E The Astrophysical Journal, 581:666–680, 2002 December 10 # 2002. The American Astronomical Society. All rights reserved. Printed in U.S.A. FORMATION OF PROTOPLANET SYSTEMS AND DIVERSITY OF PLANETARY SYSTEMS Eiichiro Kokubo Division of Theoretical Astrophysics, National Astronomical Observatory, Osawa, Mitaka, Tokyo, 181-8588, Japan; [email protected] and Shigeru Ida Department of Earth and Planetary Sciences, Tokyo Institute of Technology, Ookayama, Meguro-ku, Tokyo, 152-8551, Japan Received 2002 May 14; accepted 2002 August 8 ABSTRACT We investigate the formation of protoplanet systems from planetesimal disks by global (N ¼ 5000 and 10,000 and 0:5 AU < a < 1:5 AU, where N is the number of bodies and a is the distance from a central star) N-body simulations of planetary accretion. For application to extrasolar planetary systems, we study the wide variety of planetesimal disks of the surface mass density solid ¼ 1 ða=1 AUÞ g cm2 with 1 ¼ 1, 10, 100 and ¼ 1=2; 3=2; 5=2. The results are all consistent with the prediction from the ‘‘ oligarchic growth ’’ model. We derive how the growth timescale, the isolation (final) mass, and the orbital separation of protoplanets depend on the initial disk mass (1) and the initial disk profile (). The isolation mass increases in pro3=2 1=2 portion to 1 , while the number of protoplanets decreases in proportion to 1 . The isolation mass ð3=2Þð2Þ , which means it increases with a for < 2 while it decreases with a for > 2. The depends on a as a growth timescale increases with a but decreases with 1. Based on the oligarchic growth model and the conventional Jovian planet formation scenario, we discuss the diversity of planetary systems. Jovian planets can form in the disk range where the contraction timescale of planetary atmosphere and the growth timescale of protoplanets (cores) are shorter than the lifetime of the gas disk. We find that for the disk lifetime 108 yr, several Jovian planets would form from massive disks with 1 e30 with Uranian planets outside the Jovian planets. Only terrestrial and Uranian planets would form from light disks with 1 d3. Solar system–like planetary systems would form from medium disks with 1 ’ 10. Subject headings: planetary systems — planetary systems: formation — solar system: formation On-line material: color figures When the mass of a protoplanet exceeds a critical value, runaway growth slows down (Ida & Makino 1993) and then the growth mode shifts to oligarchic growth (Kokubo & Ida 1998), where among protoplanets larger bodies tend to grow more slowly than smaller ones, while the mass ratios between protoplanets and planetesimals still increase. As a result, in the late stage, similar-sized protoplanets grow oligarchically, while most planetesimals remain small. In this stage, orbital separations of protoplanets are kept wider than about 5 Hill radii of the protoplanets through orbital repulsion (Kokubo & Ida 1995). The final stage of terrestrial planet formation would be giant impacts among the protoplanets (e.g., Chambers & Wetherill 1998; Agnor, Canup, & Levison 1999; Kominami & Ida 2002), and that of Jovian planet formation would be, according to the conventional ‘‘ core instability ’’ scenario, gas accretion onto cores (protoplanets) while accreting planetesimals (e.g., Mizuno 1980; Bodenheimer & Pollack 1986; Pollack et al. 1996; Ikoma, Nakazawa, & Emori 2000). Uranian planets would be failed protoplanets that could not capture the massive gas envelope (e.g., Lissauer et al. 1995). Most of the previous studies of planetary accretion focused on solar system formation. In other words, the minimum-mass protoplanetary disk model was assumed in which the disk mass Mdisk ’ 0:01–0.02 M (Weidenschilling 1977; Hayashi 1981) and the gas-to-dust ratio is the solar abundance. This model contains the smallest amount of the solid component (dust) that can reproduce the present solar system. The distribution of planetesimals was then assumed 1. INTRODUCTION In the standard scenario of solar system formation, the building blocks of planets are kilometer-sized bodies called planetesimals (e.g., Safronov 1969; Hayashi, Nakazawa, & Nakagawa 1985). Planetesimals are formed from the solid component (dust) in the solar nebula (a protoplanetary disk). While orbiting the Sun, planetesimals accrete to form terrestrial (rocky) and Uranian (icy) planets and cores of Jovian (gaseous) planets. This process is called planetary accretion. Planetary accretion is an important process of planet formation that controls the basic structure of a planetary system and the formation timescale of planets. In order to clarify solar system formation, many authors investigated planetary accretion using a gasdynamic approach (e.g., Greenberg et al. 1978; Wetherill & Stewart 1989; Weidenschilling et al. 1997) and N-body simulation (e.g., Aarseth, Lin, & Palmer 1993; Kokubo & Ida 1996, 1998, 2000). The current understanding of planetary accretion is summarized as follows. In the early stage of planetary accretion, a larger planetesimal grows faster than smaller ones, resulting in the runaway growth of the largest planetesimal (Greenberg et al. 1978; Wetherill & Stewart 1989; Kokubo & Ida 1996). This is because in a planetesimal system, the larger a planetesimal becomes, the larger its growth rate becomes by gravitational focusing and dynamical friction from smaller planetesimals. Protoplanets (runaway planetesimals) are formed through runaway growth. As protoplanets grow, they start to interact with one another. 666 FORMATION OF PROTOPLANET SYSTEMS to be consistent with the dust distribution. The observation of protoplanetary disks confirmed that the typical mass of protoplanetary disks is on the order of 0.01 M (e.g., Beckwith & Sargent 1996). However, it also revealed that the disk mass ranges from 0.1 to 10 times the minimum-mass disk model. The radial profile of the disk surface density can also differ from that of the minimum-mass disk model. Dust evolution in a gas disk was investigated in various models of protoplanetary disks by Kornet, Stepinski, & Rozyczka (2001). They numerically studied the formation of a planetesimal swarm from dust under the dust-gas interaction and discussed the possible diversity of planetesimal systems. We focus on the subsequent process, planetary accretion from planetesimals, in the present paper. In order to study the formation of extrasolar planetary systems, we need to generalize the planetary accretion theory. It is necessary to study planetary accretion in less and more massive protoplanetary disks than the minimummass disk model. The variety of the mass distribution in protoplanetary disks would potentially produce diverse planetary systems. In particular, the number and location of gas giant planets should depend on disk mass distribution because the core mass depends on disk surface density and gas giants form where the core mass exceeds some critical value according to the core instability scenario. The recently found extrasolar gas giants close to the central star or those in eccentric orbits might be formed in much more massive disks than the minimum-mass disk (e.g., Lin & Ida 1997). In the previous N-body simulations of planetary accretion, Kokubo & Ida (1996, 1998, 2000) adopted a radially narrow ring of planetesimals where the ratio of the ring width Da to the ring radius a is Da=a 0:01–0.1. Kokubo & Ida (1998, 2000) developed the oligarchic growth model based on the result of these radially local N-body simulations. It is important to confirm the results of the local N-body simulations by means of radially global N-body simulations. The wide simulation width is necessary to see the global interaction of protoplanets and the effect of the radial distribution