1: \documentclass[12pt,preprint]{aastex}

2: 

3: \shorttitle{late formation of extrasolar giant planets} 

4: 

5: \shortauthors{Ida and Lin}

6: 

7: \begin{document}

8: 

9: \title{Toward a Deterministic Model of Planetary Formation IV:}

10: \title{Effects of Type-I Migration}

11: 

12: 

13: \author{S. Ida}

14: \affil{Tokyo Institute of Technology,

15: Ookayama, Meguro-ku, Tokyo 152-8551, Japan}

16: \email{ida@geo.titech.ac.jp}

17: 

18: \and 

19: 

20: \author{D. N. C. Lin}

21: \affil{UCO/Lick Observatory, University of California, 

22: Santa Cruz, CA 95064, USA}

23: \affil{Kavli Institute of Astronomy \& Astrophysics, Peking

24: University, Beijing, China}

25: \email{lin@ucolick.org}

26: 

27: \begin{abstract}

28: In a further development of a deterministic planet-formation model

29: (Ida \& Lin 2004), we consider the effect of type-I migration of

30: protoplanetary embryos due to their tidal interaction with their

31: nascent disks.  During the early embedded phase of protostellar disks,

32: although embryos rapidly emerge in regions interior to the ice line,

33: uninhibited type-I migration leads to their efficient self-clearing. 

34: But, embryos continue to form from residual planetesimals at

35: increasingly large radii, repeatedly migrate inward, and provide a

36: main channel of heavy element accretion onto their host stars. During

37: the advanced stages of disk evolution (a few Myr), the gas surface

38: density declines to values comparable to or smaller than that of the

39: minimum mass nebula (MMSN) model and type-I migration is no longer an

40: effective disruption mechanism for mars-mass embryos. Over wide

41: ranges of initial disk surface densities and type-I migration

42: efficiency, the surviving population of embryos interior to the ice

43: line has a total mass several times that of the Earth.  With this

44: reservoir, there is an adequate inventory of residual embryos to

45: subsequently assemble into rocky planets similar to those around the

46: Sun.  But, the onset of efficient gas accretion requires the emergence

47: and retention of cores, more massive than a few $M_\oplus$, prior to

48: the severe depletion of the disk gas.  The formation probability of

49: gas giant planets and hence the predicted mass and semimajor axis

50: distributions of extrasolar gas giants are sensitively determined by

51: the strength of type-I migration.  We suggest that the observed

52: fraction of solar-type stars with gas giant planets can be reproduced

53: only if the actual type-I migration time scale is an order of

54: magnitude longer than that deduced from linear theories.  We also show

55: that the introduction of such slower type-I migration rate makes

56: simulated planetary mass-period distribution more consistent with

57: observation.

58: 

59: 

60: \end{abstract}

61: \keywords{extrasolar planets -- planetary systems: 

62: formation -- solar system: formation}

63: 

64: \section{Introduction}

65: In the radial velocity survey of nearby stars, more than 200 

66: extrasolar planets with mass $M_p$ comparable to that of Jupiter $M_J$ have

67: been discovered. Current statistics indicates that $\eta_J>7\%$ of all

68: nearby F, G, and K dwarfs on various search target lists have gas

69: giant planets with semimajor axis $<5$AU \citep[e.g.,][]{Marcy05,

70: Mayor05}.  But the extrapolation of existing data suggests a higher

71: fraction of solar-type stars may have longer-period gas giant 

72: planets \citep{Cumming}.  

73: 

74: In the previous papers of this series \citep[][hereafter referred to

75: as papers I, II, and III] {IL04a, IL04b, IL05}, we carried out several

76: sets of simulations with a numerical prescription, which is based on

77: the sequential accretion scenario.  We used these results to explain

78: the observed mass and orbital distributions of extrasolar planets and

79: to constrain the intrinsic physics of planet formation through

80: detailed comparisons with the observed data.

81: 

82: In these previous investigation, we have taken into account the effect

83: of gas depletion by assuming that the gas surface density $\Sigma_g$

84: declines everywhere exponentially over a characteristic timescale

85: $\tau_{\rm dep} \simeq 1$--10 Myr. With an $\alpha$ prescription for

86: the turbulent viscosity, this assumed global evolution of $\Sigma_g$

87: can lead to declining accretion rates which are consistent with those

88: observed \citep{Hartmann98}. The magnitude of $\tau_{\rm dep}$ sets a

89: strong constraint on the gas giant planet building efficiency

90: \citep[e.g.,][]{IL04a}.

91: 

92: We have also assumed that the dust surface density ($\Sigma_d$) in

93: these disks is preserved from the initial value we have adopted.  The

94: dust grains' mass inferred from the mm observation of continuum

95: radiation from protostellar disks \citep[e.g.,][]{BS96} appears to

96: have a wide dispersion centered around the value of the minimum-mass

97: solar nebula (MMSN) model \citep{Hayashi81}.  We adopted a similar

98: distribution for $\Sigma_d$ centered around that of MMSN.  This

99: assumption, though it greatly simplified our treatment, is less well

100: justified.  In the scaling of $\Sigma_d$ with that of MMSN, we have

101: inherited the assumption that all the heavy elements were locally

102: retained during the epoch of planet formation. 

103: 

104: There are two potential avenues for the depletion of heavy elements:

105: hydrodynamic gas drag of small dust grains and type-I migration of

106: cores due to the tidal interaction with their nascent disks.  Here,

107: "cores" mean the protoplanetary embryos which formed as a result of

108: runaway/oligarchic growth of accretion of planetesimals \citep{KI98,KI02}.

109: These cores are generally not sufficiently massive to initiate runaway

110: gas accretion.  Without much additional growth before the disk gas

111: is severely depleted, these cores would become either rocky planets or

112: ice giants.  

113: 

114: In the present paper, we consider the dominant processes after the

115: formation of planetary embryos which are hydrodynamically decoupled

116: from disk gas motions.  Prior to the phase of rapid gas accretion,

117: these embedded cores and their surrounding gas exchange angular

118: momentum as they engage in tidal interaction \citep{GT80,

119: LP79}. Analytic studies suggest that isolated cores lose angular

120: momentum to the disk exterior to their orbits faster than that they

121: gain from the disk interior to their orbits. This torque imbalance

122: leads to a ``type-I'' migration. From the linear analysis, the

123: characteristic orbital decay time scale of Earth-mass cores at several

124: AU's in a MMNM is much shorter than 1 Myr \citep{Ward86, Tanaka02}.

125: Accordingly, embryos formed well outside the ice line may migrate to

126: the present locations of the Earth and Venus. But, the mostly

127: refractory compositions of the terrestrial planets in the present-day

128: solar system suggest that they probably did not migrate to their

129: present location from regions well beyond the ice line.  

130: 

131: Today, there remains considerable uncertainties in the efficiency of

132: this process (see \S 2.2).  Nevertheless, their formation could have

133: been preceded by a much larger population of cores which did migrate

134: into the Sun.  In view of this uncertainty, we carry out a parametric

135: study on the decline of the cores' accretion rate associated with a

136: $\Sigma_d$ reduction due to the type-I migration of cores (see below).

137: This effect has been neglected in our previous papers.

138: 

139: Heavy elements are also depleted from the disks through ``type-II''

140: migration.  Due to runaway gas accretion onto a core, the growing gas

141: giant planet eventually acquires sufficient mass to open a gap in the

142: disk and lock itself into a ``type-II'' migration with the viscous

143: evolution of the disk gas \citep{LP85,LP93}.  In the context of solar

144: system formation, \citet{Lin86, Lin95} has speculated that, when the

145: solar nebula was massive, several gas giants may have formed,

146: undergone type-II migration, and eventually merged with the Sun.  When

147: the disk's $\Sigma_g$ decreases below that of the MMSN model, the

148: migration has stalled.  According to this scenario, Jupiter and Saturn

149: could be the last survivors of several generations of protoplanets

150: \citep[also see][]{Papaloizou06}, although it may not be easy to form

151: terrestrial planets after preceding gas giants have passed through the

152: inner disk region \citep{Armitage03}.  A resurgence of this migration

153: and survival scenario \citep{Lin96} have been stimulated by the

154: observational discoveries of close-in gas giant planets \citep[e.g.,]

155: []{Marcy05,Mayor05}. But, a majority of the known extrasolar planets

156: have periods much longer than a few days.  We showed in Papers I and

157: II that their observed logarithmic period distribution can be used to

158: infer comparable timescales for giant planets' migration and disk depletion.

159: However, based on the relative abundance between planets with $a<0.06$

160: AU and those between 0.2-2AU's from their host stars, we have to

161: assume that up to 90\% of the gas giants which migrated to the

162: proximity of their host stars may have perished.

163: 

164: Here, we focus on the repeated clearing due to type-I migration.  

165: We suggest that this self-regulated clearing mechanism limits and

166: determines the amount of heavy elements retained by the cores and

167: residual planetesimals (also see discussion on metallicity homogeneity

168: of open clusters in \S 4.1).  It also results in late formation of

169: gas giants and the marginal probability of gas giants' formation

170: through a series of failed attempts.

171: 

172: The scenario we consider here is similar to the hypothesis proposed by

173: \citet{CW06} in the context of satellite formation around Jovian

174: planets. They suggested that the total mass of the surviving

175: satellites is self-regulated by their type-I migration through their

176: nascent circum-planetary disk which is continuously replenished.  In

177: the context of planet formation, \citet{McNeil05} carried out N-body

178: simulations of terrestrial planets' accretion and growth including the

179: effect of type-I migration. They found that it is possible to retain

180: sufficient amount of solid materials (planetesimals and embryos) to

181: assemble Earth-mass planets within a disk lifetime $\sim 10^6$ years

182: provided the type-I migration rate is slightly slower than that

183: predicted by the linear theory.  In these simulations, the initial

184: disk mass in terrestrial planet regions is assumed to be 3--4 times of

185: that of MMSN.  Based on an analytic approximation for type-I

186: migration, \citet{Kominami06} simulated the evolution of $\Sigma_d$

187: under various disk conditions.  Their results indicate that the

188: depletion region expands outwards, starting from the disk's inner

189: boundary.  They also found that in disks with same dust-to-gas ratio,

190: the asymptotic surface density of the retained embryos decreases with

191: the initial value of $\Sigma_d$ (and $\Sigma_g$) because type-I

192: migration is faster in more massive disks.

193: 

194: Here we consider a much large range of initial conditions. With a

195: comprehensive numerical prescription, we consider the possibility of

196: multiple generation of embryo formation. We show that in relatively

197: massive disks although type-I migration is more effective for

198: individual planetesimals and embryos, it does not necessarily lead to

199: more rapid depletion of the residual population.  In these disks,

200: migration starts with smaller individual masses but more generations

201: of embryo form and parish before the initial supply of planet-building

202: blocks is severely depleted. Our results indicate that the total

203: retainable mass of heavy elements only weakly depends on the initial

204: disk mass.

205: 

206: Although type-I migration places a mass limit on the retainable

207: embryos, it does not quench the formation of Earth-like terrestrial

208: planets because they can be assembled from retainable low-mass embryos

209: after the gas is depleted \citep{Kominami06}. Gas giant planets,

210: however, must form in gas-rich disks which appears to be depleted on the

211: time scale $\tau_{\rm dep} \sim$ a few Myr.  In addition, they must

212: acquire $\sim 10M_{\oplus}$ cores prior to the onset of efficient gas

213: accretion.  Although they can form rapidly outside the ice lines on

214: massive disks, early generation of such massive cores quickly migrate

215: into their host stars.  But, \citet{Thommes06} showed that cores

216: formed after the disk gas has been severely depleted may withstand the

217: disruption by the declining type-I migration. In \S3 of this paper, we

218: present results which are consistent with that obtained by

219: \citet{Thommes06}. We also incorporate gas accretion onto cores and

220: the effect of type-II migration and find that under some

221: circumstances, there is adequate residual gas to promote the efficient

222: gas accretion and the formation of gas giant planets.  This ``late

223: formation'' scenario is conceptually consistent with that inferred

224: from the noble gas enrichment in Jupiter \citep{Guillot06}.  Based on

225: this model, we are able to reproduce the observed mass-period

226: distribution of the known gas giant planets around solar type stars

227: provided the efficiency of type-I migration is at least an order of

228: magnitude slower than that derived from the linear theory (see \S3).

229: Our conclusion is consistent with the studies by 

230: \citet{libert05} who

231: also found that a necessary condition for the formation of Jupiter and

232: Saturn is a 30-times reduction in the type-I migration rate.

233: 

234: In metal-poor disks, the formation of critical mass cores requires a

235: relatively large gas surface density.  In these disks, type-I

236: migration is more effective in clearing the cores prior to the onset

237: of gas accretion.  Consequently, the formation of gas giant planets is

238: suppressed.  Assuming the disks' initial metallicity is identical to

239: that of their host stars, this effect sharpens the dependence of

240: the predicted formation efficiency of gas giants on the stellar

241: metallicity which is well established in the observational data

242: \citep{Fischer05}.

243: 

244: Since type-I and II migrations are the essential processes which

245: determine the properties of emerging planets, we briefly recapitulate,

246: in \S2, their basic physical principles.  We also incorporate a

247: quantitative prescription of these processes in our existing

248: comprehensive model for planet formation.  The model for the new

249: addition, type-I migration and the decrease in planetesimal surface 

250: density due to the migration, is explained in

251: detail.  In \S3, we carry out a systematic study on how cores'

252: migration may affect the formation of terrestrial planets and gas

253: giant planets.  We show that type-I migration delays formation of gas

254: giants, and that it leads to the mass and semimajor axis ($M_p-a$)

255: distribution consistent with that of the known gas giants around G

256: dwarfs.  We also study dependence of the the distribution on planet

257: formation parameters, metallicity.  The introduction of relatively

258: slow type-I migration results in the metallicity dependence that may

259: be consistent with observation.  Finally, we summarize our results and

260: discuss their implications in \S4.

261: 

262: 

263: \section{Planet formation and migration model}

264: \label{sec:prescription}

265: In this paper, we simulate the $M_p-a$ distribution of extrasolar

266: planets, taking into account the effects of cores' type-I migration as

267: well as type-II migration.  In Papers I-III, we outlined in detail a

268: quantitative prescription which we used to model the evolution of

269: planetesimals, gas accretion onto protoplanetary cores, the

270: termination of gas giant planet growth, and type-II migration.  In

271: this section we briefly recapitulate the dependence of both types of

272: migration on the background disk properties.

273: 

274: In this paper, we newly incorporate the effects of type-I migration

275: and examine the impact of type-I migration on the formation of

276: terrestrial planets and the asymptotic properties of emerging gas

277: giant planets.  In light of the theoretical uncertainties, we

278: introduce a prescription which captures the main determining factors

279: of type-I migration and enables us to consider a wide range in the

280: magnitude of its efficiency factor.

281: We here consider only the case of $M_* = 1M_{\odot}$ to 

282: focus on the effects of type-I migration around solar-type stars,

283: although we retain the dependences on stellar mass $M_*$ and 

284: its luminosity $L_*$.

