Time Scales,}
815: ed. J. U. Brackhill and B. I. Cohen (Academic Press, New York),
816: p. 377.
817: \bibitem[Aarseth et al.(1993)]{ALP93}
818: Aarseth, S.~J., Lin, D.~N.~C., Palmer, P.~L. 1993, ApJ, 403, 351
819:
820:
821: \bibitem[Campbell et al.(1988)]{Cam1988} Campbell, B., Walker,
822: G.~A.~H., \& Yang, S.\ 1988, \apj, 331, 902
823:
824:
825: \bibitem[]{}
826: Desidera, S., Barbieri, M., 2007, A\&A, 462, 345
827:
828:
829: \bibitem[Eggenberger et al.(2004)]{Egg2004} Eggenberger, A., Udry, S.,
830: \& Mayor, M.\ 2004, \aap, 417, 353
831:
832: \bibitem[Fukushige et al.(2005)]{fmk-05}
833: Fukushige, T., Makino, J., \& Kawai, A.\ 2005, PASJ, 57, 1009
834:
835:
836: \bibitem[Griffin et al.(2002)]{Gri2002} Griffin, R.~F.,
837: Carquillat, J.-M., \& Ginestet, N.\ 2002, The Observatory, 122, 90
838:
839:
840: \bibitem[Heppenheimer(1978)]{hep78} Heppenheimer, T., 1978,
841: A\&A, 65, 421
842: %
843: \bibitem[]{} Hatzes, A.P., Cochran, W.D., Endl, M., {\it et al.} 2003,
844: ApJ. 599, 1382
845:
846:
847: \bibitem[Ito \& Tanikawa(2001)]{ito01}
848: Ito, T., Tanikawa, K. 2001, PASJ, 53, 143
849:
850: \bibitem[]{} Inaba, S., Tanaka, H., Nakazawa, K., Wetherill, G. W., \& Kokubo, E. 2001, Icarus, 149, 235
851:
852: \bibitem[Kokubo \& Ida(1996)]{KI96}
853: Kokubo, E. \& Ida, S. 1996, Icarus, 123, 180
854:
855: \bibitem[Kokubo \& Ida(1998)]{KI98}
856: ---------. 1998, Icarus, 131, 171
857:
858: \bibitem[Kokubo \& Ida(2000)]{KI00}
859: ---------. 2000, Icarus, 143, 15
860:
861: \bibitem[Kokubo \& Ida(2002)]{KI02}
862: ---------. 2002, ApJ, 581, 666
863:
864:
865: \bibitem[Kokubo et al. 1998]{kokubo}
866: Kokubo, E., Yoshinaga, K., \& Makino, J. 1998, MNRAS,
867: 297, 1067
868:
869: \bibitem[Makino 1991]{maki91} Makino,~J. 1991b, PASJ, 43, 859
870:
871: \bibitem[Makino \& Aarseth(1992)]{Makino:92}
872: Makino, J., \& Aarseth, S.~J.\ 1992, PASJ, 44, 141
873:
874:
875: \bibitem[Makino et al 1998]{makietal98}
876: Makino, J., Fukushige, T., Funato, Y., \& Kokubo, E. 1998, New Astronomy, 3, 411
877:
878: \bibitem[makino et al 2003]{makinoetal2003} Makino,~J., Fukushige,~T., Koga,~M., \& Namura,~K. 2003,
879: PASJ, 55, 1163
880:
881:
882:
883: \bibitem[Marzari \& Scholl(2000)]{mascho00} Marzari F., Scholl H., 2000,
884: ApJ, 543, 328
885:
886:
887: \bibitem[Mason et al.(2001)]{Msn2001} Mason, B.~D., Hartkopf, W.~I.,
888: Holdenried, E.~R., \& Rafferty, T.~J.\ 2001, \aj, 121, 3224
889:
890: \bibitem[]{}Papaloizou, J. C. B. 2005, A\&A, 432, 757
891:
892: \bibitem[Quintana et al. (2002)]{qui02} Quintana, E. V., Lissauer,
893: J. J., Chambers, J. E., \& Duncan, M. J. 2002, ApJ, 576, 982
894:
895: \bibitem[Quintana \& Lissauer (2006)]{qui06} Quintana, E. V., \&
896: Lissauer, J. J. 2006, Icarus, 185, 1
897:
898: \bibitem[]{}Rafikov, R. R. 2003, AJ, 125, 942
899:
900: \bibitem[Raghavan et al.(2006)]{Raghavan:06}
901: Raghavan, D., Henry, T.\ J., Mason, B.\ D., Subasavage, J.\ P., Jao,
902: W.-C., Beaulieu, T.\ D., \& Hambly, N.\ C. 2006, ApJ, 646, 523
903:
904: \bibitem[See(1896)]{See1896} See, T.~J.~J.\ 1896, AN 142, 43
905:
906:
907: \bibitem[Th\'ebault et al.(2004)]{the04} Th\'ebault, P.,
908: Marzari, F., Scholl, H., Turrini, D., Barbieri, M.,2004,
909: A\&A, 427, 1097
910:
911: \bibitem[Th\'ebault et al.(2006)]{the06} Th\'ebault, P.,
912: Marzari, F., Scholl, H., 2006, Icarus, 183, 193
913:
914: \bibitem[Whitmire et al.(1998)]{whit98} Whitmire, D., Matese, J., Criswell, L.,
915: 1998, Icarus, 132, 196
916: %
917:
918: %\bibitem[]{} G. W. Wetherill and S. Inaba, 2000, Space Sci. Rev., 92, 311
919:
920:
921: \bibitem[]{} Wilson, R. E. 1953, General Catalogue
922: of Stellar Radial Velocities (Washington, DC: Carnegie
923: Inst. Washington)
924:
925: \bibitem[Zacharias et al.(2004)]{UCAC2} Zacharias, N., Urban,
926: S.~E., Zacharias, M.~I., Wycoff, G.~L., Hall, D.~M., Monet, D.~G., \&
927: Rafferty, T.~J.\ 2004, \aj, 127, 3043
928:
929: \bibitem[Zucker et al.(2003)]{Zuc2003} Zucker, S., Mazeh, T., Santos,
930: N.~C., Udry, S., \& Mayor, M.\ 2003, \aap, 404, 775
931:
932: \bibitem[Zucker et al.(2004)]{Zuc2004} Zucker, S., Mazeh, T., Santos,
933: N.~C., Udry, S., \& Mayor, M.\ 2004, \aap, 426, 695
934: \end{thebibliography}
935:
936: \end{document}