of planetesimals on planetary accretion. In the present paper, we perform radially global N-body simulations of planetary accretion with two major motivations: (1) to confirm the oligarchic growth model by global N-body simulations and (2) to apply the oligarchic growth model to various disk models. We adopt a planetesimal disk with Da=a 1 and consider planetary accretion from not only the minimum-mass disk model but also various disk models. We investigate the dependence of the characteristics of protoplanet systems on the disk models through global N-body simulations along with analytical arguments. Recently, Thommes, Duncan, & Levison (2002b) applied the oligarchic growth model to the formation of giant planets in the outer solar system. They developed a semianalytical model of the evolution of planetesimals and compared their model with global semi–N-body simulations of planetary accretion in which planetesimalplanetesimal interactions are neglected. Their simulations are in agreement with their model for the growth timescale of protoplanets, but their masses tend to be smaller than the prediction of their model, which may be due to the edge effect of the simulation range. The migration of planetesimals due to gas drag is considered in their model. However, it is not taken into account for simplicity in the present study. We will discuss the effect of this simplification later. 667 In x 2, we summarize oligarchic growth of protoplanets (Kokubo & Ida 1998, 2000). The method of calculation is presented in x 3. The results of N-body simulations are presented in xx 4 and 5. We discuss the diversity of planetary systems based on the oligarchic growth model in x 6. Section 7 is devoted to summary and discussion. 2. OLIGARCHIC GROWTH OF PROTOPLANETS In the oligarchic growth stage, protoplanets grow by accreting planetesimals while gravitationally interacting with one another. We summarize the oligarchic growth of protoplanets in the standard disk model (Kokubo & Ida 1998, 2000) and then generalize it for general power-law disk models. 2.1. Disk Model We introduce the power-law disk model that consists of solids (rock and ice) and gas. We express the solid surface mass density as a g cm2 ; ð1Þ solid ¼ fice 1 1 AU where a is the distance from a central star, 1 is the reference surface density at 1 AU, is the power index of the radial distribution, and fice is the ice factor that expresses the increase of solids by ice condensation over the snow boundary asnow at which the disk temperature equals the ice condensation temperature ’170 K. The location of asnow depends on the disk temperature profile. We adopt the temperature profile for an optically thin disk given by 1=2 1=4 a L 2 K; ð2Þ T ’ 2:8 10 1 AU L where L is the luminosity of the central star (Hayashi 1981). For L ¼ L , we have asnow ’ 2:7 AU. In the minimum-mass disk model, 1 ’ 7, ¼ 3=2, and fice ¼ 1 ða < asnow Þ and 4.2 ða > asnow Þ (Hayashi 1981). Note that the variation of L corresponds to the shift of asnow . The amount of gas is proportional to that of solids with a constant gas-to-dust ratio fgas , a gas ¼ fgas 1 g cm2 : ð3Þ 1 AU We adopt fgas ¼ 240 (Hayashi 1981). For the temperature profile equation (2), the density distribution of the gas disk on the disk equator is given by 5=4 a g cm3 ; ð4Þ gas ¼ 1 1 AU where 1 is p the ffiffiffi reference density at 1 AU defined as 1 ¼ fgas 1 =ð H1 Þ and H1 is the scale height of the gas disk at 1 AU. For L ¼ L , we have H1 ¼ 0:047 AU and 1 ¼ 2:0 109 ð fgas =240Þð1 =10Þ. Kornet et al. (2001) studied the dust migration in a gas disk, which leads to changing distribution of planetesimals. We can apply the following discussion to their end state of planetesimal distribution. In this case, the gas-to-dust ratio should be changed according to dust migration. 668 KOKUBO & IDA 2.2. Growth Timescale In the oligarchic growth stage, protoplanets grow in a swarm of planetesimals. The growth mode of protoplanets is orderly growth where mass ratios of neighbor protoplanets tend to be unity. Still in this stage, however, the mass ratios between protoplanets and planetesimals increase. As a result, the mass distribution is divided into a small number of large protoplanets and a large number of small planetesimals. The growth rate of a protoplanet with mass M and radius R is given by (e.g., Ida & Makino 1993) dM 2GMR ’ Csolid 2 2 ; dt he ia ð5Þ he2 i1=2 where G is the gravitational constant, is the rms eccentricity of planetesimals, is the Kepler angular velocity, and C is an accretion acceleration factor of a few that reflects the effects of the eccentricity and inclination distributions of planetesimals and the solar gravity (e.g., Ida & Nakazawa 1988, 1989; Greenzweig & Lissauer 1992). The growth timescale of protoplanets M=ðdM=dtÞ is given by (e.g., Kokubo & Ida 2000) 1 1=3 1=3 p C M Tgrow ’ 4:8 103 f 1 h~e2 i 2 1026 g 2 g cm3 1=2 1 solid a M 1=6 yr ; ð6Þ 1 AU 10 g cm2 M where h~e2 i1=2 is the reduced rms eccentricity given by h~e2 i1=2 ¼ he2 i1=2 =h and h is the reduced single Hill radius of a protoplanet defined as M 1=3 ; ð7Þ h¼ 3M p is the material density of planetesimals, M is the mass of the central star, and we introduced the radius enhancement factor f (see x 3.3). In the oligarchic growth stage, he2 i1=2 increases in proportion to M 1=3 , which means that h~e2 i1=2 is independent of M (Ida & Makino 1993), while in the runaway growth stage, he2 i1=2 is independent of M, so h~e2 i1=2 / M 1=3 (e.g., Kokubo & Ida 1996). 2.3. Orbital Separation The orbital separation b of protoplanets is kept ~be5 by orbital repulsion of adjacent protoplanets (Kokubo & Ida 1995), where ~ b ¼ b=rH and rH is the mutual Hill radius of protoplanets given by 2M 1=3 a: ð8Þ rH ¼ 3M Orbital repulsion is a coupling effect of gravitational scattering between protoplanets and dynamical friction from planetesimals. In protoplanet-protoplanet (distant) gravitational scattering, protoplanets tend to increase their orbital eccentricities and inclinations and expand their orbital separation. After scattering, dynamical friction from planetesimals damps the eccentricities and the inclinations. As a result, protoplanets expand their orbital separation, keeping nearly circular noninclined orbits. Vol. 581 As protoplanets grow, the orbital separation measured by the Hill radius decreases since rH / M 1=3 and the orbital repulsion increases the orbital separation. This means that there is a typical orbital separation realized in the oligarchic growth stage. Kokubo & Ida (1998) found that the typical orbital separation of protoplanets is ~b ’ 10 and it hardly depends on the disk and protoplanet parameters according to the f ¼ 4–6, gas-free calculations. Kokubo & Ida (2000) confirmed this value with f ¼ 1 (real scale) calculations with gas. In the following, we use ~b ¼ 10 for scaling of results. 