The eccentricity-mass distribution of exoplanets:: signatures of different formation mechanisms?

We examine the distributions of eccentricity and host star metallicity of exoplanets as a function of their mass. Planets with M sin i greater than or similar to 4 M-J have an eccentricity distribution consistent with that of binary stars, while planets with M sin i less than or similar to 4 M-J are less eccentric than binary stars and more massive planets. In addition, host star metallicities decrease with planet mass. The statistical significance of both of these trends is only marginal with the present sample of exoplanets. To account for these trends, we hypothesize that there are two populations of gaseous planets: the low-mass population forms by gas accretion onto a rock-ice core in a circumstellar disk and is more abundant at high metallicities, and the high-mass population forms directly by fragmentation of a pre-stellar cloud. Planets of the first population form in initially circular orbits and grow their eccentricities later, and may have a mass upper limit from the total mass of the disk that can be accreted by the core. The second population may have a mass lower limit resulting from opacity-limited fragmentation. This would roughly divide the two populations in mass, although they would likely overlap over some mass range. If most objects in the second population form before the pre-stellar cloud becomes highly opaque, they would have to be initially located in orbits larger than similar to 30 AU, and would need to migrate to the much smaller orbits in which they are observed. The higher mean orbital eccentricity of the second population might be caused by the larger required intervals of radial migration, and the brown dwarf desert might be due to the inability of high-mass brown dwarfs to migrate inwards sufficiently in radius.