FORMATION OF THE TERRESTRIAL PLANETS

The early phases of formation in the inner solar system were dominated by collisions and short-range dynamical interactions among planetesim als. But the later phases, which account for most of the differences a mong planets, are unsure because the dynamics are mote subtle. Jupiter 's influence became more important, leading to drastic clearing out of the asteroid belt and the stunting of Mars's growth. Further in, the effect of Jupiter- both directly and indirectly, through ejection of m ass in the outer solar system- was probably to speed up the process wi thout greatly affecting the outcome. The great variety in bulk propert ies of the terrestrial bodies indicate a terminal phase of great colli sions, so that the outcome is the result of small-N statistics, Mercur y, 65 percent iron, appears to be a residual core from a high-velocity collision. All planets appear to require a late phase of high energy impacts to erode their atmospheres: including the Earth, to remove CO2 so that its ocean could form by condensation of water. Consistent wit h this model is that the largest collision, about 0.2 Earth masses, wa s into the proto-Earth, although the only property that appears to req uire it is the great lack of iron in the Moon. The other large differe nces between the Earth and Venus, angular momentum (spin plus satellit e) and inert gas abundances, must arise from origin circumstances, but neither require nor forbid the giant impact, Venus's higher ratio of light to heavy inert gases argues for it receiving a large icy impacto r, about 10(-6) Earth masses from far out, requiring some improbable d ynamics to get a low enough approach velocity. Core formation in both planets probably started rather early during accretion. Some geochemic al evidences argue for the Moon coming from the Earth's mantle, but ar e inconclusive. Large scale melting of the mantle by the giant impact would plausibly have led to stratification. But the ''lock-up'' at the end of turbulent mantle convection is a trade-off between rates: crys tallization of constituents of small density difference versus overall freezing. Also, factors such as differences in melting temperatures a nd densities, melt compressibilities, and phase transitions may have h ad homogenizing effects in the subsequent mantle convection.