2.4. Equilibrium Eccentricity In the oligarchic growth stage, the random velocity (eccentricity e and inclination i) of planetesimals is raised by viscous stirring by protoplanets and is damped by gas drag. The random velocity of planetesimals is estimated as v ’ he2 i1=2 vK , where vK is the Kepler velocity. The timescale of viscous stirring by protoplanets is given by (Ida & Makino 1993) TVS ’ G 2 n v3 ; 2 M M ln ð9Þ where the surface number density of protoplanets is estimated as nM ’ 1 2ab2ai ð10Þ and ln is the Coulomb logarithm. The gas drag timescale for a planetesimal with mass m and radius r is given by Tgas ’ 2m ; CD r2 gas u ð11Þ where CD is the drag coefficient and u is the relative velocity between the planetesimal and gas (Adachi, Hayashi, & Nakazawa 1976). Equating the above two timescales, we obtain the equilibrium eccentricity as 2=15 ~ 1=5 1=15 p b m 2 1=2 h~e ieq ’ 5:6 23 3 10 10 g 2 g cm 1=5 1=5 1=5 gas CD a ; 1 AU 1 2 109 g cm3 ð12Þ where we used u ’ v, he2 i1=2 ¼ 2hi2 i1=2 , and ln ¼ 3 weakly de(Stewart & Ida 2000). Note that h~e2 i1=2 eq pends on m. 2.5. Isolation Mass By extrapolating the oligarchic growth to the end of protoplanet accretion, we estimate the final mass or ‘‘ isolation mass ’’ of protoplanets. For simplicity, the present oligarchic growth model assumes that planetary accretion proceeds locally; in other words, no considerable migration of protoplanets and planetesimals occurs (for the application of planetesimal migration to the oligarchic growth model, see Thommes et al. 2002b). Given the orbital separation of protoplanets, the oligarchic growth model predicts the No. 1, 2002 FORMATION OF PROTOPLANET SYSTEMS isolation mass as (Kokubo & Ida 2000) Miso ’ 2absolid !3=2 !3 !3=2 ~ b solid a ¼ 0:16 10 1 AU 10 g cm2 1=2 M M ; M ð13Þ planet systems formed through oligarchic growth. We discuss the diversity of protoplanet systems in x 5. We confirm the above oligarchic growth model expanded to the power-law disk models by global N-body simulations in the subsequent sections. 3. METHOD OF CALCULATION where M is the Earth mass. In this case, the mutual Hill radius of protoplanets with Miso is given by !1=2 !1=2 ~ b solid rH ’ 0:69 10 10 10 g cm2 2 a M 1=2 AU : 1 AU M 669 We perform global N-body simulations of planetary accretion starting from various initial distributions of planetesimals. For simplicity, we consider gas-free cases. The effects of the neglect of the gas disk are discussed in x 7. 2 3.1. Disk Model and Initial Conditions ð14Þ In the oligarchic growth mode, the number of protoplanets in a certain radial range is inversely proportional to b ¼ ~brH and thus rH . 2.6. Characteristics of Protoplanets for Power-Law Disk Models We apply the oligarchic growth model to the power-law disk models given by equations (1) and (4). From equations (6) and (12)–(14), the growth timescale of protoplanets, the equilibrium eccentricity of planetesimals, the isolation mass of protoplanets, and their Hill radius for the power-law disk models are given by Tgrow !2 !1=3 !1 h~e2 i1=2 M fice 1 ’ 1:7 10 f 1026 g 6 10 1=2þ 1=6 a M yr ; ð15Þ 1 AU M 5 1 h~e2 i1=2 eq !1=5 !1=15 !1=5 ~ fgas b m ’ 5:6 10 1023 g 240 ð1=5Þþ1=20 1=5 1 a ; 1 AU 10 Miso ’ 0:16 ~ b 10 !3=2 a 1 AU fice 1 10 ð3=2Þð2Þ ~ b 10 ð16Þ TABLE 1 Initial Conditions of Planetesimal Disks !3=2 M M 1=2 M ; ð17Þ !1=2 !1=2 fice 1 rH ’ 0:69 10 10 ð1=2Þð4Þ a M 1=2 AU ; 1 AU M 2 We adopt an axisymmetric surface density distribution of a planetesimal disk given by equation (1) with inner and outer cutoffs, ain and aout . We set ain ¼ 0:5 AU and aout ¼ 1:5 AU in all models. In most models, planetesimals have equal masses. In a model (model 4), we adopt the power-law mass distribution, ndm / m dm with dynamic range mmax =mmin ¼ 20. The number of bodies is N ¼ 5000 and 10,000. We set the internal density of planetesimals p ¼ 2 g cm3. The initial conditions of planetesimal disks are summarized in Table 1. In Table 1, ¼ 1 represents an equal-mass distribution. In all the disk models, the initial eccentricities and inclinations of planetesimals are given by the Rayleigh distribution with dispersions he2 i1=2 ¼ 2hi2 i1=2 ¼ 0:02 (Ida & Makino 1992a). We call the disk model with 1 ¼ 10 and ¼ 3=2 the standard model, which is similar to the minimum-mass disk model and was often adopted in the previous N-body simulations. Because observationally inferred disk mass ranges from 0.1 to 10 times the minimum-mass disk model, we calculate the cases with 1 ¼ 1 and 100. We also consider the disks with flatter ( ¼ 12) and more centrally concentrated ( ¼ 5=2) radial density profiles. For each model, we perform two runs with different random numbers for the initial distributions. Because the results are similar for the two runs, we show the results of one of the two runs in the following. ð18Þ where we used p ¼ 2 g cm3 , C ¼ 2, and CD ¼ 1. The above estimates imply interesting characteristics of proto- Model 1 N Mdisk (g) f 1............... 2............... 3............... 4............... 5............... 6............... 7............... 8............... 1 10 10 10 100 ’10 10 ’10 3/2 3/2 3/2 3/2 3/2 1/2 3/2 5/2 1 1 1 5/2 1 1 1 1 10000 10000 10000 10000 10000 5000 5000 5000 1.5 1027 1.5 1028 1.5 1028 1.5 1028 1.5 1029 1.5 1028 1.5 1028 1.5 1028 10 10 6 6 6 6 6 6 Note.—Parameters 1, , , N, Mdisk , and f stand for the reference surface density at 1 AU, the power index of radial distribution, the power index of mass distribution, the number of planetesimals, the total mass of the planetesimal disk, and the radius enhancement factor, respectively. 670 KOKUBO & IDA 3.2. Orbital Integration The orbits of planetesimals are calculated by numerically integrating the equation of motion of planetesimals, N X xi xj dvi xi ¼ GM Gmj ; 3 dt jxi j3 jx i xj j j6¼i ð19Þ where x and v are the position and velocity of planetesimals, respectively. The right-hand side of equation (19) represents the gravity of the central star and the mutual gravitational interaction of planetesimals. We set M ¼ M . For numerical integration, we use the modified Hermite scheme for planetary N-body simulation (Kokubo, Yoshinaga, & Makino 1998) with the hierarchical time step (Makino 1991). The most expensive part of the Hermite scheme is the calculation of the mutual gravitational force and its time derivative, whose cost increases in proportion to N2. We calculate the force and its first time derivative by directly summing up interactions of all pairs on the specialpurpose computer for N-body simulation, HARP/GRAPE (Makino, Kokubo, & Taiji 1993; Makino et al. 1997). Vol. 581 times the initial planetesimal mass form on nearly circular noninclined orbits. Protoplanet growth propagates from inner to outer disk. This is because the growth timescale is smaller for small a since the surface number density of planetesimals is higher and the orbital period is smaller for small a as shown by equation (15). For ¼ 3=2, Tgrow / a3 since h~e2 i1=2 / a1=2 in a gas-free case (eq. [21]). Planetesimal accretion is complete in the inner disk, while it is still on the way in the outer disk. The eccentricities and inclinations of protoplanets are kept small because of dynamical friction from planetesimals. The velocity dispersion of planetesimals is as large as the surface escape velocity of protoplanets. The mass and orbital separation of the protoplanet system are consistent with the oligarchic growth model. The resultant characteristics of protoplanets in models 2, 3, 4, and 7 (1 ¼ 10; ¼ 3=2) are all similar. In the following, we investigate the evolution of spatial, velocity, and mass distributions of the system and the characteristics of protoplanets in model 3. 