285: 

286: Note that inclusion of type-I migration and the minor changes 

287: in truncation conditions of growth of gas giants and the disk model 

288: that are described below do not change 

289: the conclusions that have been derived in previous papers:

290: (i) the existence of "planet desert" (a lack of intermediate mass planets 

291: at $\la$ a few AU) in Paper I, (ii) the metallicity dependence 

292: (a fraction of stars with giant planets increases with the metallicity) 

293: in Paper II, and (iii) the stellar mass dependence 

294: (giant planets are much less abundant around lower-mass stars) in Paper III.

295: The results with inclusion of type-I migration corresponding to (i) and (ii) are shown 

296: in Figures \ref{fig:ma_C1} and \ref{fig:metal_dep}, respectively.

297: The stellar mass dependence with type-I migration

298: is briefly commented on in \S 3.2.3 and 

299: will be discussed in detail in a separate paper.

300: 

301: 

302: 

303: \subsection{Parameterized disk models}

304: 

305: The main objectives of this series of papers is to examine the

306: statistical properties of emerging planets under a variety of disk

307: environments rather than to study individual processes which regulate

308: the disk structure.  For computational convenience, we introduced, in

309: Papers I-III, multiplicative factors ($f_d$ and $f_g$) to scale

310: $\Sigma_g$ and $\Sigma_d$ such that

311: \begin{equation}

312: \left\{ \begin{array}{ll} 

313: \Sigma_d & = \Sigma_{d,10} \eta_{\rm ice} 

314: f_d (r/ {\rm 10 AU})^{-q_d}, 

315: \label{eq:sigma_dust} \\ 

316: \Sigma_g & = \Sigma_{g,10} f_{g} (r/ {\rm 10 AU})^{-q_g},

317: \label{eq:sigma_gas}

318: \end{array} \right.

319: \end{equation}

320: where normalization factors $\Sigma_{d,10} = 0.32 {\rm g/cm}^2$ and

321: $\Sigma_{g,10} = 75 {\rm g/cm}^2$ correspond to 1.4 times of

322: $\Sigma_g$ and $\Sigma_d$ at 10AU of the MMSN model, and the step

323: function $\eta_{\rm ice} = 1$ inside the ice line at $a_{\rm ice}$

324: (eq.~[\ref{eq:a_ice}]) and 4.2 for $r > a_{\rm ice}$ [the latter can

325: be slightly smaller ($\sim 3.0$) \citep{Pollack94}].

326: 

327: We show below that the disk metallicity [Fe/H]$_d$ is an important

328: parameter which regulates the survival of protoplanetary cores during

329: their type-I migration.  Dependence of disk metallicity is attributed

330: to distribution of $f_{d,0} = f_{g,0} 10^{{\rm [Fe/H]}_d}$.  Solar

331: metallicity corresponds to [Fe/H]$_d = 0$ and $f_{d,0} = f_{g,0}$.

332: The main advantage of these parameterized disk structure models is

333: that they enable us to efficiently simulate and to examine the

334: dominant dependence of planet formation and dynamical evolution on the

335: disk structure.

336: 

337: In self-consistent treatment of accretion flow, the disk temperature

338: is determined by an equilibrium between the viscous dissipation and

339: heat transport \citep{alpha}.  We neglect the detailed energy balance

340: \citep{Chiang97, Garaud07} and adopt the equilibrium temperature

341: distribution in highly optically thin disks prescribed by

342: \citet{Hayashi81} such that

343: \begin{equation}

344: T = 280 \left(\frac{r}{1{\rm AU}}\right)^{-1/2}

345:     \left(\frac{L_*}{L_{\odot}}\right)^{1/4} {\rm K}.

346: \label{eq:temp_dist}

347: \end{equation}

348: In this simple prescription, we set the ice line to be that determined

349: by an equilibrium temperature (eq.~[\ref{eq:temp_dist}]) in optically

350: thin disk regions \citep{Hayashi81},

351: \begin{equation}

352: a_{\rm ice} = 2.7 (L_\ast/L_\odot)^{1/2} {\rm AU},

353: \label{eq:a_ice}

354: \end{equation}

355: where $L_*$ and $L_{\odot}$ are the stellar and solar luminosity.  The

356: magnitude of $a_{\rm ice}$ may be modified by the local viscous

357: dissipation \citep{Lecar06} and stellar irradiation \citep{Chiang97,

358: Garaud07}.  These effects do not greatly modify the disk structure

359: during the late evolutionary stages and they will be incorporated in

360: subsequent papers.

361: 

362: \subsubsection{Evolution of $f_g$}

363: 

364: The surface density of the gas can be determined by a diffusion

365: equation \citep{LP85},

366: \begin{equation}

367: \frac{\partial \Sigma_g}{\partial t} - {1 \over r}

368: \frac{\partial}{\partial r} \left[ 3 r^{1/2}

369: \frac{\partial}{\partial r} (\Sigma_g \nu r^{1/2}) \right] = 0.

370: \label{eq:surf_density_evol}

371: \end{equation}

372: where we neglected the sink terms due to photoevaporation and

373: accretion onto the cores.  For computational convenience, we adopt the

374: standard constant-$\alpha$ prescription for viscosity \citep{alpha},

375: \begin{equation}

376: \nu= \alpha c_s H,

377: \label{eq:alpha}

378: \end{equation}

379: where $c_s$, $H=\sqrt{2}c_s/\Omega_{\rm K}$, and $\Omega_{\rm K}$ are

380: the sound speed, disk scale height, and the Keplerian angular frequency.

381: Although the value of $\alpha$ is not clear, $\alpha \sim 10^{-3}$ may

382: be consistent with the observational data of accretion rates of T

383: Tauri disks \citep[e.g.,][]{Hartmann98} and the results of MHD

384: simulations \citep[e.g.,][]{Sano04}.  We here use $\alpha = 10^{-3}$

385: as a nominal value.

386: 

387: With the $\alpha$ prescription and eq.~(\ref{eq:temp_dist}), $\nu

388: \propto r$ and eq.~(\ref{eq:surf_density_evol}) reduced to a linear

389: partial differential equation for which self-similarity solutions have

390: been presented by \citet{Lynden-Bell74}. The numerical solution for

391: disk-evolution equation [{\it i.e.}, eq.~(\ref{eq:surf_density_evol})

392: with $\alpha = 10^{-3}$], 

393: starting with $\Sigma_g \propto r^{-1.5}$ for

394: $r < r_m$ and an exponential cutoff at $r_m = 10$AU, is illustrated in

395: Figure \ref{fig:disk_vis_evol}a.

396: 

397: After a brief initial transition, the numerical solution quickly

398: approaches to $\Sigma_g \propto r^{-1}$ with an asymptotic exponential

399: cut-off, which is the self-similar solution obtained by

400: \citet{Lynden-Bell74} for a linear viscosity prescription and by

401: \citet{Linbod82} for an $\alpha$ model. As shown in

402: Figure~\ref{fig:disk_vis_evol}b, $\Sigma_g$ in the region $r < r_m$

403: decreases as uniformly independent of $r$.  In the self-similar

404: solution, $\Sigma_g$ at $r < r_m$ decays as $\Sigma_g \propto

405: (t/\tau_{\rm dep} + 1)^{-3/2}$, where

406: \begin{equation}

407: \tau_{\rm dep} = \frac{r_m^2}{3\nu(r_m)} 

408: \simeq 3 \times 10^6 

409:     \left(\frac{\alpha}{10^{-3}}\right)^{-1}

410:     \left(\frac{r_m}{100{\rm AU}}\right)

411:     \left(\frac{M_*}{M_{\odot}}\right)^{-1/2}

412:     {\rm yrs}.

413: \end{equation}

414: In the above, we used eqs.~(\ref{eq:temp_dist}) and

415: (\ref{eq:alpha}).

416:  

417: This self-similar nature of the solution is preserved if we assume a

418: spatially uniform exponential depletion of the disk ({\it i.e.}

419: $\Sigma_g \propto \exp(-t/\tau_{\rm dep})$ with a $r$-independent

420: $\tau_{\rm dep}$), as in Papers I-III, although in the actual

421: self-similar solution, $\Sigma_g \propto t^{-3/2}$ for $t > t_{\rm

422: dep}$.  If the effect of photoevaporation is taken into account,

423: $\Sigma_g$ decays rapidly after it is significantly depleted, so that

424: the exponential decay could be more appropriate.  Thus, we adopt

425: \begin{equation}

426: f_{g} = f_{g,0} \exp(-t/\tau_{\rm dep}),

427: \label{eq:gas_exp_decay}

428: \end{equation}

429: in the disk model in eq.~(\ref{eq:sigma_gas}).  Even if we fix the

430: $\alpha$ value, $\tau_{\rm dep}$ has uncertainty, because we do not

431: have enough knowledge about $r_m$, and $r_m$ should have some

432: dispersion.  We use $\tau_{\rm dep}$ as a parameter and set it to be

433: in a range of $10^6$--$10^7$ yrs, which may be consistent

434: with observation.

435: Corresponding to the self-similar solution, we consider here disk

436: models with $\Sigma_g \propto r^{-1}$ ($q_g = 1$) in addition to the

437: case with $\Sigma_g \propto r^{-3/2}$ ($q_g = 3/2$) that was assumed

438: in Papers I-III corresponding to the MMSN model.  

439: 

440: 

441: \subsubsection{Evolution of $f_d$}

442: 

443: We start each model with $\Sigma_d \propto r^{-3/2}$ ($q_d = 3/2$).

444: In principle, the evolution of $\Sigma_d$ should be treated

445: independently from $\Sigma_g$ even though small dust particles are

446: thermally and dynamically coupled to the gas.  Physically, the

447: magnitude of $\Sigma_d$ is determined by the grains' growth rate and

448: the planetesimals' retention efficiency.  In view of the large

449: uncertainties in these processes, we adopt the radial gradient of

450: conventional MMSN model, $q_d = 3/2$.  Different radial gradient

451: between $\Sigma_d$ and $\Sigma_g$ could be produced by inward

452: migration of dust grains due to gas drag, which tends to make inner

453: disk more metal-rich while outer disk metal-poor \citep{Stepinski97,

454: Kornet}.

455: 

456: In Papers I-III, $f_d$ was assumed to be constant with time until core

457: mass reaches isolation mass.  In the present paper, we take into

458: account planetesimal clearing due to the cores' accretion.  Since we

459: also include type-I migration of the cores, $f_d$ at a given location

460: (semimajor axis $a$) continuously decreases with time as planetesimals

461: are accreted by cores which in term undergo orbital decay.

462: Note that in the present paper, $a$ is identified as $r$, since

463: we neglect evolution of orbital eccentricities.

464: 

465: The full width of feeding zone of a core is given by 

466: \citep{KI98,KI02}

467: \begin{equation}

468: \begin{array}{lll}

469: \Delta a_{\rm c} & = & 10 r_{\rm H} \\

470:   & = & 10 \left( \frac{2M_{\rm c}}{M_{\ast}} \right)^{1/3}a, 

471: \end{array}

472: \end{equation}

473: where $r_{\rm H}$ is two-body Hill radius.  Following Papers I-III, we

474: take into account of the expansion of feeding zones due to collisions

475: among the isolated cores after significant depletion of disk gas

476: \citep{Kominami02,ZLS07}, although this effect influences

477: the results only slightly (it is important in inner regions

478: in which cores are significantly depleted by type-I migration). 

479: 

480: An increase in the cores' mass is uniformly subtracted from the mass

481: in its feeding zone, keeping $q_d$ locally.  When the cores migrate

482: out from the feeding zone, the reduction of $f_d$ at the initial

483: location of the cores is stopped.  But, the cores can continue to

484: accrete planetesimals along their migration path.  In any given system,

485: when a core reaches 0.03AU (occasionally, it can attain sufficient

486: mass to accrete gas and become a gas giant during the course of its

487: migration), a next-generation core is launched at the pre-migration

488: radius with 100 smaller mass.  (Since core growth is faster during

489: earlier stage of disk evolution (eq.~[\ref{eq:m_grow0}]), the choice

490: of initial mass of the next-generation core does not affect the total

491: core accretion timescale.)  Thereafter the growth of the core and the

492: depletion of the planetesimals in its feeding zone resume.  

493: 

494: This process repeats until the residual planetesimals are depleted.

495: Along the cores' migrating paths, they accrete residual planetesimals

496: (until they reach $a < a_{\rm dep,mig}$ given by

497: eq.~[\ref{eq:a_depletion}]; see below)

498: but cannot reduce $f_d$ as fast as the cores formed {\it in

499: situ}. Therefore we neglect the evolution of $f_d$ due to the

500: planetesimal accretion by the migrating cores.  The clearing of the

501: residual planetesimal disk by migrating gas giants, is also neglected

502: because it rarely occurs and is unlikely to affect the overall

503: distribution of the emerging terrestrial planets (see \S 3.2.2).  The

504: overall evolution of $\Sigma_d$ due to core accretion and migration is

505: described in \S 3.1.

506: 

507: \subsection{Core growth and type-I migration}

508: 

509: We briefly recapitulate a comprehensive analysis on the growth of

510: planetesimals and cores.  Readers can find a systematic derivation of

511: these results and the appropriate references from Paper I.  During

512: their growth through cohesive collisions, cores' mass accretion rate

513: (of planetesimals) at any location $a$ and time $t$ is described by

514: (Paper I)

515: \begin{equation}

516: \begin{array}{l}

517: dM_{\rm c}/dt = M_{\rm c}/\tau_{\rm c,acc}; \\

518: \tau_{\rm c,acc} = 2.2 \times 10^5 \eta_{\rm ice}^{-1}

519: f_d^{-1} f_{\rm g}^{-2/5} 10^{(2/5)(3/2-q_g)}

520: \left( \frac{a}{1{\rm AU}} \right)^{27/10 + (q_d-3/2) + (2/5)(q_g-3/2)}

521: \left(\frac{M_\ast}{M_{\odot}} \right)^{-1/6}

522: {\rm yrs},

523: \end{array}

524: \label{eq:m_grow0}

525: \end{equation}

526: where $f_d$ and $f_g$ change with time.  In the derivation of the

527: above expression, we have adopted the mass of the typical field

528: planetesimals to be $m=10^{20}$g. The numerical factor $10^{(2/5)

529: (3/2-q_g)}$ comes from the fact that we scale surface densities at

530: 10AU in eq.~(\ref{eq:sigma_gas}).  

