Host: The Japanese Society for Planetary Sciences: Local Organizing Committee for 2006 Fall Meeting
Remnant planetesimals might have played an important role to reduce enhanced orbital eccentricities of the terrestrial planets formed by giant impacts. However, it is not clarified how much small planetesimals exist in the giant impact phase, because simulations in the runaway growth and giant impact phases have been conducted rather independently. We conduct direct N-body simulations through the accretion of planets beginning from a planetesimal disk. The runway phase is simulated with the parallel tree-code PKDGRAV (Richardson et al. 2000), up to 100,000 yrs with adopting artificially enhanced radii. The giant impact phase is simulated with the hybrid symplectic code Mercury (Chambers 1999) up to 100Myr. Fixing the total mass and angular momentum to the values for the current terrestrial planets, we vary the width of the initial planetesimal disk (0.01-0.5AU) and the number of planetesimals (1000-5000). We find that in most of the cases a planetesimal disk results in three planets with similar size, although we sometimes obtain two planets for an initially narrow disk. There is a weak correlation between the initial width of a planetesimal disk and the final separation of planets. Also, orbital eccentricities of final planets are small enough for the initial width of 0.5AU.