4.2. Spatial Distribution 3.3. Accretion The evolution of the surface mass density distribution is plotted in Figure 2. We found that the overall density profile hardly changes in the simulation period. The bumpy structure of the profile is due to the discrete distribution of protoplanets and not important. This constancy of the density profile is necessary for the oligarchic growth model to hold. The reason for the almost time independent density distribution is that the diffusion timescale of the disk due to gravitational scattering among planetesimals is much larger than the growth timescale of protoplanets. Using the viscous stirring timescale of planetesimals, we estimate the diffusion timescale as !2 !2 !1 Da he2 i1=2 m 10 Tdiff ’ 1:3 10 1 AU 1023 g 0:01 5=2 1 solid a yr ; ð20Þ 1 AU 10 g cm2 We adopt an f-fold radius of bodies to accelerate the accretion process and thus save calculation time. The use of an f-fold radius changes the growth timescale but not the growth mode as long as f is not too large (for details, see Kokubo & Ida 1996). The timescale of planetary accretion is reduced by a factor of f2 at most in gas-free cases. It should be noted that the use of f affects the stage of giant impacts among protoplanets, where the orbital elements of planets formed by giant impacts relatively strongly depend on f. In the present calculation, we do not follow this stage. For simplicity, we assume that two planetesimals always accrete when they contact. In accretion, the position and velocity of the center of mass are conserved. The lack of collisional fragmentation or rebound seems to make no significant change in the growth mode of protoplanets (Wetherill & Stewart 1993). Note that migration of fragments due to gas drag may change the surface density profile, which we discuss later. It is worth noting that we should be careful when we use f in the outer disk where planetesimals are loosely bound to the system and thus their ejection due to scattering by protoplanets is effective. In this case, the radius enhancement allows accretion to continue past the point where, in reality, it would stall. where Da is the width of the disk (for derivation, see the Appendix). For small planetesimals that form the main disk, Tdiff is much longer than Tgrow of protoplanets. Even if f ¼ 1 is adopted and scattering by protoplanets is considered, Tdiff 4Tgrow still holds. Thus, the density profile remains almost constant. 4. FORMATION OF PROTOPLANET SYSTEMS 4.3. Velocity Distribution We investigate in detail the result of the standard disk model (1 ¼ 10 and ¼ 3=2) from which the solar system may form. With the global N-body simulations, we confirm the oligarchic growth model constructed based on the results of local N-body simulations. The evolution of the rms eccentricity and inclination distributions of planetesimals is shown for four disk zones (a < 0:75 AU, 0:75 AU < a < 1:0 AU, 1:0 AU < a < 1:25 AU, 1:25 AU < a) in Figure 3. The initial values are the same for all the zones. Viscous stirring increases the eccentricity and the inclination, while dynamical friction reduces those of large bodies. The inner disk is heated up first for the same reason as the short growth timescale for the inner disk. For me1025 g, as a result of dynamical friction, d loghe2 i1=2 =d log m ’ d loghi2 i1=2 =d log m ’ 12, which means the equipartition of the random energy (e.g., Stewart & Wetherill 1988; Ida & Makino 1992b). The ratio of eccentricity to inclination for planetesimals is always e=i ’ 2. 4.1. Overall Evolution Figure 1 shows the snapshots of the system on the a-e and a-i planes for model 3. Starting with 10,000 equal-mass (1:5 1024 g) planetesimals, we have calculated the system for 4 105 yr. We employed a sixfold radius of planetesimals. The number of bodies decreases from 10,000 to 333. In 4 105 yr, 12 protoplanets whose mass is larger than 100 No. 1, 2002 FORMATION OF PROTOPLANET SYSTEMS 671 Fig. 1a Fig. 1b Fig. 1.—Snapshots of a planetesimal system on the (a) a-e plane and (b) a-i plane for model 3. The circles represent planetesimals, and their radii are proportional to the radii of the planetesimals. The system initially consists of 10,000 equal-mass (1:5 1024 g) planetesimals in 0.5 AU a 1:5 AU. The numbers of planetesimals are 2612 (5 104 yr), 1409 (105 yr), 673 (2 105 yr), and 333 (4 105 yr). The filled circles represent protoplanets that are larger than 100 times the initial mass, and the length of the line on a protoplanet is 10rH . This relaxation of the orbital elements is characteristic of the Keplerian disk (Ida, Kokubo, & Makino 1993). In a gas-free system, the random velocity of planetesimals is determined mainly by viscous stirring. When the mass of a protoplanet exceeds 100 times the mass of the typical planetesimal, the viscous stirring by the protoplanet becomes dominant (Ida & Makino 1993). In this case, the random velocity of planetesimals becomes as large as the surface escape velocity of the protoplanet vesc ’ eesc vK , where eesc is the escape eccentricity given by (e.g., Safronov 1969) 1=2 1=3 M a 1=2 : ð21Þ eesc ’ 0:081f 1026 g 1 AU For M ¼ 1027 g, a ¼ 1 AU, and f ¼ 6, eesc ’ 0:07 (~eesc ’ 13), which is consistent with the simulation. Note that the eccentricities and inclinations of the protoplanets are always kept small. For growing protoplanets in a planetesimal swarm, this is due to dynamical friction from planetesimals. The relatively large orbital separations ~b ’ 10, and thus the weak gravitational interaction is the reason for the isolated protoplanets (in longer timescales, however, their eccentricities and inclinations can become large because of long-term mutual perturbation). 4.4. Mass Distribution The evolution of the mass distribution is plotted for the four disk zones in Figure 4. The mass distribution first relaxes to the power-law distribution with power index d log nc =d log m ’ 1:5, where nc is the cumulative number of bodies, which is characteristic of runaway growth (Makino et al. 1998; Kokubo & Ida 2000). Then the slope of the mass distribution gradually becomes gentle because in this system there is no supply of small planetesimals. The evolution of the mass distribution of large planetesimals or protoplanets agrees with the result of the statistical growth equation by Wetherill & Stewart (1993). The equivalence between the results of the statistical approach and N-body simulation is shown in detail in Inaba et al. (2001). 672 KOKUBO & IDA Vol. 581 It is clearly shown that the mass distribution is divided into two components, small planetesimals with mass m 1 10 1024 g and large protoplanets with mass M 1027 g. This bimodal mass distribution is the result of runaway and oligarchic growth of protoplanets. 4.5. Characteristics of Protoplanets Fig. 2.—Radial profile of the surface mass density of planetesimals at t ¼ 0 yr (solid line), 5 104 yr (dotted line), 105 yr (short-dashed line), 2 105 yr (long-dashed line), and 4 105 yr (dotted-dashed line) for model 3. The evolution of the protoplanet system on the a-M plane is plotted in Figure 5. We found that 12 protoplanets of about the predicted isolation mass M 0:1 M are formed with the orbital separation ~b ¼ 13:7 2:5, which is almost consistent with the oligarchic growth model. The two curves in Figure 5 are analytical estimates of the protoplanet growth for ~b ¼ 10 and 15 based on the oligarchic growth model. Using he2 i1=2 ¼ eesc , from equation (5) we predict the growth of a protoplanet in the oligarchic growth stage in a gas-free system as " solid 1=3 2 C MðtÞ ’ M0 þ 32f 2 10 g cm2 3=2 #3 a t g; ð22Þ 1 AU 1 yr where M0 is the initial mass of the protoplanet. The Fig. 3b Fig. 3a Fig. 3.—Snapshots of the rms (a) eccentricity and (b) inclination distributions against mass at t ¼ 5 104 , 105, 2 105 , and 4 105 yr for model 3; he2 i1=2 and hi2 i1=2 are calculated in four divided zones: a < 0:75 AU (circles), 0:75 AU < a < 1:0 AU ( pentagons), 1:0 AU < a < 1:25 AU (squares), and 1:25 AU < a (triangles). No. 1, 2002 FORMATION OF PROTOPLANET SYSTEMS Fig. 4.—Cumulative number of bodies is plotted against mass at t ¼ 5 104 , 105, 2 105 , and 4 105 yr for model 3. The mass distribution is calculated in four divided zones: a < 0:75 AU (solid line), 0:75 AU < a < 1:0 AU (dotted line), 1:0 AU < a < 1:25 AU (short-dashed line), and 1:25 AU < a (long-dashed line). The initial number of bodies is 3037, 2656, 2223, and 2084 from inner to outer zones. 673 Fig. 5.—Mass of protoplanets with mass larger than 100 times the initial mass is plotted against semimajor axis at t ¼ 5 104 , 105, 2 105 , and 4 105 yr for model 3. The two curves are the analytical estimates of the protoplanet growth based on the oligarchic growth model for ~b ¼ 10 (solid line) and 15 (dotted line). 