531: 

532: Based on the results of previous numerical simulations (see references

533: in Paper I), we assume that runaway growth phase is quickly

534: transferred to oligarchic growth phase \citep{KI98}.  During the

535: oligarchic growth phase, the cores' accretion timescale is dominated

536: by their late-stage growth ($\tau_{\rm c,acc} \propto M_{\rm

537: c}^{1/3}$) so that dependence on their initial mass is negligible.  In

538: eq.~(\ref{eq:m_grow0}), we adopt $M_{\rm c}(0) = m=10^{20}$g.  When

539: $f_g < 10^{-3}$, we use gas-free accretion rate rather than that in

540: Eq.~(\ref{eq:m_grow0}) (Paper I),

541: \begin{equation}

542: \tau_{\rm c,acc} = 2 \times 10^7 \eta_{\rm ice}^{-1}

543: f_d^{-1} \left( \frac{a}{1{\rm AU}} \right)^{3 + (q_d-3/2)}

544: \left(\frac{M_\ast}{M_{\odot}} \right)^{-1/2}

545: {\rm yrs}.

546: \label{eq:m_grow_e}

547: \end{equation}

548: 

549: In Papers I-III, we artificially terminate the cores' accretion at

550: $M_{\rm c}= M_{\rm c,iso}$, where the isolation mass is given by

551: \begin{equation}

552: M_{\rm c,iso} \simeq

553: 0.16 \eta_{\rm ice}^{3/2} f_d^{3/2} 

554: \left(\frac{a}{1\mbox{AU}}\right)^{3/4-(3/2)(q_d-3/2)} 

555: \left(\frac{M_\ast}{M_{\odot}} \right)^{-1/2} M_{\oplus}.

556: \label{eq:m_iso0}

557: \end{equation}

558: In the present paper, we do not need to adopt an artificial

559: termination to the core growth because when $M_{\rm c}$ approaches

560: isolation $M_{\rm c,iso}$, {\rm in situ} core accretion is

561: automatically slowed down by the reduction of $f_d$. However, in outer

562: regions, ejection of planetesimals by large protoplanets limits their

563: accretion at $M_{\rm e,sca}$ (scattering limit; Paper I). We take this

564: effect into account by adopting an artificial termination for the

565: cores' accretion of planetesimals when they attain a mass $M_{\rm

566: e,sca} \simeq 1.4 \times 10^3 (a/1{\rm AU})^{-3/2}(M_{\ast}/M_{\odot})

567: M_{\oplus}$.

568: 

569: Imbalance in the tidal torques from outer and inner disks causes type-I

570: migration of a core.  Through 3D linear calculation, \citet{Tanaka02}

571: derived the time scale of type-I migration,

572: \begin{equation} 

573: \begin{array}{ll}

574: \tau_{\rm mig1} & = \frac{a}{\dot{a}} 

575: = \frac{1}{C_1}

576:   \frac{1}{2.728 + 1.082 q_g}

577:   \left(\frac{c_s}{a \Omega_{\rm K}}\right)^{2} 

578:   \frac{M_*}{M_p}

579:   \frac{M_*}{a^2 \Sigma_g}

580:   \Omega_{\rm K}^{-1} \\

581:  & \simeq 5 \times 10^4 \times 10^{(3/2-q_g)} \frac{1}{C_1 f_g} 

582:   \left(\frac{M_{\rm c}}{M_{\oplus}} \right)^{-1} 

583:   \left(\frac{a}{1{\rm AU}}\right)^{q_g} 

584:   \left(\frac{M_*}{M_{\odot}}\right)^{3/2}

585:   \;{\rm yrs}.  

586: \end{array}

587: \label{eq:tau_mig1} 

588: \end{equation} 

589: where we used eq.~(\ref{eq:temp_dist}) and $q_g = 1$--1.5.  

590: 

591: In eq.~(\ref{eq:tau_mig1}), we introduce an scaling factor $C_1$ to

592: allow for the retardation of type-I migration due to non-linear

593: effects. In the expression of \citet{Tanaka02}, $C_1 = 1$.  Many

594: simulations have been carried out \citep[see e.g.,][]{Papaloizou06}

595: with different numerical methods and resolutions, protoplanetary

596: potentials and orbits, and disk structures.  They produced a

597: considerable range of the timescale of type-I migration ($\tau_{\rm

598: mig 1}$).  Retardation processes for this type-I migration include

599: variation in the surface density and temperature gradient

600: \citep{Masset06b}, intrinsic turbulence in the disk \citep{Laughlin04,

601: Nelson04}, self-induced unstable flow \citep{Koller04, Li05}, and

602: non-linear radiative and hydrodynamic feedbacks \citep{Masset06a}.

603: Under some circumstances, $C_1$ is reduced to be $\la 0.1$

604: (Dobbs-Dixon et al. 2007, in preparation).  In light of these

605: uncertainties, we adopt $C_1$ as a parameter and mainly examine the

606: slower migration cases with $C_1 < 1$.

607: 

608: Type-I migration can occur before core mass reaches $M_{\rm c,iso}$.

609: In this case, $\Sigma_d$ (equivalently, $f_d$) decreases but does not

610: completely vanish.  From the decreased $\Sigma_d$, new cores accrete

611: with reduced growth rates and isolation masses.  The cycle of core

612: growth, type-I migration, and reduction of $\Sigma_d$ continues until

613: $\Sigma_d$ decreases to the levels from which only cores small enough

614: not to migrate \citep{Kominami06}.  We present a detailed

615: discussion on the self-regulation mechanism in later sections.

616: 

617: Under some conditions, the migrating cores can capture and accumulate

618: planetesimals along their paths \citep{Ward_Hahn, TI99}.  N-body

619: simulation by \citet{Kominami06}, however, showed that the trapping of

620: planetesimals by the cores is tentative and it does not

621: significantly reduce their accretion rates.  In our simulations, we

622: use the accretion rate for migrating cores that is the same as the

623: rate for non-migrating cores (eq.~[\ref{eq:m_grow0}]).  For cores in

624: the systems with $q_g = 1$ and $q_d = 1.5$,

625: \begin{equation}

626: \begin{array}{l}

627: dM_{\rm c}/dt = M_{\rm c}/\tau_{\rm c,acc} \; ; \\

628: \tau_{\rm c,acc} = 3.5 \times 10^5 \eta_{\rm ice}^{-1} 

629: f_d^{-1} f_{g}^{-2/5} 

630: \left( \frac{a}{1{\rm AU}} \right)^{5/2}

631: \left(\frac{M_\ast}{M_{\odot}} \right)^{-1/6} 

632: {\rm yrs}, 

633: \end{array}

634: \label{eq:tau_c_acc2} 

635: \end{equation}

636: \begin{equation}

637: \begin{array}{l}

638: da/dt = a/\tau_{\rm mig1}\; ; \\

639:       \tau_{\rm mig1} \simeq 1.6 \times 10^5 C_1^{-1} f_g^{-1} 

640:   \left(\frac{M_{\rm c}}{M_{\oplus}} \right)^{-1} 

641:   \left(\frac{a}{1{\rm AU}}\right)

642:   \left(\frac{M_*}{M_{\odot}}\right)^{3/2}

643:   \;{\rm yrs}.  

644: \end{array}

645: \label{eq:tau_mig12} 

646: \end{equation}

647: 

648: The magnitude of $f_d$ is consistently decreased by the increase in

649: $M_{\rm c}$ (see \S\ref{subsec:sigma_evol}).  Since the accretion rate

650: is determined by the instantaneous local value of $f_d$, it limits the

651: mass of the migrating cores in a gaseous medium.  Prior to the severe

652: depletion of the disk gas, we quench the cores' accretion of

653: planetesimals at $a < a_{\rm dep,mig}$ (eq.~[\ref{eq:a_depletion}])

654: on the basis that there is an inadequate supply of residual

655: planetesimals in these locations to significantly add to their

656: masses.  But, during the gas depletion, we assume $f_{g}$ decays

657: exponentially. After disk gas is significantly depleted ($f_g  \la

658: 10^{-3}$), Eq.~(\ref{eq:m_grow_e}) is used for $\tau_{\rm c,acc}$.  We

659: set also $da/dt$ is set to be zero at disk inner edge ($\sim 0.03$AU).

660: 

661: \subsection{Formation of gas giant planets}

662: 

663: Prescriptions for formation of gas giant planets are the same as those

664: used in Paper I-III, although the cores' rate of planetesimal

665: accretion is revised from Paper I-III by the incorporation of type-I

666: migration and planetesimal depletion. After the formation of these gas

667: giants, we assume all residual planetesimals in the gap is cleared as

668: consequence of dynamical instabilities. These planetesimals are either

669: accreted by the gas giants or scattered elsewhere \citep{Zhou07}. We

670: neglect the emergence of second-generation cores close to the orbit of

671: gas giants.

672:  

673: In principle, cores with mass much less than that of the Earth can

674: accrete gas.  But, unless the heat released during gas and

675: planetesimal accretion is diffused and radiated away, quasi thermal

676: and hydrodynamic equilibrium would be established to prevent further

677: flow onto the cores. Around low-mass cores, the temperature and

678: density of the envelope are low so that heat cannot be easily

679: redistributed through their envelopes.  But as the cores grow through

680: planetesimal bombardment beyond a mass

681: \begin{equation}

682: M_{\rm c,hydro} \simeq

683: 10 \left( \frac{\dot{M}_{\rm c}}{10^{-6}M_{\oplus}/ {\rm yr}}\right)^{0.25}

684: M_{\oplus},

685: \label{eq:crit_core_mass}

686: \end{equation}

687: both the radiative and convective transport of heat become efficient

688: to allow their envelope to contract dynamically \citep{Ikoma00}.  In

689: the above equation, we neglected the dependence on the opacity in the

690: envelope (see Paper I).  In regions where the cores have already

691: attained isolation, their planetesimal-accretion rate $\dot M_{\rm c}$

692: is much diminished \citep{Zhou07} and $M_{\rm c,hydro}$ can be

693: comparable to an Earth mass $M_\oplus$.  But, gas accretion also

694: releases energy and its rate is still regulated by the efficiency of

695: radiative transfer in the envelope such that

696: \begin{equation}

697: \frac{dM_{\rm p}}{dt} \simeq \frac{M_{\rm p}}{\tau_{\rm KH} },

698: \label{eq:mgsdot}

699: \end{equation}

700: where $M_{\rm p}$ is the planet mass including gas envelope.  In Paper

701: I, we approximated the Kelvin-Helmholtz contraction timescale

702: $\tau_{\rm KH}$ of the envelope with

703: \begin{equation}

704: \tau_{\rm KH} \simeq 10^{k1} 

705: \left(\frac{M_{\rm p}}{M_{\oplus}}\right)^{-k2} \; {\rm yrs}.

706: \label{eq:tau_KH}

707: \end{equation}

708: In order to take into account the uncertainties associated with

709: planetesimal bombardment, dust sedimentation and opacity in the

710: envelope, we adopt a range of values $k1 = 8$--10 and $k2 = 3$--4 (see

711: Papers I and II).  Here we use $k2 = 3$ and treat $k1$ as a parameter.

712: 

713: Gas accretion onto the core is quenched when the disk is depleted

714: either locally or globally.  We assume that gas accretion is

715: terminated if either the thermal condition or global depletion

716: condition is satisfied.  A (partial) gap is formed when the rate of

717: tidally induced angular momentum exchange by the planet with the disk

718: exceeds that of the disk's intrinsic viscous transport \citep{LP85}

719: (the viscous condition),

720: \begin{equation} 

721: M_p > M_{\rm g,vis} 

722: \simeq 30 \left(\frac{\alpha}{10^{-3}}\right) 

723: \left(\frac{a}{1{\rm AU}}\right)^{1/2} 

724: \left(\frac{L_\ast}{L_{\odot}}\right)^{1/4} M_{\oplus}.

725: \label{eq:m_gas_vis} 

726: \end{equation} 

727: Planets with $M_p > M_{\rm g,vis}$ have sufficient mass to induce the

728: partial clearing of the disk near their orbit.  The tidal torque on

729: either side of the gap becomes sufficiently strong to induce the

730: planets to adjust their positions within the gap.  This feedback

731: process leads to a transition from type-I to type-II migration.

732: Therefore, we adopt the viscous condition for the onset of type-II

733: migration.  In Papers I-III, we adopted $\alpha = 10^{-4}$ as a

734: nominal value. In order to match the simulated $M_p-a$ distribution to

735: the observed data, eq.~(\ref{eq:m_gas_vis}) was arbitrarily multiplied

736: by a scaling factor $A_\nu = 10$.

737: 

738: The previously adopted value of $\alpha$ is smaller than that inferred

739: from the models of protostellar disk evolution \citep{Hartmann98}.  In

740: this paper, we use $\alpha = 10^{-3}$ without imposing the scaling

741: factor.  As a result, the condition for the onset of type-II migration

742: is the same as that in Papers I-III.

743: 

744: A clear gap would be formed and gas accretion would be terminated when

745: the planet's Hill radius becomes larger than disk scale height

746: \citep{LP85} (the thermal condition) and it is given by (Paper I)

747: \begin{equation}

748: M_p > M_{\rm g,th}

749: \simeq 0.95 \times 10^3 \left(\frac{a}{1{\rm AU}}\right)^{3/4}

750: \left(\frac{L_\ast}{L_\odot}\right)^{3/8}

751: \left(\frac{M_\ast}{M_\odot}\right)^{-1/2} M_\oplus.

752: \label{eq:m_gas_th}

753: \end{equation}

754: While the thermal condition $r_{\rm H} > 1.5 H$ was used in Papers I-III,

755: we here used the condition $r_{\rm H} > 2 H$, in order to

756: make clear that the asymptotic mass of gas giants is determined

757: by global depletion of disk gas rather than the local thermal condition

758: in the cases with inclusion of type-I migration 

759: in which gas giants are formed in relatively late stages.

760: (It can also reflect the effect of gas flow into the gap mentioned below.)

761: 

762: Some numerical simulations indicate that a residual amount of gas may

763: continue to flow into the gap after both the viscous and thermal

764: conditions are satisfied\citep{D'Angelo03,Kley06,Tanigawa07}.

765: However, recent numerical simulations \citep{Dobbs-Dixon07} also show

766: that the azimuthal accretion flow from the corotation regions onto the

767: planet is effectively quenched despite a diminishing flux of gas into

768: the gap.  Since the thermal condition usually requires larger $M_p$

769: than the viscous condition ($M_{\rm g,th} > M_{\rm g,vis}$), we adopt

770: it as the criterion for the termination of gas accretion in the

771: determination of the asymptotic mass of gas giant planets.  But, we

772: take into a residual amount of gas which may leak through the gap and

773: provide an effective avenue for angular transfer between the inner and

774: outer regions of the disk via the gas giant's corotation resonance 

775: (\S 2.4).

776: 

777: Gas accretion may also be limited by the diminishing amount of

778: residual gas in the entire disk even for planets with 

779: $M_{\rm p} < M_{\rm p,th}$.  In Papers I-III,

780: we assumed that the maximum available mass is determined 

781: simply by $M_{\rm g,no iso} \sim \pi a^2 \Sigma_g$.

782: However, because the global limit plays a more important role

783: when type-I migration is incorporated, we use a more appropriate

784: condition $M_{\rm g,no iso} \sim \int_0^{2a} 2\pi a \Sigma_g da$

785: (We neglect disk gas inflow from far outer regions).

