Testing Bimodal Planet Formation

We suggest that the observed break in giant-planet frequency as a function of host metallicity at Z = 0.02 may be a reflection of bimodal planet formation. We search for signatures of this bimodality in the distributions of the planet eccentricities, periods, masses, and multiplicity. However, the low-metallicity sample is at present too small to test for any but the most severe differences in these two putative populations. Subject headings: planetary systems – planetary systems: formation – stars: abundances 1. TEST OF BIMODALITY Fischer, Valenti & Marcy (2003) have demonstrated that the frequency of extra-solar giant planets is a strong function of metallicity Z. Santos, Israelian & Mayor (2004) have confirmed this result and have further shown that there is a sharp break in frequency at Z = 0.02, which can be represented algebraically by, f = 0.025 + 16(Z − 0.02)Θ(Z − 0.02), (1) where Θ is a step function. See their figure 7. We suggest that the two terms of this equation correspond to two different modes of giant-planet formation, the first being metallicity-independent and the second being strongly dependent on metallicity. For example, in its simplest form, the FIG. 1.— Eccentricity e vs. metallicity Z for the 109 planets of the 98 stars with metallicities determined by Santos et al. (2004) with the exception of HD 47536b. If the break in the frequency of planets at Z = 0.02 reflects bimodal planet formation, one might expect this to be reflected in a similar break in the eccentricity distribution at this boundary (dashed line). No such break is visible, but the present statistics would be sufficient to reveal only the most glaring differences. Electronic address: fields,[email protected] disk-instability model of Boss (1995) would not be expected to depend on metallicity. By contrast, the more conventional picture of gas accretion onto rock-ice cores might well be very sensitive to metallicity. If the two mechanisms generated planets with substantially different distributions in eccentricity, period, or mass, then these should be revealed in the observed distributions of these properties as functions of metallicity. That is, the planets with Z < 0.02 should entirely reflect one distribution, while those with Z ≥ 0.02 should predominantly reflect the other. To test this possibility, we make three plots. In Figures 1, 2, and 3, we show the eccentricities e, periods P, and mass functions M sin i of the 109 planets cataloged by Santos et al. (2004) versus the metallicities of their 97 host stars. We exclude HD 47536 as its planet’s properties are not well established. We take planet properties from The Geneva Extrasolar Planet Search Programmes website1. FIG. 2.— Same as Fig. 1, but for period P vs. metallicity Z. 1 http://obswww.unige.ch/∼udry/planet/planet.html