5. DIVERSITY OF PROTOPLANET SYSTEMS analytical estimate of the protoplanet growth is given by M ¼ min½MðtÞ; Miso . The results of the N-body simulation agree well with the analytical estimates with C ¼ 2. Protoplanets grow to the isolation mass, and the ‘‘ accretion wave ’’ propagates from inner to outer disk. Note that when M is small the analytical model overestimates the rms eccentricity of planetesimals and thus underestimates the growth rate. This is because small growing protoplanets cannot stir up neighbor planetesimals up to eesc within Tgrow since Tgrow of the small protoplanets is shorter than TVS . This effect leads to the fact that the protoplanets in the N-body simulation grow faster than those of the analytical model when they are small. We confirmed the oligarchic growth of protoplanets in the standard disk model by the global simulation ranging from 0.5 to 1.5 AU. The reason the oligarchic growth model holds is that accretion proceeds locally; in other words, the radial distribution of planetesimals is not greatly affected by planetesimal dynamics and accretion. In this section, a wide variety of power-law disk models are adopted and the results of N-body simulations are compared with those predicted by the oligarchic growth model. We investigate the dependence of characteristics of protoplanet systems on the disk mass (1) and the disk profile (). 5.1. Disk Mass Dependence First we investigate the 1 dependence of protoplanet systems. The variation of 1 corresponds to the variation of the disk mass when is fixed. Figures 6 and 7 show the snapshots of the protoplanet systems on the a-e and a-M planes for 1 ¼ 1 (model 1), 10 (model 3), and 100 (model 5). The power index of the radial distribution is fixed as ¼ 3=2. In Figure 7, the isolation mass of protoplanets for ~b ¼ 10 and ~b ¼ 15 is shown for 1 ¼ 1, 10, 100. In the disk with 1 ¼ 100, at 105 yr, six large protoplanets with M ’ 1 5 M are formed with the orbital separation ~b ¼ 11:5 1:7, which is in agreement with the oligarchic growth model. Two more relatively small protoplanets are 674 KOKUBO & IDA Vol. 581 Fig. 7.—Same as Fig. 6 but on the a-M plane for protoplanets. The two lines show the estimated isolation mass for ~b ¼ 10 (solid line) and 15 (dotted line). Fig. 6.—Snapshots of planetesimal systems on the a-e plane for model 1 (1 ¼ 1) at 6 105 yr, model 3 (1 ¼ 10) at 4 105 yr, and model 5 (1 ¼ 100) at 105 yr. The circles represent planetesimals, and their radii are proportional to the radii of the planetesimals. The systems initially consist of 10,000 equal-mass [1:5 ð1023 1025 Þ g] planetesimals in 0.5 AU a 1:5 AU. Filled circles represent protoplanets that are larger than 100 times the initial mass, and the length of the line on a protoplanet is 10rH . on the outside of the Figure 6 region, ða; eÞ ¼ ð1:48; 0:43Þ and (1.73, 0.16), which is the result of close encounters between protoplanets. The eccentricities of protoplanets in the 1 ¼ 100 case are relatively highly pumped up because the mass of protoplanets is relatively large and thus the gravitational interaction is relatively strong. Note that the predicted isolation mass of protoplanets around 1 AU is several M , which may cause the onset of gas accretion from the gas disk with the result that the protoplanets become Jovian planets (e.g., Ikoma et al. 2000). We will discuss this case in the next section. In the disk with 1 ¼ 1, at 6 105 yr, 31 protoplanets b ¼ 15:1 5:1. with M ’ 0:003 0:01 M are formed with ~ This orbital separation is a little bit larger than the typical value ~ b ¼ 10. This difference of orbital separation is due to protoplanet-protoplanet collisions. For protoplanet sysb ’ 10, and e i 0:01, the orbital tems with M 1026 g, ~ instability timescale is on the order of 105 yr (Yoshinaga, Kokubo, & Makino 1999). This means that, to some extent, the protoplanet-protoplanet collisions already took place to form larger protoplanets with larger orbital separation than those predicted by the oligarchic growth model especially in the inner disk. If protoplanets with separation ~b ¼ 10 collide, the resultant new separation for the merged protoplanet becomes ~b ’ 15, whose instability timescale is much larger than the simulation period if e i 0:01. It is clearly shown that the number of protoplanets decreases with 1 while the mass of protoplanets increases with 1. The oligarchic growth model predicts that the isola3=2 tion mass of protoplanets depends as Miso / 1 and the 1=2 1 (eqs. [17] number of protoplanets nM / rH / 1 and [18]). The results of the N-body simulations are consistent with this prediction. 5.2. Disk Profile Dependence Next we consider the -dependence of protoplanet systems. We compare three cases with ¼ 12 (model 6), 3/2 (model 7), and 5/2 (model 8) with the fixed disk mass Mdisk ¼ 1:5 1028 g in between 0.5 and 1.5 AU. It should be noted that in a real protoplanetary disk, is not a free parameter but determined by some physical processes in the disk. Because we do not know how the initial planetesimal distribution is determined, it is of importance to explore the -space. Figure 8 shows the protoplanet mass against semimajor b ¼ 10 axis for t ¼ 2 105 yr with the isolation mass for ~ and 15. As the oligarchic growth model predicts, the results show that the mass of protoplanets depends on a as M / að3=2Þð2Þ , with ~b ¼ 10–15. Note that in the outer disk, accretion is still on the way. The oligarchic growth model means that if the initial disk is centrally condensed ( > 2) Miso decreases with a and if the initial disk has a flatter distribution ( < 2) Miso increases with a. In the standard model ¼ 3=2, we have larger protoplanets on larger semimajor axes. The radial profile index is an important factor that determines the mass distribution of No. 1, 2002 FORMATION OF PROTOPLANET SYSTEMS 675 planetary systems formed from the diverse protoplanet systems. 6. DIVERSITY OF PLANETARY SYSTEMS The oligarchic growth model describes the growth of solid (rocky and icy) protoplanets. Based on the oligarchic growth model together with the gas giant planet formation scenario, we discuss the possible diversity of planetary systems. The gas giant planet is a dominant member of planetary systems that controls the basic architecture of planetary systems. We use the oligarchic growth model for the power-law disks, equations (15), (16), and (17), and assume that the orbital separation of protoplanets is scaled by the Hill radius and ~b ¼ 10. For simplicity, we also assume that all the solid materials are finally incorporated into planets in situ. Thus, in the following discussion, the isolation mass and the growth timescale of protoplanets are the maximum and the minimum values. We discuss the effect of planetesimal migration in x 7. The disk parameters are set as fice ¼ 4:2, fgas ¼ 240, and asnow ¼ 2:7 AU. We adopt planetesimals with m ¼ 1023 g and set M ¼ M . 6.1. Formation of Planets from Protoplanets Fig. 8.—Mass of protoplanets plotted against semimajor axis at t ¼ 2 105 yr for models 6 ( ¼ 1=2), 7 ( ¼ 3=2), and 8 ( ¼ 5=2) with the isolation mass for ~ b ¼ 10 (solid line) and 15 (dotted line). protoplanets, which leads to formation of diverse planetary systems as will be shown in the next section. In summary, we confirmed that oligarchic growth holds for the power-law disk models with 1 ¼ 1 100 and ¼ 1=2 5=2. The dependence of protoplanet systems on 1 and is schematically illustrated in Figure 9. Given the initial distribution of planetesimals, we can immediately obtain the protoplanet systems that are formed from the planetesimals based on the oligarchic growth model. The variation of 1 and leads to diversity of protoplanet systems. In the next section, we discuss possible diversity of α 2 a a a a Σ1 Fig. 9.