786: For $q_g = 1$,

787: \begin{equation}

788: M_{\rm g,no iso} \simeq  3.5 \times 10^2 f_{\rm g,0} 

789: \exp \left(-\frac{t}{\tau_{\rm dep}}\right) 

790: \left(\frac{a}{1\mbox{AU}}\right)^{1/2} M_{\oplus}.

791: \label{eq:m_gas_non_iso}

792: \end{equation}

793: When $M_{\rm g,noiso}$ diminishes below $M_{\rm p}$ or when $M_{\rm

794: p}$ exceeds $M_{\rm g,th}$, gas accretion is terminated.  

795: 

796: \subsection{Type-II migration}

797: 

798: In our previous analysis in Papers I-III, we adopted a simply analytic

799: prescription for type-II migration: (i) before disk gas mass decays to

800: the value comparable to the planet mass, the planet migrates with

801: (unperturbed) disk accretion and (ii) when disk gas mass is comparable

802: to the planet mass, a fraction ($C_2 \sim 0.1$) of the total (viscous

803: plus advective) angular momentum flux transported by the disk gas

804: (which is assumed to be independent of the disk radius) is utilized by

805: the planet in its orbital evolution. 

806: 

807: If the planet's tidal torque can severely clear a gap in the vicinity

808: of its orbit, $\Sigma_g$ in inner disk region would decrease faster

809: than that in the outer region.  The full torque asymmetry leads to

810: $C_2 \simeq 1$ in case (ii).  But, when the truncation condition is

811: marginally satisfied, the disk interior to the planet's orbit have a

812: total mass $\ga M_p$ such that it can effectively replenish the

813: angular momentum lost by planet to the outer disk region.  There may

814: also be a leakage of gas through the gap region \citep{D'Angelo03}

815: which would suppress the degree of torque asymmetry.  The

816: protoplanet's corotation resonance may drive an effective angular

817: momentum transfer across the two disk regions separated by the

818: protoplanet.  

819: 

820: Protoplanets are formed in disk regions where the midplane is inactive

821: to magneto-rotational instabilities.  It is possible that the modest

822: accretion rate onto the host stars flow through this region via an

823: active layer which is exposed to external ionizing photons and cosmic

824: ray particles \citep{Gammie96}.  Different values of effective $\alpha$'s

825: may contribute to the mass flow through the disk and the planet-disk

826: interaction, especially if there is some leakage across the gap

827: region.  All of these possible scenarios can lead to $C_2 \ll 1$.

828: 

829: The uninterrupted replenishment of gas into the corotation region may

830: also maintain a finite vortensity gradient and an unsaturated

831: corotation torque \citep{Masset06a} which may reduce the efficiency of

832: type-II migration from disk gas accretion even in case (i)

833: \citep{Crida07}.  However, the results of another set of 2D numerical

834: hydrodynamic simulations \citep{D'Angelo06} essentially reproduces the

835: the 1D simulation \citep{LP86} in which the contribution from the

836: corotation resonance is neglected.

837: 

838: In order to take into account of these uncertainties, we reduce here

839: the migration rate by a factor $C_2$ for cases (ii) as in Papers I-III,

840: \begin{equation}

841: \begin{array}{ll}

842: \tau_{\rm mig2,ii} 

843:  & {\displaystyle = \frac{1}{C_2} 

844: \frac{(1/2)M_p \Omega_{\rm K}(a) a^2}

845: {3 \pi \Sigma_{g}(r_m) r_m^2 \nu_m \Omega_{\rm K}(r_m)}} \\

846:  & {\displaystyle \simeq 5 \times 10^5 f_g^{-1} 

847: \left(\frac{C_2 \alpha}{10^{-4}} \right)^{-1} 

848: \left(\frac{M_{\rm p}}{M_{\rm J}} \right) 

849: \left(\frac{a}{1{\rm AU}}\right)^{1/2} 

850: \left(\frac{M_*}{M_{\odot}}\right)^{-1/2}}

851: \;{\rm yrs}.  

852: \end{array}

853: \label{eq:tau_mig2b} 

854: \end{equation} 

855: When $\tau_{\rm mig2,ii}$ is shorter than the migration timescale for (i), 

856: we use the latter timescale, 

857: \begin{equation}

858: \begin{array}{ll}

859: \tau_{\rm mig2,i} 

860:  & {\displaystyle  = \mid \frac{a}{\dot{a}} \mid 

861: = \frac{a}{(3/2) \nu / a} }\\

862:  & {\displaystyle \simeq 0.7 \times 10^5

863: \left(\frac{\alpha}{10^{-3}}\right)^{-1}

864: \left(\frac{a}{1{\rm AU}}\right)

865: \left(\frac{M_*}{M_{\odot}}\right)^{-1/2}}

866: \;{\rm yrs}.

867: \end{array}

868: \label{eq:tau_mig2a} 

869: \end{equation}

870: Due to more accurate estimations,

871: the numerical factors in $\tau_{\rm mig2,i}$ and

872: $\tau_{\rm mig2,ii}$ slightly differ from those

873: in Papers I-III for the same values of $C_2$ and $\alpha$.  

874: However, there are still uncertainties in the formula

875: for $\tau_{\rm mig2,ii}$.  Furthermore, 

876: it is not easy to theoretically evaluate the

877: value of $C_2$ as well as $\alpha$.  

878: Hence, we vary the

879: magnitude of $C_2$ and compare the simulated results with the observed

880: data to obtain a calibration. 

881: For $\alpha \sim 10^{-3}$, in the limit that 

882: $C_2 \sim 1$, most of gas giants are removed from the regions beyond

883: 1AU, which is inconsistent with observed data of extrasolar planets.

884: As shown in Paper II, in order to reproduce $M_p$--$a$ distribution of

885: observed extrasolar planets, disk depletion timescale $\tau_{\rm dep}$

886: must be $\sim \tau_{\rm mig2}$ at a few AU.

887: In the present paper, we mostly adopt 

888: $C_2 \alpha = 10^{-4}$.  

889: For $\alpha \sim 10^{-3}$, it corresponds to $C_2 \sim 1/10$.

890: As a result, $\tau_{\rm mig2,ii}$ we adopt here is almost

891: identical to that used in Papers I-III.  

892: 

893: 

894: \section{Effects of type-I migration}

895: 

896: In our previous simulations in Papers I-III, the effect of type-II

897: migration was included but that of type-I migration was neglected. In

898: this section, we investigate the effects of type-I migration on the

899: planets' $M_p$--$a$ distribution.

900: 

901: \subsection{Surface density evolution due to type-I migration}

902: \label{subsec:sigma_evol}

903: 

904: In order to analyze the asymptotic $M_p-a$ distribution from the Monte

905: Carlo simulations, we first present the results on the $\Sigma_d$

906: reduction due to type-I migration.  The relevant time scale here is a

907: function of $C_1, f_g$ and $M_p$ (eq.~[\ref{eq:tau_mig1}]). The

908: magnitude of the cores' $M_p$ is a function of $f_d$, $f_g$ and $t$.

909: 

910: Since type-I and II migrations involve the tidal interaction of

911: embedded cores with the disk gas, the dynamical clearing of the

912: residual planetesimals is suppressed after the gas is severely

913: depleted (except for outer regions in which ejection by massive

914: embryos can be efficient).  Since $t \sim \tau_{\rm dep}$ is a

915: critical stage for the build-up and retention of the cores and the

916: onset of gas accretion onto the cores, we are particularly concerned

917: with the residual $\Sigma_d$--distribution at this stage.  

918: The asymptotic $\Sigma_d$--$a$

919: distribution is determined by $C_1, f_{g,0}$ and $\tau_{\rm dep}$.

920: In this subsection, we show the simulation results with $f_{g,0} = f_{d,0}$,

921: $q_d = 1.5$ and $q_g = 1.0$ at $t = 0$.  

922: The results with other reasonable values of $f_{d,0}/f_{g,0}$, $q_d$

923: and $q_g$ are qualitatively similar.  With these initial conditions,

924: we can compute the emergence of cores during the epoch of gas

925: depletion.

926: 

927: Figure \ref{fig:sigma_type1_mig1_fg3}a shows $\Sigma_d$ at $t = 10^5,

928: 10^6$, and $10^7$ years (dashed, dotted, and solid lines) with $C_1 =

929: 1$ and $f_{g,0} = 3$.  These results correspond to surviving

930: protoplanets at $t \sim \tau_{\rm dep}$ for $\tau_{\rm dep} = 10^5,

931: 10^6$, and $10^7$ years, although depletion of $f_g$ on time scales

932: $\tau_{\rm dep}$ is not taken into account here (in the Monte Carlo

933: simulations in section 3.2, the exponential decay,

934: eq.~(\ref{eq:gas_exp_decay}), is considered).  We generated 1000 semi

935: major axes with a log-uniform distribution in the range of 0.05--50 AU

936: and simulated the growth and orbital migration of cores there.  

937: In these simulations, gas accretion onto cores is neglected

938: in order to clearly see the effects of type-I migration.

939: The dynamical interactions among the cores is also neglected.  

940: With N-body simulations, \citet{Kominami05} showed that the

941: dynamical interactions with other cores and planetesimals do not

942: change the cores' type-I migration speed.  The results presented in

943: this panel confirm the finding of N-body simulation by

944: \citet{Kominami06}. They also show that $\Sigma_d$ is depleted in an

945: inside-out manner.  The clearing of planetesimals, along the paths of

946: previous generations of cores, limits the growth of the migrating

947: cores and the gravitational interactions between them.  In order to

948: take into account the effect of dynamical clearing in the Monte Carlo

949: simulations (to be presented in next subsection), we terminate the

950: growth of cores after their semimajor axes have decreased below

951: $a_{\rm dep,mig}$ given by eq.~(\ref{eq:a_depletion}).

952: 

953: Figure \ref{fig:sigma_type1_mig1_fg3}d shows that the total mass of

954: the remaining population of cores is $\sim 0.1M_{\oplus}$.  For these

955: retained cores, $\tau_{\rm mig1} > 3 \tau_{\rm c,acc}$ 

956: ($\tau_{\rm c,acc} = \dot{M}_{\rm c}/M_{\rm c}$), 

957: where a factor 3 reflects an actual

958: timescale to reach $M_{\rm c}$ because $\tau_{\rm c,acc} \propto

959: M_{\rm c}^{1/3}$.  From eqs.~(\ref{eq:tau_c_acc2}) and

960: (\ref{eq:tau_mig12}), the maximum mass of remaining cores ($M_{\rm

961: c,max}$) is given by

962: \begin{equation}

963: M_{\rm c,max} \simeq

964: 0.21 C_1^{-3/4} 

965: \left(\frac{f_{g,0}}{3}\right)^{3/10}

966: \left(\frac{\eta_{\rm ice} f_{d,0}}{f_{g,0}}\right)^{3/4}

967: \left( \frac{a}{1{\rm AU}} \right)^{-9/8}

968: \left(\frac{M_\ast}{M_{\odot}} \right)^{5/4} M_{\oplus}.

969: \label{eq:m_c_max}

970: \end{equation}

971: The above expression with $f_{g,0}=f_{d,0}$ reproduces the result at

972: $\ga a_{\rm dep,mig} \sim 1$AU in Figure~

973: \ref{fig:sigma_type1_mig1_fg3}d as well as the weak dependence of

974: $M_{\rm c,max}$ on $f_{g,0}$ and $C_1$

975: (Figures~\ref{fig:sigma_type1_mig1_fg30}d and \ref

976: {fig:sigma_type1_mig01_fg3}d).  In order to further examine the

977: $f_{g,0}$ and $C_1$ dependences, we carry out a set of models with

978: $f_{g,0} = 30$ ($C_1 = 1$) and $C_1 = 0.1$ ($f_{g,0} = 3$) in Figures

979: \ref{fig:sigma_type1_mig1_fg30} and \ref{fig:sigma_type1_mig01_fg3}.

980: In the former case, the disks is marginally self-gravitating and

981: contain a significant fraction of the central stars' mass.  The

982: emergence of relatively massive cores is more sensitively determined

983: by $C_1$, the inefficiency of type-I migration.  A large amount of

984: solid materials in the disk does not efficiently promote the formation

985: of massive cores, because the associated dense gas increases type-I

986: migration speed, as already pointed out by \citet{Kominami06}.

987: 

988: In the inner regions at $a \la 0.3$AU, the cores' accretion proceeds

989: on very short time scales and they reach their isolation mass near

990: their birth place (Figure \ref{fig:sigma_type1_mig1_fg3}c).

991: Thereafter, most of these cores migrate into their host stars within

992: $10^5$ years.  Consequently, the local $\Sigma_d$ is essentially

993: depleted by the formation and migration of the very first generation

994: cores (panel c).  This domain is determined by

995: the condition $M_{\rm c,max} \ga M_{\rm c,iso}$.  From

996: eqs.~(\ref{eq:m_iso0}) and (\ref{eq:m_c_max}), this condition implies

997: \begin{equation}

998: a \la a_{\rm iso,mig} \simeq 0.45 C_1^{-2/5} 

999: \left(\frac{f_{g,0}}{3}\right)^{-2/3}

1000: \left(\frac{\eta_{\rm ice} f_{d,0}}{f_{g,0}}\right)^{-2/5}

1001: \left(\frac{M_\ast}{M_{\odot}} \right)^{14/15}{\rm AU}.

1002: \label{eq:a_1st_gene}

1003: \end{equation}

1004: This expression approximately reproduces the result in

1005: Figure~\ref{fig:sigma_type1_mig1_fg3} and it accounts for the

1006: dependence on $C_1$ and $f_{g,0}$ (see

1007: Figures~\ref{fig:sigma_type1_mig1_fg30}c and

1008: \ref{fig:sigma_type1_mig01_fg3}c).  The location of $a_{\rm iso,mig}$

1009: does not vary with the slope of the surface density distribution

1010: because both competing processes are determined by local properties of

1011: the disk.

1012: 

1013: In the limit of efficient type-I migration (with $C_1=1$), most of the

1014: cores undergo orbital decay before they attain their isolation mass in

1015: the intermediate regions (for Figures~\ref{fig:sigma_type1_mig1_fg3},

1016: 0.3AU $\la a \la 10$AU).  Consequently a significant fraction of

1017: $\Sigma_d$ remains to promote the formation of subsequent-generation

1018: cores.  The results of our simulations (Figure

1019: \ref{fig:sigma_type1_mig1_fg3}c) show that many generations of cores

1020: may emerge at the same disk location.  This repeated formation and

1021: self-destruction process is even more efficient in massive disks where

1022: $f_{g,0} \gg 1$ (see Figure \ref{fig:sigma_type1_mig1_fg30}c).  

1023: 

1024: In the inner region, surface density is significantly depleted by the

1025: N generations of core formation and disruption where

1026: \begin{equation}

1027: N_{\rm gene,1} \simeq \frac{M_{\rm c,iso}}{M_{\rm c,max}}

1028: \simeq 4.1 C_1^{3/4} 

1029: \left(\frac{f_{g,0}}{3}\right)^{6/5}

1030: \left(\frac{\eta_{\rm ice} f_{d,0}}{f_{g,0}}\right)^{3/4}

1031: \left(\frac{a}{1{\rm AU}}\right)^{15/8}

1032: \left(\frac{M_\ast}{M_{\odot}} \right)^{-7/4}.

1033: \label{eq:N_gene1}

1034: \end{equation}

1035: As the planetesimal building blocks become depleted, the cores formed

1036: at later epochs have masses smaller than $M_{\rm c,max}$.  Therefore

1037: the above expression for $N_{\rm gene,1}$ slightly under estimates the

1038: number of populations of cores which may emerge.  

1039: 

1040: At large disk radii, the number of generation is limited by disk

1041: depletion time.  In this region,

1042: \begin{equation}

1043: N_{\rm gene,2} = \frac{t}{\tau_{\rm c,mig}(M_{\rm c,max})}

1044: \simeq 3.9 C_1^{1/4} 

1045: \left(\frac{f_{g,0}}{3}\right)^{13/10}

1046: \left(\frac{\eta_{\rm ice} f_{d,0}}{f_{g,0}}\right)^{3/4}

1047: \left(\frac{a}{1{\rm AU}}\right)^{-17/8}

1048: \left(\frac{t}{10^6{\rm years}}\right)

1049: \left(\frac{M_\ast}{M_{\odot}} \right)^{-1/4}.

1050: \label{eq:N_gene2}

1051: \end{equation}

1052: This estimate completely reproduces the results in Figures~

1053: \ref{fig:sigma_type1_mig1_fg3}c, \ref{fig:sigma_type1_mig1_fg30}c, and

1054: \ref{fig:sigma_type1_mig01_fg3}c.  The actual number of generation is

1055: given by min$(N_{\rm gene,1}, N_{\rm gene,2})$.  Significant depletion

1056: of the original inventory of heavy elements occurs in the limit

1057: $N_{\rm gene,1} \la N_{\rm gene,2}$, that is, within the location

1058: \begin{equation}

1059: a \la a_{\rm dep,mig} 

1060: \simeq C_1^{-1/8} 

1061: \left(\frac{f_{g,0}}{3}\right)^{1/40}

1062: \left(\frac{t}{10^6{\rm years}}\right)^{1/4}

1063: \left(\frac{M_\ast}{M_{\odot}} \right)^{3/8}{\rm AU}.

1064: \label{eq:a_depletion}

1065: \end{equation}

1066: This boundary of disruption zone is in excellent agreement with the

1067: critical location within which $\Sigma_d$ (equivalently, from $f_{d}$)

1068: has reduced from its initial values by an order of magnitude.  Note

1069: that the dependences of $a_{\rm dep,mig}$ on $C_1$ and $f_{g,0}$ are

1070: very weak.  As long as $N_{\rm gene,2} > 1$, the disruption zone is

1071: confined to $a \sim (t/10^6{\rm years})^{1/4}$ AU, independent of disk

1072: surface density and migration speed $C_1$ (in the limit of small

1073: $C_1$, $N_{\rm gene,2} < 1$ and $N_{\rm gene,1} < 1$,

1074: so that no depletion occurs).

1075: 

1076: These results indicate that, during the early epoch of disk evolution,

1077: cores form and migrate repeatedly to clear out the residual

1078: planetesimals. This self-regulated process provides an avenue for the

1079: host stars to acquire most of the heavy elements retained by the

1080: planetesimals. In large, well-mixed molecular clouds where star

1081: clusters form, this self-regulated mechanism would lead to stellar

1082: metallicity homogeneity (\S 4.1). The inner disk region contains cores

1083: which have started their type-I migration but not yet reached to the

1084: disks' inner edge. The average value of $\Sigma_d$ at these locations,

1085: including the contribution of these migrating cores, is two orders of

1086: magnitude smaller than its initial value.

1087: 

1088: Eventually, the disk gas is so severely depleted that relatively

1089: massive cores no longer undergo significant amount of type-I

1090: migration. As any given disk radius, the condition for retaining 90\%

1091: of the initial solid surface density is $N_{\rm gene,2} \simeq

1092: 0.1N_{\rm gene,1}$.  We find, from eqs.~(\ref{eq:N_gene1}),

1093: (\ref{eq:N_gene2}), and (\ref{eq:a_depletion}), that this condition is

1094: satisfied in regions with

1095: \begin{equation}

1096: a \ga a_{\rm surv,mig} 

1097: = 10^{1/4} a_{\rm dep,mig} \simeq 2 a_{\rm dep,mig}.

1098: \label{eq:a_survival}

1099: \end{equation}

1100: Type-I migration leads to a transition in the $\Sigma_d$ distribution

1101: at this orbital radius.

1102: 

1103: Inside the ice line, type-I migration limits the mass of individual

1104: surviving cores.  But these cores can coalesce through giant impacts

1105: during and after the severe depletion of the disk gas

1106: \citep{Kominami02, IL04a}.  Provided the total mass of residual

1107: planetesimals and cores is $\sim O(1)M_{\oplus}$ at 1AU $\la a \la$ a

1108: few AU, Earth-mass terrestrial planets may form near 1AU.  Previous

1109: simulations \citep{Chambers98, Agnor99, Kominami02, Raymond04} show

1110: that the most massive terrestrial planets tend to form in inner

1111: regions of the computational domain where the isolated cores are

1112: initially placed. In these simulations, strong gravitational

1113: scattering process can inject planetesimals close to the host stars to

1114: form smaller planets with relatively close-in orbits.

1115: 

1116: In general, type-I migration leads to clearing of planetesimals close

1117: to their host stars and sets the inner edge of the cores' population

1118: at $\sim 1$AU.  The lack of planets inside the Mercury's orbit in our

1119: Solar system might also be attributed to this result

1120: \citep{Kominami06}.  In addition, the self-regulated clearing process

1121: also leads to $f_d \sim O(1)$ near 1AU for wide variety of initial

1122: conditions ($f_{d,0}$).  This residual distribution of heavy elements at

1123: $\tau_{\rm dep}$ ensures the formation of Mars to Earth-sized

1124: terrestrial planets in habitable zones (see \S 3.2).  Even in the

1125: limit that the disks' initial $\Sigma_d$ distribution is much larger

1126: than that of MMSN, the reduction of $\Sigma_d$ at $\la$ a few AU

1127: inhibits {\it in situ} formation of gas giants interior to the ice

1128: line.  In contrast to the results in Papers I-III, the inclusion of

1129: type-I migration suppresses the rapid and prolific formation of hot

1130: Jupiter which in term facilitates the formation and retention

1131: probability of habitable terrestrial planets.

1132: 

1133: We also simulated several models with $q_d = 2$ which correspond to a

1134: steeply declining initial surface density distribution. Such a steep

1135: gradient of $\Sigma_d$ could be produced by the inward migration of dust

1136: due to the hydrodynamic drag on them by the disk gas 

1137: \citep{Stepinski97, Kornet}.  In this model, a significant amount of

1138: solid mass is contained in inner disk regions where cores quickly

1139: form and undergo type-I migration.  Nearly all the initial mass of

1140: solid components in disks is accreted by the host stars through the

1141: self-regulation by type-I migration.  The results of this model is

1142: consistent with the discovery of metallicity homogeneity among the

1143: stars in young open clusters (see \S 4.1).

1144: 

1145: The above discussions indicate that the formation of Mars to

1146: Earth-sized habitable planets depends only weakly on type-I migration

1147: speed.  However, it is critical for the formation of cores of gas

1148: giant planets, because $M_{\rm c,max} \propto C_1^{-3/4}$

1149: (eq.~[\ref{eq:m_c_max}]).  For giant planets to actually form,

1150: sufficiently massive cores must be able to accrete gas on time scales

1151: at least shorter than a few folding time of $\tau_{\rm dep}$.  For

1152: $\tau_{\rm dep} \sim 10^6$--$10^7$ years, we deduce, from

1153: eq.~(\ref{eq:mgsdot}) with $k1 =9$ and $k2=3$, that gas giant

1154: formation is possible only for $M_{\rm c} \ga$ a few $M_{\oplus}$.

1155: But, type-I migration suppresses the emergence of such massive cores

1156: in disk regions with relatively large $\Sigma_g$.  The results with

1157: $C_1=1$ in Figures \ref{fig:sigma_type1_mig1_fg3}d and

1158: \ref{fig:sigma_type1_mig1_fg30}d show that the asymptotic core masses

1159: are generally much smaller than that needed to launch efficient gas

1160: accretion even though there is little decline in the magnitude of

1161: $\Sigma_d$.

1162: 

1163: In the case of $C_1=0.1$ (Figure \ref{fig:sigma_type1_mig01_fg3}d),

1164: the cores formed at a few AU originally have $\sim M_{\oplus}$.  

1165: Provided the planetesimals along their migration path

1166: are not captured onto their mean motion resonances, these cores can

1167: grow up to a few $M_{\oplus}$ through accretion during migration.  The

1168: core mass is marginal for rapid gas accretion. As shown in

1169: eq.~(\ref{eq:m_c_max}), in metal-rich disk regions (where $f_{d,0}/

1170: f_{g,0} > 1$), more massive cores can be retained with the same amount

1171: of solid materials.  In this expression, the effects of gas depletion

1172: has not been taken into account. But, this process is included in the

1173: numerical simulations in \S3.2 and it also enhances the mass of

1174: retainable cores, especially those which emerge during the advanced

1175: stages of gas depletion.  These results suggest that gas giants can be

1176: formed for $C_1 \la 0.1$.  (\citet{Alibert05} derived a similar

1177: condition for formation of Jupiter and Saturn.)  In the next

1178: subsection, we will show that $C_1 \la 0.1$ reproduces mass and semi

1179: major axis distribution of extrasolar gas giants comparable to those

1180: observed.

1181:  

1182: \subsection{Mass - semimajor axis distributions}

1183: 

1184: \subsubsection{Dependence on type-I migration speed}

1185: 

1186: We now consider the signature of type-I migration on the $M_p-a$

1187: distribution for a population of emerging planets.  In the Monte Carlo

1188: simulations, we first generate a 1,000 set of disks with various

1189: $f_{d,0}$ (the initial value of $f_d$) and $\tau_{\rm dep}$.  We adopt

1190: the same prescriptions for the distributions of $f_{d,0}$ and

1191: $f_{g,0}$ as those in Paper II.  For the gaseous component, we assume

1192: $f_{g,0}$ has a log normal distribution which is centered on the value

1193: of $f_{g,0} = 1$ with a dispersion of 1 ($\delta \log_{10} f_{g,0} =

1194: 1.0$) and upper cut-off at $f_{g,0} = 30$, independent of the stellar

1195: metallicity.  For the heavy elements, we choose $f_{d,0}= 10^{{\rm

1196: [Fe/H]}_d} f_{g,0}$, where [Fe/H]$_d$ is metallicity of the disk.  We

1197: assume these disks have the same metallicity as their host stars.

1198: 

1199: Following our previous papers, we also assume $\tau_{\rm dep}$ has

1200: uniform log distributions in the ranges of $10^6$--$10^7$ yrs.  For

1201: each disk, 15 $a$'s of the protoplanetary seeds are selected from a

1202: uniform log distribution in the ranges of 0.05--$50$AU, assuming that

1203: averaged orbital separation between planets is 0.2 in log scale (the

1204: averaged ratio of semimajor axes of adjacent planets is $\simeq 1.6$).

1205: This procedure is the same as that adopted in Paper II.  Constant

1206: spacing in the log corresponds to the spacing between the cores is

1207: proportional to $a$, which is the simplest choice and a natural

1208: outcome of dynamical isolation at the end of the oligarchic growth.

1209: The log constant spacing for planets and cores with similar masses

1210: also maximizes dynamical stability.  In the present paper, we neglect

1211: dynamical interaction between planets (this issue will be addressed in

1212: future papers) and the growth of individual planets are integrated

1213: independently.  Although the choice of averaged orbital separation is

1214: arbitrarily, it would not change overall results with regard to the

1215: effects of type-I migration.

1216: 

1217: In all the simulations presented here, $\alpha = 10^{-3}$ 

1218: and $M_{\ast} = 1M_{\odot}$ are assumed.

1219: Since the on-going radial velocity surveys are focusing

1220: on relatively metal-rich stars, we present the results with [Fe/H] =

1221: 0.1 in most cases.  The dependence on [Fe/H] is shown in

1222: Figures~\ref{fig:metal_dep}.  In order to directly compare with

1223: observations, we determine the fraction of stars with currently

1224: detectable planets as $\eta_J$.  In the determination of $\eta_J$, we

1225: assume the detection limit is set by the magnitude of radial velocity

1226: ($v_r > 10$m/s) and orbital periods ($T_K < 4$ years).

1227: According to the following uncertainty, we exclude close-in planets 

1228: with $a < 0.05$AU in the evaluation of $\eta_J$.

1229: 

1230: We artificially terminate type-I and II migration near disk inner edge

1231: at a 2 day period ($\sim 0.03$AU for $M_{\ast} = 1 M_{\odot}$) 

1232: in a similar way to Papers I-III.  

1233: We have not specified a survival criterion for the

1234: close-in planets because we do not have adequate knowledge about

1235: planets' migration and their interaction with their host stars near

1236: inner edge of their nascent disks.  Hence, we record all the planets

1237: which have migrated to the vicinity of their host stars.  In reality,

1238: a large fraction of the giant planets migrated to small disk radii may

1239: either be consumed \citep[e.g.,][]{Sandquist98} or tidally disrupted

1240: \citep[e.g.,][]{Trilling98, Gu03} by their host stars.  Cores that

1241: have migrated to inner edge may also coagulate and form super-earths

1242: \citep[e.g.,][]{Terquem07}.  We also neglect such core coagulation

1243: near inner edge.

1244: 

1245: In a set of fiducial models, we adopt $M_{\ast} = 1 M_{\odot}$, [Fe/H]

1246: = 0.1 and $(k1,k2)=(9,3)$ for the gas giants' growth rate

1247: (eq.~[\ref{eq:tau_KH}]).  The analytic deductions in the previous

1248: subsection suggest that relatively massive cores can be retained to

1249: form gas giants provided $C_1 \la 0.1$.  Figures \ref{fig:ma_C1} show

1250: the predicted $M_p$--$a$ distributions for $C_1 = 0$ (panel b), 

1251: $C_1 = 0.01$ (panel c), $C_1 = 0.03$ (panel d), $C_1 = 0.1$ (panel e), 

1252: and $C_1 = 0.3$ (panel f).  In order to directly compare

1253: the theoretical predictions 

1254: with the observed data, we plot (in panel a) $M_p$ which is a factor

1255: of 1.27 times the values of $M_p \sin i$ determined from radial

1256: velocity measurements (http://exoplanet.eu/).  