—Schematic illustration of the diversity of protoplanet systems on the 1- plane. Four examples of protoplanet systems are illustrated. [See the electronic edition of the Journal for a color version of this figure.] Protoplanets form through oligarchic growth in the gas disk. The fate of protoplanets depends on their mass and growth timescale and the lifetime of the gas disk. From the observation of protoplanetary disks, the lifetime of protoplanetary disks Tdisk is estimated as 106–107 yr (Strom, Edwards, & Skrutski 1993; Zuckerman, Forveille, & Kastner 1995) and the amount of the gas necessary for a Jovian planet would remain for a few times longer than the estimated disk lifetime (Thi et al. 2001). The protoplanet system with ~b ¼ 10 is orbitally stable as long as 0.1%–1% of the gas disk remains because the gas disk–planet interaction damps the eccentricities of protoplanets (Iwasaki et al. 2002). In other words, in the gas disk, no accretion among protoplanets occurs. Thus, we can use the isolation mass and the growth timescale for the isolation mass to consider the final stage of planet formation. A gas giant or Jovian planet is formed by gas accretion onto a protoplanet (solid core) by the gravity of the protoplanet. In the core instability scenario, the onset of gas accretion occurs when the mass of a core exceeds a critical mass Mcr ’ 5 15 M (e.g., Mizuno 1980; Bodenheimer & Pollack 1986; Pollack et al. 1996). Note that the cores of Jovian planets are not formed by accretion among protoplanets in the gas disk as mentioned above. In the standard disk model, terrestrial planets form through giant impacts among protoplanets whose mass is smaller than the critical core mass after most of the gas disk is depleted (e.g., Kominami & Ida 2002). Uranian planets would be protoplanets themselves whose growth timescale is longer than the lifetime of the gas disk in their vicinity (e.g., Kokubo & Ida 2000). In the following, we consider the diversity of planetary systems according to this concept of final planet formation. 6.2. Conditions for Gas Giant Planet Formation Gas accretion begins when the growth timescale of a protoplanet becomes longer than the contraction timescale of the planetary atmosphere (Ikoma et al. 2000). Planetesimal 676 KOKUBO & IDA Vol. 581 accretion plays the role of heat source to support the planetary atmosphere. At the end of the oligarchic growth stage, planetesimal accretion almost ceases and the growth timescale becomes formally infinity. Then gas accretion begins. The contraction timescale Tcont is given by 8 Miso 5=2 9 Tcont 10 10 yr ; ð23Þ M which decreases rapidly with the isolation mass of protoplanets (Ikoma et al. 2000). If Miso is relatively small, gas accretion proceeds slowly. However, gas accretion becomes increasingly rapid as the total planetary mass increases (e.g., Pollack et al. 1996; Ikoma et al. 2000). The lifetime of protoplanetary disks leads to two conditions for the formation of gas giant planets: (1) the contraction timescale must be shorter than the lifetime of the disk: Tcont < Tdisk ; ð24Þ which is equivalent to the conventional condition that Miso > Mcr , where we define critical core mass as the minimum mass satisfying Tcont < Tdisk ; and (2) in order to capture the gas, protoplanet growth must be completed before the depletion of the gas: Tgrow < Tdisk : Fig. 10a ð25Þ In other words, only protoplanets with Tcont < Tdisk (Miso > Mcr ) and Tgrow < Tdisk can become gas giants. This means that there is the limited range of semimajor axis for gas giant planet formation. We adopt Tcont ¼ 108 ðMiso =M Þ5=2 yr and scale results with Tdisk ¼ 108 yr in the following discussion. These values are favorable for gas giant planet formation. The qualitative feature of the following discussion, however, hardly changes as long as Tcont and Tdisk are in reasonable ranges. 6.3. Habitat Segregation of Planets In Figure 10, we plot the isolation mass of protoplanets against the semimajor axis for 1 ¼ 1, 10, 100 and ¼ 3=2; 5=2. As discussed in the previous section, Miso increases with a for ¼ 3=2, while Miso decreases with a for ¼ 5=2. The isolation mass jumps at the snow border. Note that asnow sets a rough boundary between terrestrial (rocky) and Uranian (icy) planets. Substituting equations (16) and (17) into equation (15), we obtain the growth timescale of protoplanets with Miso in the gas disk ~ 1=10 9=10 b 1 1=2 Tgrow ’ 3:2 105 fice 10 10 ð9þ16Þ=10 a yr : ð26Þ 1 AU In Figure 11, Tgrow and Tcont for Miso are drawn against the semimajor axis for 1 ¼ 1, 10, 100 and ¼ 3=2, 5/2. The dependence of the contraction timescale on the semimajor axis for < 2 is qualitatively different from that for > 2. 5=2 15=4 ð15=4Þð2Þ a . Thus, Tcont For Miso , Tcont / Miso / 1 decreases with a for ¼ 3=2, while Tcont increases with a for ¼ 5=2. This difference leads to different formation ranges for Jovian planets. In the following, we focus on the ¼ 3=2 case, which has the same radial profile as the standard disk model. The Fig. 10b Fig. 10.—Isolation mass of protoplanets with ~b ¼ 10 against the semimajor axis for disks with 1 ¼ 1, 10, and 100 for (a) ¼ 3=2 and (b) ¼ 5=2. The discontinuity at asnow ¼ 2:7 AU (snow border) is due to ice condensation. application of a similar discussion to other disks is, however, straightforward. For < 2, the condition Tcont < Tdisk sets the inner boundary for gas giant formation. The condition Tgrow < Tdisk draws the outer boundary. From equation (24), we obtain the inner boundary as amin gas ’ 2 12fice Tdisk 108 yr !8=15 ~b 10 !2 1 10 !2 AU : ð27Þ No. 1, 2002 FORMATION OF PROTOPLANET SYSTEMS 677 max Fig. 12.—Habitat segregation of planets based on amin gas , agas , and the snow border asnow for disks with Tdisk ¼ 107 yr (dotted line) and 108 yr (solid line) against 1 for ¼ 3=2 and ~b ¼ 10. Fig. 11a 1 in Figure 12. The territories of terrestrial, Jovian, and Uranian planets are determined by the following rule: max Jovian planets can form in amin gas dadagas , and for terrestrial and Uranian planets, adasnow and aeasnow except for the Jovian planet zone, respectively. This is the basic habitat segregation of planets for < 2 disks and consistent with the present solar system. The short disk lifetime narrows the Jovian territory as shown in equations (27) and (28). The qualitative segregation pattern, however, is independent of the disk lifetime. In the standard disk model 1 ’ 10, it is difficult to form the cores of Jovian planets in the solar system within Tdisk ¼ 107 yr. For that, a massive (1 50) disk is required, which is consistent with Thommes et al. (2002b). It is also possible to form them in a disk with Tdisk ¼ 108 yr and 1 10. The increase of the disk lifetime has the same effect on the habitat segregation of planets as the increase of the disk mass. In the rest of this section, as an example, we investigate the Tdisk ¼ 108 yr case in detail. Fig. 11b Fig. 11.