1257: This correction factor corresponds to the average

1258: value $1/ \langle \sin i \rangle = 4/\pi$ for a sample of planetary

1259: systems with randomly oriented orbital planes.  To compare with 

1260: the theoretical results with $M_{\ast} = 1M_{\odot}$, 

1261: we plot only the data of planets around stars 

1262: with $M_{\ast} = 0.8$--$1.2 M_{\odot}$ observed by radial velocity surveys.

1263: 

1264: All results show "planet desert," 

1265: which is a lack of intermediate-mass ($M_p \sim 10$--$100M_{\odot}$)

1266: planets at $\la$ a few AU.

1267: However, formation probability of gas giants dramatically

1268: changes with $C_1$. The fraction of stars with gas giants 

1269: $\eta_J$ changes from $22.9\%$ for the model with $C_1 = 0$ 

1270: to $0.2\%$ for $C_1 = 0.3$ (see Table \ref{tab:1}).  

1271: In the observed data, $\eta_J \sim 5$--8$\%$ around 

1272: [Fe/H]$\sim 0.1$ (Figure~\ref{fig:metal_dep}).

1273: 

1274: In contrast, the distributions of retained gas giants are similar 

1275: to each other except for relatively large $C_1$.

1276: The theoretical predictions are also consistent with the

1277: observed distribution in panel a.  We carry out a KS test for

1278: statistical similarity between the simulated models and the observed

1279: data for the parameter domain of 0.1AU $< a < 2.5$ AU and 

1280: $M_{\rm p} > 100M_{\oplus}$.  

1281: This range corresponds to a maximum rectangular

1282: region in which planets are detectable by radial surveys with

1283: precision $v_r > 10$m/s and duration $<4$ years. Since we have not

1284: imposed any criterion for determining the survival probability of

1285: short-period planets, the predicted population of planets with 

1286: $a < 0.1$AU are excluded in the quantitative statistical significance test.

1287: Except for the model with $C_1=0.3$ (panel f),

1288: these models are statistically

1289: similar to the observed data within 

1290: a significant level $Q_{\rm KS} \ga 0.3$ for both semimajor axis

1291: and mass cumulative distribution functions.  

1292: In particular, the model with $C_1=0.03$ (panel d) shows

1293: an excellent agreement with $Q_{\rm KS} = 0.86$ for 

1294: the mass function.  

1295: 

1296: In models with $C_1=0.3$, only the low-mass cores can survive type-I

1297: migration.  The envelope contraction time scales for these low-mass

1298: cores are generally much longer (eq.~[\ref{eq:tau_KH}]) than the gas

1299: depletion time scales.  Consequently, $\eta_J$ is very small ($<1\%$) 

1300: for the simulated model with $C_1 = 0.3$. Since type-II

1301: migration occurs after planets have acquired a mass which is adequate

1302: to open up gaps, the close-in planets with $\ga 100 M_{\oplus}$ are

1303: rare for models with $C_1=0.3$, in contrast to the models that type-I

1304: migration is neglected or sufficiently reduced (panels b to e). 

1305: 

1306: Assuming the survival fraction of close-in planets is independent

1307: their $M_p$, the magnitude of $C_1$ can be calibrated from the observed

1308: mass distribution close-in planets.  In Figure~\ref{fig:close-in}a, we

1309: plot the mass function for all the planets which are halted

1310: artificially at $a=0.03$ AU.  In this panel, we neglect any further

1311: evolution including both disruption and collisions.  We also consider

1312: an alternative limit that after the gas depletion, the

1313: multiple-generation of cores which migrated to the proximity of any

1314: given star are able to merge into a single terrestrial planet with a

1315: mass $M_{\rm mer}$ (Fig.~\ref{fig:close-in}b).  These results suggest

1316: the potential findings of many hot earths, including Neptune-mass

1317: planets, with either transit or radial velocity surveys.

1318: 

1319: The above results clearly highlights the competing effect of type-I

1320: migration and gas accretion.  In these models, we approximate the gas

1321: accretion process with $(k1, k2) = (9,3)$.  The early models of

1322: proto-gas-giant planet formation \citep{Pollack96} yield slower growth

1323: rates and they are better fitted with $(k1, k2) = (10,3.5)$.  But,

1324: recent revisions \citep{Ikoma00, Hubickyj05} of these models indicate

1325: that the protogas giants' growth rates can be significantly enhanced

1326: by the opacity reduction associated with grain growth or boundary

1327: conditions at different regions of the disk \citep{Ikoma01}.  The

1328: Kelvin-Helmholtz contraction timescale may also be reduced by

1329: turbulent heat transport in the outer envelope of the protoplanets.

1330: In view of these uncertainties, we also simulated models with $k1 = 8$

1331: and 10, with $k2 = 3$ for all cases.  The predicted $\eta_J$ are

1332: listed in Table \ref{tab:1}.  For models with $C_1 = 0.03$--0.1 in which

1333: type-I migration marginally suppresses the formation of gas giants,

1334: the magnitude of $\eta_J$ depends sensitively on the minimum mass for

1335: the onset of dynamical gas accretion (which is represented by $k1$).

1336: The results in Table \ref{tab:1} indicate that a smaller value of $k1

1337: (=8)$ can lead to a significant increase in $\eta_J$ because smaller

1338: mass cores can initiate the runaway gas accretion within $\tau_{\rm

1339: dep} = 10^6$--$10^7$ years (eq.~[\ref{eq:tau_KH}]).

1340: 

1341: \subsubsection{Preservation of terrestrial planets}

1342: 

1343: Although type-I migration 

1344: is an effective mechanism to suppress the emergence of

1345: gas giant planets, earth-mass planets can form at $\sim 1$AU even in

1346: the limit of $C_1 \sim 1$.  (But most terrestrial planets with $a \la

1347: 0.3$AU would be eliminated by a full-strength type-I migration.)  

1348: Not all of these planets formed interior

1349: to the ice line may be retained.  A fraction of these planets may be

1350: trapped onto the mean motion resonance of migrating gas giant planets

1351: and be forced to migrate with them \citep{Zhou05}. In the 

1352: relatively unlikely event of inefficient eccentricity 

1353: damping and very fast type-II migration, 

1354: some of the embryos along the path of the migrating gas

1355: giants may also be scattered to large radial distances and form later

1356: generation terrestrial planets \citep{Raymond06}.

1357: 

1358: The survival of terrestrial planets depends on their post-formation

1359: encounter probability with migrating giant planets.  In the absence of any

1360: type-I migration, this probability is modest. But the inclusion of a

1361: small amount of type-I migration significantly reduce the fraction of

1362: stars with massive close-in gas giants because the retention of the

1363: progenitor cores becomes possible only at the late stages of disk

1364: evolution when the magnitude of $\Sigma_g$ is reduced.  With a limited

1365: supply of the residual disk gas, the growth of gas giants and their

1366: type-II migration are suppressed.  

1367: 

1368: We find that repeated migration of gas giants is less common in models

1369: with $C_1 \ga 0.01$ than those with $C_1 = 0$. The low type-II

1370: migration probability reduced the need for efficient disruption of

1371: largely accumulated close-in planets (see Paper II). It also ensures

1372: that most of the terrestrial planets formed in the habitable zones are

1373: not removed by the migrating gas giants.  Note that type-I migration

1374: also inhibits {\it in situ} formation of gas giants near 1AU (see \S

1375: 3.1).  Thus, a small amount of type-I migration facilitates formation

1376: and retention of terrestrial planets in habitable zones in extrasolar

1377: planetary systems, rather than inhibits them.

1378: 

1379: \subsubsection{Metallicity dependence}

1380: 

1381: We also study the dependence on other parameters, metallicity

1382: ([Fe/H]).  As fiducial models, we set $C_1=0.03$ and $k1 =9$ to compare

1383: the metallicity dependence.  The most massive cores which can form and

1384: be retained prior to gas depletion have masses $M_{\rm c,max}$ given

1385: by eq.~(\ref{eq:m_c_max}).  Smaller $f_{g,0}$, larger $f_{d,0}$, or

1386: smaller $C_1$ increases $M_{\rm c,max}$ and would enhance formation of

1387: gas giants.  The models in Figures \ref{fig:ma_C1} have already

1388: illustrated the dependence on $C_1$.

1389: 

1390: In Paper II, we showed that relatively large values of [Fe/H] (or

1391: equivalent $f_{d,0}/f_{g,0}$) enhance the growth rates and increase

1392: the isolation masses for the cores, for the disks with the same $f_{g,0}$.  

1393: Fast emergence of massive cores

1394: can lead to the rapid onset of gas accretion.  Based on that model, we

1395: found $\eta_J$ increases with [Fe/H], and the simulated of $\eta_J$

1396: dependence on the [Fe/H] is qualitatively consistent with observed

1397: data \citep{Fischer05}.

1398: 

1399: Here we study the effect of type-I migration on the $\eta_J$--[Fe/H]

1400: correlation.  In Figures~\ref{fig:ma_C1}, [Fe/H] = 0.1 is assumed.  We

1401: carried out similar simulations with various [Fe/H] and the simulated

1402: $\eta_J$--[Fe/H] relation is plotted in Figure~{\ref{fig:metal_dep}).

1403: A comparison between the results here and those in Paper II shows that

1404: type-I migration enhances the $\eta_J$--[Fe/H] correlation. In

1405: metal-poor disks, $f_{d,0}/f_{g,0} < 1$ and $\eta_J$ is significantly

1406: reduced by type-I migration.  But in more metal-rich disks, relatively

1407: massive cores can be retained before the disk gas is severely

1408: depleted.  The resultant steep dependence is in a better agreement

1409: with the observed data (open circles in the plot) than in the

1410: dependence without type-I migration (filled circles with dashed line),

1411: although different choice of assumed averaged orbital separation 

1412: may change $\eta_J$ slightly.

1413: 

1414: Formed planetary systems are affected by stellar mass $M_*$ 

1415: as well as metallicity. 

1416: The dependence on $M_*$ was studied in Paper III,

1417: using a simple model without the effects of type-I migration.

1418: In the paper, we predicted that gas giants are much more rare 

1419: around M type dwarfs than around FGK dwarfs while

1420: super-Earths are abundant around M type dwarfs, 

1421: which is consistent with radial velocity survey 

1422: and microlensing survey \citep{Beaulieu06}.

1423: Adding the type-I migration to the simple prescription in Paper III, 

1424: we found that the above conclusions do not change.

1425: Around M type dwarfs, super-earths at 1--3AU, 

1426: which are inferred to be abundant by microlensing survey, 

1427: survive type-I migration.

1428: In a separate paper, we will address the details of the $M_*$ dependence

1429: of planetary systems,

1430: taking into account the $M_*$ dependences of many physical quantities.

1431: 

1432: 

1433: \section{Summary and Discussions}

1434: 

1435: In our previous Monte Carlo simulations of planet formation processes

1436: (Papers I, II, and III), we neglected the effects of type-I migration.

1437: In the present paper, we have investigated its effects on formation of

1438: terrestrial planets and cores of gas giants.  We found that type-I

1439: migration provides a self-clearing mechanism for planetesimals in the

1440: terrestrial-planet region. Although the planetesimal disk at $a \la

1441: 1$AU is significantly cleared, the total mass of residual planetary

1442: embryos at regions within a few AU is comparable to the Earth's mass, 

1443: almost independent of the disk and migration parameters. Earth-like planets

1444: can be assembled in habitable zones after the depletion of the disk gas.  

1445: 

1446: But this self-regulated clearing process does prevent giant planets

1447: from forming at $\sim 1$AU even in very massive disks. In general, the

1448: clearing of cores leads to the late formation of gas giants.  When the

1449: surface density of the disk gas is reduced below that of the MMNM,

1450: type-I migration would no longer be able to remove cores which are

1451: sufficiently massive to initiate the onset of rapid gas

1452: accretion. This late-formation tendency also reduces the fraction of

1453: gas giant which undergo extensive type-II migration. Since migrating

1454: gas giants capture and clear cores along their migration paths, type-I

1455: migration also facilitates the retention of Earth-mass planets in the

1456: habitable zones.

1457: 

1458: In the limit that type-I migration operates with an efficiency

1459: comparable to that deduced from the traditional linear torque analysis

1460: ({\it i.e.} with $C_1 = 1$), all cores would be cleared prior to the

1461: gas depletion such that gas giant formation would be effectively

1462: suppressed.  However, the catastrophic peril of type-I migration would

1463: be limited by at least a ten-fold reduction in the type-I migration 

1464: speed ($C_1 \la 0.1$).  In this limit, a substantial fraction of the cores may

1465: survive and gas giants would form and be retained with an efficiency

1466: ($\eta_J$) comparable to that observed.

1467: 

1468: The above discussions indicate that the formation probability of gas

1469: giants is delicately balanced by various competing processes. Since

1470: these processes have comparable efficiency, the fraction of solar type

1471: stars with gas giants appears to be ``threshold'' quantity. Small

1472: variations in the strength of one or more of these effects can

1473: strongly modify the detection probability of extrasolar planets.  For

1474: example, $\eta_J$ is observed to be a rapidly increasing function of

1475: their host stars' metallicity \citep{Fischer05}.  Although we were able

1476: to reproduce this observed trend in Paper II, the simulations

1477: presented here indicate that this effect is enhanced by a small amount

1478: of type-I migration because it is more effective in metal-poor disks.

1479: 

1480: The $M_p$ - $a$ distribution in Figures~\ref{fig:ma_C1} also show its

1481: sensitive dependence on various other model parameters.  The most

1482: noticeable dependence is the relative frequency between gas and ice

1483: giant planets at several AU from their host stars.  These

1484: distributions also predict a modest population of close-in cores, a

1485: fraction of which may survive and be observable as hot earths.

1486: 

1487: \subsection{Metallicity homogeneity of open clusters}

1488: 

1489: In modern paradigm of star formation \citep{Shu87}, most of the stellar

1490: content is processed through protostellar disks.  Gas diffusion in

1491: these disks is regulated by the process of turbulent transport of

1492: angular momentum whereas the flow of heavy elements is determined by

1493: orbital evolution of their main carriers, {\it i.e.} through gas drag

1494: of grains and tidal interaction of planetesimals and cores with the

1495: gas. In general the accretion rates of these two components are

1496: not expected to match with each other.  In fact, the formation of gas

1497: giant planets around $\sim$10\% of the solar type stars and the common

1498: existence of debris disks requires the retention of heavy elements

1499: with masses at least comparable to that of the MMSN.  Yet, stars in

1500: young stellar clusters are chemically homogenous \citep{Wilden02,

1501: Quillen02, Shen05}.  The observationally determined upper limit in the

1502: metallicity dispersion ($< 0.03$--0.04 dex) among the stars in the

1503: Pleiades and IC4665 open clusters implies a total residual heavy

1504: element mass (including the planets) to be less than twice that in

1505: Solar system planets.

1506: 

1507: Clues on the resolution of this paradox can be found in the

1508: protostellar disks.  Recent models of the observed millimeter

1509: continuum SED's suggest that some fraction of classical T Tauri disks

1510: has dust mass $\ga$ 10 times the total heavy-element mass in Solar

1511: system \citep{Hartmann06}.  Their host stars would acquire significant

1512: metallicity dispersion if a large fraction of the heavy elements in

1513: these disks is retained as terrestrial planets and cores while most of

1514: their gas components is accreted by the stars.  Hence, an efficient

1515: process to clear heavy elements is required.

1516: 

1517: Although, due to gas drag, dust grains can undergo inward migration

1518: and be accreted onto their host stars, we here assume that the

1519: formation of planetesimals is a more efficient self-clearing process.

1520: In active protostellar disks where $\Sigma_g$ and $\Sigma_d$ are much

1521: larger than those of MMSN, planetesimals quickly grow into cores which

1522: undergo rapid type-I migration. Through a series of numerical

1523: simulations, we show that most cores formed interior to $a_{\rm

1524: dep,mig} \sim 1$ AU are lost before the disk gas is severely depleted.

1525: On a larger spatial and time scales, this process is also effective in

1526: transporting most of the original heavy elements to their host stars.

1527: Provided their progenitor molecule clouds is thoroughly mixed, this

1528: self-regulated clearing process would ensure stars in young clusters

1529: acquire nearly uniform metallicity.

1530: 

1531: \subsection{Diversity of giant planets around solar-type stars}

1532: 

1533: Among the stars with known gas giants, a large fraction of them show

1534: signs of additional planets. This special multiplicity function is

1535: associated with the formation of gaps around the first-born gas

1536: giants.  Beyond the outer edge of the gap, a positive pressure

1537: gradient leads the gas to attain a local super Keplerian velocity.

1538: Dust particles and planetesimals accumulate in this region and grow

1539: into sufficiently massive cores to initiate the formation of

1540: additional planets\citep{Bryden00}.  It is also possible that the

1541: planet formation is a threshold phenomenon, {\it i.e.} the conditions

1542: needed to form a multiple-planet systems are marginally more stringent

1543: than that for the formation of single gas giants.

1544: 

1545: The simulations presented here consider the formation probability of

1546: individual planets.  Although the impact of type-I migration on the

1547: residual disk and the formation of multiple generation cores have

1548: been taken into account, we have neglected the impact of gas giant

1549: formation on the residual disks.  Nevertheless, this algorithm can be

1550: used to qualitatively describe a threshold scenario for the formation 

1551: of multiple planets.

1552: 

1553: The total formation timescale of gas giants is $\tau_{\rm form} \sim

1554: \tau_{\rm c,acc} + \tau_{\rm KH}$.  In comparison with both type-II

1555: migration and the disk depletion timescales, we find that, provided

1556: $\tau_{\rm c,acc} < \tau_{\rm mig1} $,

1557:    \begin{itemize}

1558:    \item Solar-system-like giant planets would form,

1559:          if $\tau_{\rm form} \sim \tau_{\rm dep}$, because 

1560:          gas giants would not have

1561:          enough time to undergo extensive type-II migration.

1562:    \item Eccentric giants would form, 

1563:          if $\tau_{\rm form} < \tau_{\rm dep}$ and 

1564:          $\tau_{\rm mig2} > \tau_{\rm form}$.  The second condition implies 

1565:          that nearby second-generation gas giants are likely to emerge, 

1566:          either by spontaneously satisfying the formation conditions or due 

1567:          to an induced formation process\citep{Bryden00}, before they 

1568:          undergo any significant type-II migration.  Orbital instability 

1569:          of closely packed multiple-planet systems would excite their 

1570:          eccentricities\citep{Rasio96, Weidenschilling96, LI97, ZLS07}.

1571:    \item Close-in giants would form if $\tau_{\rm form} < \tau_{\rm dep}$, and

1572:          $\tau_{\rm mig2} < \tau_{\rm form}$. 

1573:          They undergo type-II migration before neighboring gas giants form.

1574:    \end{itemize}

1575: In the second case, if $\tau_{\rm mig2} < \tau_{\rm dep}$ and orbital

1576: instability does not occur on a time scale $\sim \tau_{\rm dep}$, the

1577: multiple planets could be locked into mean-motion resonances during

1578: migration.  According to the above conditions, the regions in which

1579: close-in, eccentric, and solar-system-like giant planets are likely to

1580: form are schematically plotted in the $f_{d,0}$--$a_0$ plane ($a_0$ is

1581: the initial semimajor axis) in Figure~\ref{fig:areas}, 

1582: neglecting type-I migration.

1583: Bright gray, dark gray, and modest gray regions correspond to

1584: close-in, eccentric, and solar-system-like giants regions,

1585: respectively.  A more quantitative set of simulations will be 

1586: presented in a subsequent paper.

1587: 

1588: 

1589: \vspace{1em} 

1590: \noindent ACKNOWLEDGMENTS.  We thank for detailed helpful comments by

1591: an anonymous referee.  This work is supported by NASA (NAGS5-11779,

1592: NNG04G-191G, NNG06-GH45G), JPL (1270927), NSF(AST-0507424,

1593: PHY99-0794), and JSPS.

1594: 

1595: \begin{thebibliography}{}

1596: 

1597: \bibitem[Agnor et al.(1999)]{Agnor99}

1598: Agnor, C. B., Canup, R. M. \& Levison, H. F. 1999,

1599: Icarus, 142, 219

1600: 

1601: \bibitem[Alibert et al.(2005)]{Alibert05}

1602: Alibert, Ya., Mousis, O., Mordasini, C. \& Benz, W.

1603: 2005, \apjl, 626, 57

1604: 

1605: \bibitem[Armitage et al.(2002)]{Armitage02}

1606: Armitage, P.~J., Livio, M., Lubow, S.~H., \& Pringle, J.~E.

1607: 2002, MNRAS, 334, 248

1608: 

1609: \bibitem[Armitage (2003)]{Armitage03}

1610: Armitage, P.~J. 2003, ApJL, 582, 47

1611: 

1612: \bibitem[Artymowicz(1993)]{Artymowicz93}

1613: Artymowicz, P. 1993, \apj, 419, 166

1614: 

1615: \bibitem[Balbus \& Hawley(1991)]{Balbus91} 

1616: Balbus, S. A., \& Hawley, J. F., 1991,

1617: \apj, 376, 214

1618: 

1619: \bibitem[Beaulieu et al.(2006)]{Beaulieu06}

1620: Beaulieu, J.-P. et al. 2006,

1621: Nature, 439, 139

1622: 

1623: \bibitem[Beckwith \& Sargent(1996)]{BS96}

1624: Beckwith, S.~V.~W. \& Sargent, A.~I., 1996,

1625: Nature, 383, 139

1626: 

1627: \bibitem[Bodenheimer \& Pollack(1986)]{BP86}

1628: Bodenheimer, P., \& Pollack, J. B. 1986, Icarus, 67, 391

1629: 

1630: \bibitem[Boss(1997)]{Boss97}

1631: Boss, A. P. 1997, Science, 276, 1836

1632: 

1633: \bibitem[Bryden et al. (2000)]{Bryden00}

1634: Bryden, G., Rozyczka, M., Lin, D.N.C., \& Bodenheimer, P.

1635: \apj, 540, 1091

1636: 

1637: \bibitem[Burkert \& Ida(2007)]{Burker07} 

1638: Burkert, A. \& Ida, S. 2007,

1639: \apj, 660, 845

1640: 

1641: \bibitem[Calvet et al.(2000)]{Calvet}

1642: Calvet, N., Hartmann, L. \& Strom, S.~E. 2000,

1643: in Protostars and

1644: Planets IV, ed. V. Mannings, A.~P. Boss and S.~S. Russell 

1645: (Tucson:Univ. of Arizona Press), 377

1646: 

1647: \bibitem[Canup \& Ward(2006)]{CW06}

1648: Canup, R. M. \& Ward, W. R. 2006, Nature, 441, 834

1649: 

1650: \bibitem[Chiang \& Goldreich (1997)]{Chiang97}

1651: Chiang, E. I. \& Goldreich, P. 1997,

1652: \apj, 490, 368

1653: 

1654: \bibitem[Chambers \& Wetherill(1998)]{Chambers98}

1655: Chambers, J. E. \& Wetherill, G. W. 1998,

1656: Icarus, 136, 304

1657: 

1658: \bibitem[Crida \& Morbidelli(2007)]{Crida07}

1659: Crida, A. \& Morbidelli, A. 2007, 

1660: MNRAS, 377, 1324

1661: 

1662: \bibitem[Cumming(2004)]{Cumming}

1663: Cumming, A. 2004, MNRAS, 354, 1165

1664: 

1665: \bibitem[Daisaka et al.(2006)]{Kominami06}

1666: Daisaka, K. J., Tanaka, H. \& Ida, S. 2006, Icarus, 185, 492 

1667: 

1668: \bibitem[D'Angelo et la.(2003)]{D'Angelo03} 

1669: D'Angelo, G., Kley, W. \& Henning, T. 2003,

1670: \apj, 586, 540

1671: 

1672: \bibitem[D'Angelo et al.(2006)]{D'Angelo06}

1673: D'Angelo, G., Lubow, S. H. \& Bate, M. R.

1674: 2006, \apj, 652, 1698 

1675: 

1676: \bibitem[Dobbs-Dixon et al.(2007)]{Dobbs-Dixon07}

1677: Dobbs-Dixon, I., Li, S.-L. \& Lin, D. N. C. 2007,

1678: \apj, 660, 791 

1679: 

1680: \bibitem[Fischer \& Valenti (2005)]{Fischer05}

1681: Fischer, D. A. \& Valenti, J. A. 2005.

1682: \apj, 622, 1102

1683: 

1684: \bibitem[Garaud \& Lin (2007)]{Garaud07}

1685: Garaud, P. \& Lin, D. N. C. 2007,

1686: \apj, 654, 606

1687: 

1688: \bibitem[Gammie (1996)]{Gammie96}

1689: Gammie, C.F. 1996, \apj, 457, 355

1690: 

1691: \bibitem[Goldreich \& Tremaine(1980)]{GT80}

1692: Goldreich, P., \& Tremaine, S. 1980, ApJ, 241, 425

1693: 

1694: \bibitem[Gu et al.(2003)]{Gu03}

1695: Gu, P., Lin, D.~N.~C., \& Bodenheimer, P.~H. 2003,

1696: \apj, 588, 509

1697: 

1698: \bibitem[Guillot \& Hueso(2006)]{Guillot06}

1699: Guillot, T. \& Hueso, R. 2006,

1700: MNRAS, 367, L47

1701: 

1702: \bibitem[Hartmann et al.(1998)]{Hartmann98}

1703: Hartmann, L., Calvet, N., Gullbring, E., \& D'Alessio, P.

1704: 1998, \apj, 495, 385

1705: 

1706: \bibitem[Hartmann et al.(2006)]{Hartmann06}

1707: Hartmann, L., D' Aleesio, P., Calvet, N., \& Muzerolle, J.

1708: 2006, \apj, 652, 472

1709: 

1710: \bibitem[Hayashi(1981)]{Hayashi81}

1711: Hayashi, C. 1981, Prog. Theor. Phys. Suppl., 70, 35

1712: 

1713: \bibitem[Hubickyj et al.(2005)]{Hubickyj05}

1714: Hubickyj, O., Bodenheimer, P., Lissauer, J. J. 2005,

1715: Icarus, 179, 415

1716: 

1717: \bibitem[Ida \& Lin (2004a)]{IL04a}

1718: Ida, S. \& Lin, D. N. C. 2004,

1719: \apj, 604, 388 (Paper I)

1720: 

1721: \bibitem[Ida \& Lin (2004b)]{IL04b}

1722: Ida, S. \& Lin, D. N. C. 2004,

1723: \apj, 616, 567 (Paper II)

1724: 

1725: \bibitem[Ida \& Lin (2005)]{IL05}

1726: Ida, S. \& Lin, D. N. C. 2005,

1727: \apj, 626, 1045 (Paper III)

1728: 

1729: \bibitem[Ikoma et al.(2000)]{Ikoma00}

1730: Ikoma, M., Nakazawa, K. \& Emori, E. 2000, 

1731: \apj, 537, 1013

1732: 

1733: \bibitem[Ikoma et al.(2001)]{Ikoma01}

1734: Ikoma, M., Emori, H. \& Nakazawa, K. 2001, 

1735: \apj, 553, 999

1736: 

1737: \bibitem[Kley \& Dirksen(2006)]{Kley06} 

1738: Kley, W. \& Dirksen, G. 2006

1739: \aa, 447, 369

1740: 

1741: \bibitem[Kokubo \& Ida(1998)]{KI98}

1742: Kokubo, E. \& Ida, S. 1998, Icarus, 131, 171

1743: 

1744: \bibitem[Kokubo \& Ida(2002)]{KI02}

1745: Kokubo, E. \& Ida, S. 2002, \apj, 581, 666

1746: 

1747: \bibitem[Kominami \& Ida(2002)]{Kominami02}

1748: Kominami, J. \& Ida, S. 2002, Icarus, 157, 43

1749: 

1750: \bibitem[Kominami et al.(2005)]{Kominami05}

1751: Kominami, J., Tanaka, H. \& Ida, S. 2005, Icarus, 178, 540

1752: 

1753: \bibitem[Koller \& Li(2004)]{Koller04}

1754: Koller, J., Li, H. 2004, 

1755: in The Search For Other Worlds. AIP Conf. Proc., 713, pp. 63-66

1756: 

1757: \bibitem[Kornet et al.(2001)]{Kornet}

1758: Kornet, K., Stepinski, T. F., Rozyczka, M. 2001, \aa, 378, 180

1759: 

1760: \bibitem[Laughlin et al.(2004)]{Laughlin04}

1761: Laughlin, G., Steinacker, A. \& Adams, F. C.,

1762: \apj, 309, 846

1763: 

1764: \bibitem[Lecar et al.(2006)]{Lecar06}

1765: Lecar, M., Podolak, M., Sasselov, D. \& Chiang, E. 2006,

1766: \apj, 640, 1115

1767: 

1768: \bibitem[Li et al.(2005)]{Li05}

1769: Li, H. et al. 2005

1770: \apj, 624, 1003

1771: 

1772: \bibitem[Lin (1986)]{Lin86}

1773: Lin, D.~N.~C. 1986,

1774: in The solar system: Observations and interpretations. 

1775: ed. M.G. Kivelson \& E. Cliffs (NJ: Prentice-Hall), 28

1776: 

1777: \bibitem[Lin (1995)]{Lin95}

1778: Lin, D. N. C. 1995, 

1779: in Molecular Clouds and Star Formation. ed. C.