—Growth timescale of the isolation mass protoplanet with ~ b ¼ 10 and the contraction timescale of the planetary atmosphere against the semimajor axis for 1 ¼ 1 (dotted line), 10 (solid line), and 100 (dashed line) with (a) ¼ 3=2 and (b) ¼ 5=2. For ¼ 3=2, d log Tgrow =d log a > 0 and d log Tcont =d log a < 0, and for ¼ 5=2, d log Tgrow =d log a > d log Tcont =d log a > 0. The outer boundary is obtained from equation (25) as !2=59 !20=59 !18=59 ~ b Tdisk 1 10=59 max agas ’ 5:5fice AU : 10 108 yr 10 ð28Þ max We draw habitat segregation of planets based on amin gas , agas , and the snow border asnow for Tdisk ¼ 107 and 108 yr against 6.4. Disk Mass Dependence We discuss the possible diversity of planetary systems using the above habitat segregation of planets. The observation of protoplanetary disks by Beckwith & Sargent (1996) revealed that the protoplanetary disk has a wide range of mass, 0.1–10 times the mass of the minimum-mass disk model. We investigate the cases 1 ’ 1, 10, 100 with Tdisk ¼ 108 yr. 6.4.1. Light Disk (1 ’ 1) In light disks with 1 ’ 1, we have Miso ’ 0:05 M at 1 AU and 0:3 M at 10 AU. In this case, to form gas giants, the lifetime of the gas disk must be on the order of 109 yr, which is far longer than the estimated lifetime of the gas disk. Therefore, gas giants would not form at all in this case. Figure 12 shows that light disks with 1 d3 would not have any gas giants. Note that even if giant impacts among 678 KOKUBO & IDA protoplanets form more massive planets than Mcr , they cannot become gas giants since giant impacts are possible only after the depletion of most of the gas (Iwasaki et al. 2002; Kominami & Ida 2002). A planetary system formed from the light disk would consist of many relatively small solid planets, terrestrial planets inside the snow border, and Uranian planets outside the snow border. 6.4.2. Massive Disk (1 ’ 100) For the disk as massive as 1 ’ 100, Miso ’ 5 M at 1 AU, which is large enough for gas accretion within Tdisk . Gas giants can form in the inner disk (a 1 AU). Furthermore, in the massive disks, the growth timescale of protoplanets is so short that Tgrow < Tdisk even at large a. Therefore, several gas giants would form in relatively massive disks with 1 e30. Uranian planets would form outside the Jovian planets. We will discuss the massive disk case in relation to the origin of observed extrasolar planets in more detail below. 6.4.3. Medium (Standard) Disk (1 ’ 10) In the disk with 1 ’ 10, a planetary system similar to the solar system is expected. In this disk, gas giants can form only in the limited range beyond the snow border. This range depends on Tdisk . For Tdisk 108 yr, one or two gas giants may form between the snow border and about 10 AU. In this case, we have terrestrial planets, Jovian planets, and Uranian planets from inner to outer system. In Figure 13, we schematically summarize the predicted diversity of planetary systems produced by the disk mass variation for disks with < 2. It should be noted that in the oligarchic growth model we assumed the accretion in the gas disk. However, by definition, Tgrow of Uranian planets beyond the Jovian planet zone exceeds Tdisk . After the dispersal of the gas disk, the random velocity of planetesimals is pumped up as high as the escape velocity of protoplanets. This high random velocity makes the accretion process slow and inefficient and thus Tgrow longer. This accretion inefficiency is a severe problem Mdisk T cont <Tdisk Tgrow<Tdisk Vol. 581 for the formation of Uranian planets in the solar system (Levison & Stewart 2001). One possible solution to this problem is that Uranian planets form in the Jovian planet region and are subsequently transported outward (Thommes, Duncan, & Levison 1999, 2002a). 6.5. Origin of Extrasolar Planets The disk mass dependence of planetary systems suggests that the number of Jovian planets increases with the disk mass. However, initially formed Jovian planet systems would not be the final configuration of planetary systems since planetary systems with more than three giant planets may not be stable systems in the long term (e.g., Chambers, Wetherill, & Boss 1996; Marzari & Weidenschilling 2002). A planetary system of several gas giants may become orbitally unstable against long-term mutual perturbations. After the ejection of some planets or merging, orbitally stable planets in eccentric orbits would remain, which may correspond to observed extrasolar planets in eccentric orbits (Rasio & Ford 1996; Weidenschilling & Marzari 1996; Lin & Ida 1997; Marzari & Weidenschilling 2002). In addition, interactions between gas giants and a residual relatively massive gas disk may lead to significant orbital decay to a central star (e.g., Lin & Papaloizou 1993), which may correspond to extrasolar planets with short orbital periods (hot Jupiters) such as 51 Peg b (Lin, Bodenheimer, & Richardson 1996). If an extremely massive disk with 1 e200 (Mdisk e0:3 M for ¼ 3=2) is considered, Figure 12 suggests that in situ formation of hot Jupiters at a 0:05 AU such as 51 Peg b, And b, etc., may be possible. However, dust particles may be evaporated at a 0:05 AU in the disk, which inhibits planetesimal formation, and/or ultraviolet and X-ray radiation from a T Tauri star may strip the gas envelope of a young gas giant (Lin et al. 1996). Hence, the migration model may be favored for hot Jupiter formation. On the other hand, in situ formation of extrasolar planets in circular orbits around a ’ 0:2 AU such as CrB b and HD 192263 b is likely to occur in relatively massive disks with 1 e100 (Mdisk e0:15 M ). The inhibition processes for in situ formation for hot Jupiters do not apply to this case. It is difficult for the migration (Lin et al. 1996) or the slingshot model (Rasio & Ford 1996) to explain planets in circular orbits at a ’ 0:2 AU because tidal interaction or the magnetic field of a host star, which circularizes orbits, may be weak there. In situ formation in relatively massive disks may be most promising. 7. SUMMARY AND DISCUSSION a Fig. 13.—Schematic illustration of the diversity of planetary systems against the initial disk mass for < 2. The left large circles stand for central stars. The double circles (cores with envelopes) are Jovian planets, and the others are terrestrial and Uranian planets. [See the electronic edition of the Journal for a color version of this figure.] Terrestrial and Uranian planets and solid cores of Jovian planets form through accretion of planetesimals. In planetary accretion, oligarchic growth of protoplanets is a key process that controls the basic structure of planetary systems. We confirmed that the oligarchic growth model generally holds in the wide variety of planetesimal disks solid ¼ 1 ða=1 AUÞ g cm2 with 1 ¼ 1, 10, 100 and ¼ 1=2; 3=2; 5=2 by performing global N-body simulations. We derived how the characteristics of protoplanet systems depend on the initial disk mass (1) and the initial disk profile (). The oligarchic growth model gives the growth timescale and the isolation mass as equations (15) and (17), respectively, which are in good agreement with the No. 1, 2002 FORMATION OF PROTOPLANET SYSTEMS results of the N-body simulations. The isolation mass 3=2 increases in proportion to 1 , and the number of proto1=2 planets decreases in proportion to 1 . Because the isolað3=2Þð2Þ , for < 2 it increases tion mass depends on a as a with a while it decreases with a for > 2. The growth timescale increases with a but decreases with 1. These characteristics of protoplanets provide a base for considering the diversity of planetary systems. Based on the oligarchic growth model and the gas giant planet formation scenario, we discussed the diversity of planetary systems. Gas giant planets can form only in the disk range where the contraction timescale of the atmosphere of a core and the growth timescale of the core are shorter than the lifetime of the gas disk. The contraction timescale decreases with the isolation mass as 5=2 Tcont / Miso , and the growth timescale increases with the isolation mass and the semimajor axis as 1=3 Tgrow / Miso aþ1=2 . We derived the disk range where Jovian planets can form as a function of the disk parameters. For example, for ¼ 3=2 and Tdisk 108 yr, several Jovian planets would form from massive disks with 1 e30 with Uranian planets outside the Jovian planets. Only terrestrial and Uranian planets would form from light disks with 1 d3. Solar system–like planetary systems would form from medium disks with 1 ’ 10 (a more massive disk with 1 50 is required for Tdisk 107 yr). This diversity of planetary systems suggests that there exist many extrasolar terrestrial planets. The present detection probability of extrasolar planets around solar-type stars is a few percent. Because of the observational selection effect, most extrasolar planets so far discovered have relatively small semimajor axes and large masses (e.g., Marcy, Cochran, & Mayor 2000). These extrasolar planetary systems might correspond to the planetary systems formed from massive disks. A massive disk may form several giant planets. The long-term orbital instability of the planetary system and/or the (type II) migration of the giant planets due to gas disk–planet interaction may lead to a system similar to those of the extrasolar planets (Lin & Papaloizou 679 1993). The other disks with smaller mass, which are the majority of protoplanetary disks (Beckwith & Sargent 1996), may form terrestrial planets if planet formation proceeds in a similar way as described here. In the present discussion, potentially important processes due to the gas disk are not considered, namely, (type I) migration of protoplanets and migration of planetesimals due to gas drag. The migration of protoplanets due to tidal interaction with the gas disk (e.g., Goldreich & Tremaine 1980; Ward 1986, 1997; Papaloizou & Larwood 2000) would modify the oligarchic growth of protoplanets. Protoplanets can be lost, spiraling into the central star. On the other hand, Tanaka & Ida (1999) showed the possibility that the inward migration of a protoplanet accelerates the growth of the protoplanet since planetesimals with low random velocity are supplied as the protoplanet migrates. The migration speed of protoplanets is, however, still uncertain at present (Tanaka, Takeuchi, & Ward 2002). Another gas disk–planet interaction, the damping of e and i of protoplanets, does not affect the oligarchic growth. It strengthens the effects of dynamical friction from small planetesimals so that the isolation of protoplanets is reinforced. Small planetesimals also migrate because of gas drag. In particular, fragmentation of planetesimals at collisions and subsequent migration of fragments due to gas drag may change the initial spatial distribution of planetesimals. In fact, the migration of small planetesimals reduces the available disk mass for planetary accretion and thus leads to smaller masses for protoplanets than in the migrationless case as shown by Thommes et al. (2002b). It is possible to extend the oligarchic growth model to take into account the migration processes of protoplanets and planetesimals. Before integrating the effects of the migration into the oligarchic growth model, however, we need to clarify each process in more detail. We propose the present discussion on the diversity of planetary systems as a piece of framework to consider possible planetary systems in the universe. APPENDIX DIFFUSION OF A PLANETESIMAL DISK BY SELF-GRAVITY In a planetesimal disk, the random velocity of planetesimals is given by v ’ he2 i1=2 vK . From the conservation of the Jacobi energy, we have db2 4 d½ðeaÞ2 þ ðiaÞ2 4 a2 ðe2 þ i2 Þ ’ ’ ; 3 dt 3 TVS dt ðA1Þ where b is the orbital separation of planetesimals and TVS is the viscous stirring (relaxation) timescale due to planetesimalplanetesimal gravitational interactions given by v3 ; TVS ’ pffiffiffi 2G2 nm m2 ln ðA2Þ where nm is the surface number density of planetesimals (e.g., Ida 1990). Using equation (A2), we obtain the diffusion timescale Tdiff ðDaÞ2 db2 =dt ’ 1:3 1010 Da 1 AU !2 he2 i1=2 0:01 !2 m 1023 g !1 solid 10 g cm2 !1 a 1 AU !5=2 where we used nm ’ solid =ð2hi2 i1=2 amÞ, he2 i1=2 ¼ 2hi2 i1=2 , and ln ¼ 3 (Stewart & Ida 2000). M M !3=2 yr ; ðA3Þ 680 KOKUBO & IDA REFERENCES Aarseth, S. J., Lin, D. N. C., & Palmer, P. L. 1993, ApJ, 403, 351 Lissauer, J. J., Pollack, J. B., Wetherill, G. W., & Stevenson, D. J. 1985, in Adachi, I., Hayashi, C., & Nakazawa, K. 1976, Prog. Theor. Phys., 56, Neptune and Triton, ed. D. Cruikshank (Tucson: Univ. Arizona Press), 1756 37 Agnor, C. B., Canup, R. M., & Levison, H. F. 1999, Icarus, 142, 219 Makino, J. 1991, PASJ, 43, 859 Beckwith, S. V. W., & Sargent, A. I. 1996, Nature, 383, 139 Makino, J., Fukushige, T., Funato, Y., & Kokubo, E. 1998, NewA, 3, 411 Bodenheimer, P., & Pollack, J. B. 1986, Icarus, 67, 391 Makino, J., Kokubo, E., & Taiji, M. 1993, PASJ, 45, 349 Chambers, J. E., & Wetherill, G. W. 1998, Icarus, 136, 304 Makino, J., Taiji, M., Ebisuzaki, T., & Sugimoto, D. 1997, ApJ, 480, 432 Chambers, J. E., Wetherill, G. W., & Boss, A. P. 1996, Icarus, 119, 261 Marcy, G. W., Cochran, W. D., & Mayor, M. 2000, in Protostars and Goldreich, P., & Tremaine, S. 1980, ApJ, 241, 425 Planets IV, ed. V. Mannings, A. P. Boss, & S. S. Russell (Tucson: Univ. Greenberg, R., Wacker, J., Chapman, C. R., & Hartman, W. K. 1978, Arizona Press), 1285 Icarus, 35, 1 Marzari, F., & Weidenschilling, S. J. 2002, Icarus, 156, 570 Greenzweig, Y., & Lissauer, J. J. 1992, Icarus, 100, 440 Mizuno, H. 1980, Prog. Theor. Phys., 64, 544 Hayashi, C. 1981, Prog. Theor. Phys. Suppl., 70, 35 Papaloizou, J. C. B., & Larwood, J. D. 2000, MNRAS, 315, 823 Hayashi, C., Nakazawa, K., & Nakagawa, Y. 1985, in Protostars and Pollack, J. B., Hubickyj, O., Bodenheimer, P., Lissauer, J. J., Podolak, M., Planets II, ed. D. C. Black & M. S. Mathews (Tucson: Univ. Arizona & Greenzweig, Y. 1996, Icarus, 124, 62 Press), 1100 Rasio, F. A., & Ford, E. B. 1996, Science, 274, 954 Ida, S. 1990, Icarus, 88, 129 Safronov, V. S. 1969, Evolution of the Protoplanetary Cloud and Ida, S., Kokubo, E., & Makino, J. 1993, MNRAS, 263, 875 Formation of the Earth and the Planets (Moscow: Nauka) Ida, S., & Makino, J. 1992a, Icarus, 96, 107 Stewart, G. R., & Ida, S. 2000, Icarus, 143, 28 ———. 1992b, Icarus, 98, 28 Stewart, G. R., & Wetherill, G. W. 1988, Icarus, 74, 542 ———. 1993, Icarus, 106, 210 Strom, S. E., Edwards, S., & Skrutski, M. F. 1993, in Protostars and Planets III, ed. E. H. Levy & J. I. Lunine (Tucson: Univ. Arizona Press), 837 Ida, S., & Nakazawa, K. 1988, Prog. Theor. Phys. Suppl., 96, 211 Tanaka, H., & Ida, S. 1999, Icarus, 139, 350 ———. 1989, A&A, 224, 303 Tanaka, H., Takeuchi, T., & Ward, W. R. 2002, ApJ, 565, 1257 Ikoma, M., Nakazawa, K., & Emori, H. 2000, ApJ, 537, 1013 Thi, W. F., et al. 2001, Nature, 409, 60 Inaba, S., Tanaka, H., Nakazawa, K., Wetherill, G. W., & Kokubo, E. Thommes, E. W., Duncan, M. J., & Levison H. F. 1999, Nature, 402, 635 2001, Icarus, 149, 235 Iwasaki, K., Emori, H., Nakazawa, K., & Tanaka, H. 2002, PASJ, 54, 471 ———. 2002a, AJ, 123, 2862 Kokubo, E., & Ida, S. 1995, Icarus, 114, 247 ———. 2002b, Icarus, submitted ———. 1996, Icarus, 123, 180 Ward, W. R. 1986, Icarus, 67, 164 ———. 1998, Icarus, 131, 171 ———. 1997, Icarus, 126, 261 ———. 2000, Icarus, 143, 15 Weidenschilling, S. J. 1977, Ap&SS, 51, 153 Kokubo, E., Yoshinaga, K., & Makino, J. 1998, MNRAS, 297, 1067 Weidenschilling, S. J., & Marzari, F. 1996, Nature, 384, 619 Kominami, J., & Ida, S. 2002, Icarus, 157, 43 Weidenschilling, S. J., Spaute, D., Davis, D. R., Marzari, F., & Ohtsuki, K. Kornet, K., Stepinski, T. F., & Rozyczka, M. 2001, A&A, 378, 180 1997, Icarus, 128, 429 Levison, H. F., & Stewart, G. R. 2001, Icarus, 153, 224 Wetherill, G. W., & Stewart, G. R. 1989, Icarus, 77, 330 Lin, D. N. C., Bodenheimer, P., & Richardson, D. C. 1996, Nature, 380, ———. 1993, Icarus, 106, 190 606 Yoshinaga, K., Kokubo, E., & Makino, J. 1999, Icarus, 139, 328 Lin, D. N. C., & Ida, S. 1997, ApJ, 477, 781 Zuckerman, B., Forveille, T., & Kastner, J. H. 1995, Nature, 373, 494 Lin, D. N. C., & Papaloizou, J. C. B. 1993, in Protostars and Planets III, ed. E. H. Levy & J. I. Lunine (Tucson: Univ. Arizona Press), 749
© Copyright 2026 Paperzz