1780: Yuan \& J.H. You. Singapore (World Scientific), 261

1781: 

1782: \bibitem[Lin \& Bodenheimer(1982)]{Linbod82}

1783: Lin, D.~N.~C. \& Bodenheimer, P.  1982,

1784: \apj, 262, 768

1785: 

1786: \bibitem[Lin et al.(1996)]{Lin96}

1787: Lin, D.~N.~C., Bodenheimer, P. \& Richardson, D. 1996,

1788: Nature, 380, 606

1789: 

1790: \bibitem[Lin \& Ida(1997)]{LI97}

1791: Lin, D.~N.~C. \& Ida, S. 1997, \apj

1792: 

1793: \bibitem[Lin \& Papaloizou(1985)]{LP85}

1794: Lin, D.~N.~C. \& Papaloizou, J.~C.~B. 1985,

1795: in Protostars and Planets II, 

1796: ed. D.~C. Black \& M.~S.Matthew (Tucson: Univ. of Arizona Press), 981

1797: 

1798: \bibitem[Lin \& Papaloizou(1979)]{LP79}

1799: Lin, D.~N.~C. \& Papaloizou, J.~C.~B. 1979,

1800: MNRAS, 188, 191

1801: 

1802: \bibitem[Lin \& Papaloizou(1986)]{LP86}

1803: Lin, D.~N.~C. \& Papaloizou, J.~C.~B. 1986,

1804: \apj, 309, 846

1805: 

1806: \bibitem[Lin \& Papaloizou(1993)]{LP93}

1807: Lin, D. N. C., \& Papaloizou, J. C. B. 1993, 

1808: in Protostars and

1809: Planets III, ed. E. H. Levy and J. I. Lunine (Tucson:Univ. of Arizona

1810: Press), 749

1811: 

1812: \bibitem[Lynden-Bell \& Pringle(1974)]{Lynden-Bell74}

1813: Lynden-Bell, D. \& Pringle, J. E. 1974, MNRAS, 168, 603

1814: 

1815: \bibitem[McNeil et al.(2005)]{McNeil05}

1816: McNeil, D., Duncan, M. \& Levison, H. F. 2005,

1817: \aj, 130, 2884

1818: 

1819: \bibitem[Marcy et al.(2005)]{Marcy05}

1820: Marcy, G. et al. 2005, Prog. Theor. Phys. Suppl., 158, 24

1821: 

1822: \bibitem[Masset et al.(2006b)]{Masset06b}

1823: Masset, F. S., D'Angelo, G. \& Kley, W., 2006, 

1824: \apj, in press

1825: 

1826: \bibitem[Masset et al.(2006a)]{Masset06a}

1827: Masset, F. S., Morbidelli, A., Crida, A., Ferreira, J., 2006, 

1828: \apj, 642, 478

1829: 

1830: \bibitem[Mayor et al.(2005)]{Mayor05}

1831: Mayor, M., Pont, F. \& Vidal-Madjar, A. 2005, 

1832: Prog. Theor. Phys. Suppl., 158, 43

1833: 

1834: \bibitem[Nelson \& Papaloizou(2004)]{Nelson04}

1835: Nelson, R. P. \& Papaloizou, J. C. B. 2004, 

1836: MNRAS, 350, 849

1837: 

1838: \bibitem[Papaloizou and Terquem(2006)]{Papaloizou06}

1839: Papaloizou, J. C. B., Terquem, C. 2006, Rep. Prog. Phys., 69, 119

1840: 

1841: \bibitem[Pollack et al.(1994)]{Pollack94}

1842: Pollack, J. B., Hollenbach, D., Beckwith, S., Simonelli, D. P.,

1843: Roush, T., \& Fong, W. 1994, \apj, 421, 615

1844: 

1845: \bibitem[Pollack et al.(1996)]{Pollack96}

1846: Pollack, J. B., Hubickyj, O., Bodenheimer, P., Lissauer, J. J.,

1847: Podolak, M., \& Greenzweig, Y. 1996, Icarus, 124, 62

1848: 

1849: \bibitem[Quillen(2002)]{Quillen02}

1850: Quillen, A. C. 2002. \aj, 124, 400

1851: 

1852: \bibitem[Rafikov(2003)]{Rafikov03}

1853: Rafikov, R. R. 2003, \apj, 125, 922

1854: 

1855: \bibitem[Rasio \& Ford(1996)]{Rasio96}

1856: Rasio, F. A. \& Ford, E. B. 1996,

1857: Science, 274, 954

1858: 

1859: \bibitem[Raymond et al.(2004)]{Raymond04}

1860: Raymond, S. N., Quinn, T. \& Lunine, J. I. 2004,

1861: Icarus, 168, 1

1862: 

1863: \bibitem[Raymond et al.(2006)]{Raymond06}

1864: Raymond, S. N., Mandell, A. M. \& Sigurdsson, S. 2006,

1865: Science, 313, 1413

1866: 

1867: \bibitem[Sandquist et al.(1998)]{Sandquist98}

1868: Sandquist, E., Taam, R. E., Lin, D. N. C. \& Burkert, A. 1998,

1869: \apjl, 506, L65

1870: 

1871: \bibitem[Sano et al.(2000)]{Sano00} 

1872: Sano, T., Miyama, S. M., Umebayashi, T., \& Nakano, T., 2000,

1873: \apj, 543, 486

1874: 

1875: \bibitem[Sano et al.(2004)]{Sano04} 

1876: Sano, T., Inutsuka, S., Turner, N. J. \& Stone, J. M. 2004,

1877: Prog. Theor. Phys. Suppl., 155, 409

1878: 

1879: \bibitem[Shen et al.(2005)]{Shen05}

1880: Shen, Z.-X., Jones, B., Lin, D. N. C., Liu, X.-W., Li, S.-L. 2005,

1881: \apj, 635, 608

1882: 

1883: \bibitem[Shu et al.(1987)]{Shu87}

1884: Shu, F.H., Adams, F.C., \& Lizano, S. 1987, ARA\&A, 25, 23 

1885: 

1886: \bibitem[Shu et al.(1993)]{Shu93}

1887: Shu, F. H., Johnstone, D., Hollenbach, D. 1993,

1888: Icarus, 106, 92

1889: 

1890: \bibitem[Shakura \& Sunyaev(1973)]{alpha}

1891: Shakura, N. I. \& Sunyaev, R. A. 1973, A\&A, 24, 337

1892: 

1893: \bibitem[Stepinski \& Valageas(1997)]{Stepinski97}

1894: Stepinski, T. F. \& Valageas, P. 1997, A\&A, 319, 1007

1895: 

1896: \bibitem[Tanaka \& Ida(1999)]{TI99}

1897: Tanaka, H. \& Ida, S. 1999, Icarus, 139, 350

1898: 

1899: \bibitem[Tanaka et al.(2002)]{Tanaka02}

1900: Tanaka, H., Takeuchi, T. \& Ward, W. 2002, \apj, 565, 1257

1901: 

1902: \bibitem[Tanigawa \& Ikoma(2007)]{Tanigawa07} 

1903: Tanigawa, T. \& Ikoma, M. 2007,

1904: \apj, in press

1905: 

1906: \bibitem[Terquem \& Papaloizou(2007)]{Terquem07}

1907: Terquem, C. \& Papaloizou, J. C. B. 2007,

1908: \apj, 654, 1110

1909: 

1910: \bibitem[Thommes \& Murray(2006)]{Thommes06}

1911: Thommes, E. \& Murray, N. 2006, \apj, 644, 1214

1912: 

1913: \bibitem[Trilling et al.(1998)]{Trilling98}

1914: Trilling, D.~E., Benz, W., Guillot, T., Lunine, J. I.,

1915: Hubbard, W. B. \& Burrows, A. 1998, \apj, 500, 428

1916: 

1917: \bibitem[Ward(1986)]{Ward86}

1918: Ward, W. 1986, Icarus, 67, 164

1919: 

1920: \bibitem[Ward(1997)]{Ward97}

1921: ---------. 1997, Icarus, 126, 261

1922: 

1923: \bibitem[Ward \& Hahn(1995)]{Ward_Hahn}

1924: Ward, W. \& Hahn, J. 1995, ApJL, 440, 25

1925: 

1926: \bibitem[Weidenschilling \& Marzari(1996)]{Weidenschilling96}

1927: Weidenschilling, S. J. \& Marzari, F. 1996,

1928: Nature, 384, 619

1929: 

1930: \bibitem[Wilden et al.(2002)]{Wilden02}

1931: Wilden, B.S., Jones, B.F. Lin, D.N.C., \& Soderblom, D.R. 

1932: 2002, \aj, 124, 2799

1933: 

1934: \bibitem[Zhou et al.(2005)]{Zhou05}

1935: Zhou, J.-L., Aarseth, S. J., Lin, D. N. C. \& Nagasawa, M. 2005,

1936: ApJL, 631, L85

1937: 

1938: \bibitem[Zhou et al.(2007)]{ZLS07}

1939: Zhou, J.-L., Lin, D.N.C., \& Sun, Y.S. 2007, \apj, in press.

1940: 

1941: \bibitem[Zhou et al.(2007)]{Zhou07}

1942: Zhou, J.-L. \& Lin, D.N.C. 2007, \apj, in press.

1943: 

1944: \end{thebibliography}{}

1945: 

1946: \clearpage

1947: \begin{table}

1948: \input{tab1.tex}

1949: \label{tab:1}

1950: \end{table}

1951: 

1952: \clearpage

1953: 

1954: \begin{figure}

1955: \plotone{f1}

1956: \caption{

1957: Viscous evolution of a disk with $\alpha = 10^{-3}$.

1958: We set $\Sigma_g = 0$ at $r =0.1$AU and $10^5$AU.

1959: (a) The distributions of $\Sigma_g$ at $t = 0,

1960: 10^3, 10^4, 10^5, 10^6,$ and $10^7$ years (from top to bottom).

1961: (b) Time evolution of $\Sigma_g$ at planet-forming

1962: regions $r = 0.25, 0.67, 1.8, 4.7,$ and 12.5 AU

1963: (from top to bottom).

1964: }

1965: \label{fig:disk_vis_evol}

1966: \end{figure}

1967: 

1968: \begin{figure}

1969: \plotone{f2}

1970: \caption{ The evolution of $\Sigma_d$ due to dynamical sculpting by

1971: type I migration in the case of $C_1 = 1$ and $f_{g,0} = 3$.  

1972: Here we assume constant $f_g$.

1973: (a) The $\Sigma_d$-distributions at $t = 10^5, 10^6$, and $10^7$ years 

1974: are expressed by

1975: dashed, dotted, and solid lines.  The initial distribution is also

1976: shown by dotted lines.  (b) The evolution of $f_d$-distributions.

1977: (The meanings of the lines are the same as panel a.)

1978: (c) The number of generation of protoplanetary cores formed

1979: at each $a$.  (d) Planet distributions at $t = 10^7$ years.  The dashed line

1980: expresses the core isolation mass and the scattering limit.

1981:  }

1982: \label{fig:sigma_type1_mig1_fg3}

1983: \end{figure}

1984: 

1985: \begin{figure}

1986: \plotone{f3}

1987: \caption{ The same as Figure~\ref{fig:sigma_type1_mig1_fg3} except for

1988: $f_{g,0} = 30$ to approximate the early phases of massive disk

1989: evolution.  }

1990: \label{fig:sigma_type1_mig1_fg30}

1991: \end{figure}

1992: 

1993: \begin{figure}

1994: \plotone{f4}

1995: \caption{ The same as Figure~\ref{fig:sigma_type1_mig1_fg3} except for

1996: $C_1 = 0.1$ to approximate weak type I migration.  }

1997: \label{fig:sigma_type1_mig01_fg3}

1998: \end{figure}

1999: 

2000: \begin{figure}

2001: \plotone{f5}

2002: \caption{ Planetary mass and semi major axis distribution.  

2003: Units of the mass ($M_{\rm p}$) and semimajor axis ($a$)

2004: are earth mass ($M_{\oplus} = M_{\rm E}$) and AU.

2005: (a) Observational data of extrasolar planets

2006: (based on data in http://exoplanet.eu/), around stars with 

2007: $M_{\ast} = 0.8$--$1.2 M_{\odot}$ detected by the

2008: radial velocity surveys.

2009: The determined $M_p \sin i$ is

2010: multiplied by $1/ \langle \sin i \rangle = 4/\pi \simeq 1.27$,

2011: assuming random orientation of planetary orbital planes.

2012: (b) The distribution obtained from 

2013: Monte Carlo simulations without taking into account the effect of 

2014: type-I migration, (c) that includes the type-I migration with

2015: $C_1 = 0.01$, (d) $C_1 = 0.03$, (e) $C_1 = 0.1$, and

2016: (f) $C_1 = 0.3$.  The dashed lines express

2017: observational limit with radial-velocity 

2018: measure precision of $v_r = 10$m/s.  

2019: In these models, 

2020: $M_{\ast} = 1 M_{\odot}$,

2021: the magnitude of the metallicity [Fe/H] = 0.1 and 

2022: the contraction time scale parameters in eq.~[\ref{eq:tau_KH}] are 

2023: assumed to be $(k1,k2)=(9,3)$. }

2024: \label{fig:ma_C1}

2025: \end{figure}

2026: 

2027: \begin{figure}

2028: \plotone{f6}

2029: \caption{ 

2030: The metallicity dependence of fraction of stars

2031: with planets which are detectable by radial velocity surveys

2032: with measurement precision $v_r > 10$m/s and coverage periods $< 4$ years.

2033: Open circles with error bars are observed data \citep{Fischer05}.

2034: Dashed, solid, and dotted lines with filled circles

2035: are the theoretically predicted dependences

2036: with $C_1 = 0, 0.03$ and 0.3, respectively.

2037: Other parameters are

2038: $(k1,k2) = (9,3)$, and $M_{\ast} = 1 M_{\odot}$.

2039: }

2040: \label{fig:metal_dep}

2041: \end{figure}

2042: 

2043: \begin{figure}

2044: \plotone{f7}

2045: \caption{ 

2046: The predicted mass function which are halted

2047: artificially at $a=0.03$ AU.

2048: (a) The mass function for all the planets,

2049: neglecting further evolution due to both disruption 

2050: and collisions is neglected.  

2051: Filled circles, open circles, squares, and triangles

2052: represent the results with $C_1 = 0, 0.001, 0.01$, and 0.1,

2053: respectively.

2054: ($M_{\ast} = 1 M_{\odot}$, [Fe/H] = 0.1 and $(k1,k2)=(9,3)$.)

2055: (b) That includes the maximum effect of 

2056: coagulation after the gas is depleted. All the planets 

2057: are coagulated into one planet in each system.   

2058: }

2059: \label{fig:close-in}

2060: \end{figure}

2061: 

2062: \begin{figure}

2063: \plotone{f8}

2064: \caption{The parameter domains (initial disk surface density

2065: $f_{d,0}$ and initial semimajor axis of planets $a_0$), 

2066: in which close-in, eccentric, and

2067: solar-system-like giant planets are likely to form, in the limit of

2068: negligible effects of type-I migration.  Bright gray, dark

2069: gray, and modest gray regions correspond to close-in, eccentric, and

2070: solar-system-like giants regions, respectively.}

2071: \label{fig:areas}

2072: \end{figure}

2073: 

2074: \end{document}

2075: 

2076: