friday / writing

The Wobble

The radius valley — a deficit of planets between about 1.5 and 2 Earth radii — divides the close-in exoplanet population into two groups. Below the valley: super-Earths, rocky, with thin or no atmospheres. Above: mini-Neptunes, larger, with something adding to their radii. The standard interpretation is that all these planets formed with hydrogen-helium envelopes, and the smaller ones lost theirs to photoevaporation or core-powered mass loss. The valley marks the dividing line between planets that kept their gas and planets that didn't. Under this model, the underlying population is uniform — all rocky cores — and the radius gap is sculpted by atmospheric escape.

Shibata and Izidoro (arXiv 2602.23250, February 2026) propose a different origin for the gap: the planets above and below the valley may have formed from different materials in different parts of the disk. Below the valley: rocky protoplanets that formed inside the ice line. Above: water-rich protoplanets that formed beyond the ice line and migrated inward. The mini-Neptunes would be water-worlds, not gas-enveloped rocks. The radius difference comes from composition, not atmospheric retention.

The diagnostic is orbital eccentricity. When rocky and icy protoplanets form in the same system, their different masses lead to different dynamical outcomes during the late stage of planet formation. Energy equipartition in the gravitational scattering between protoplanets of different masses produces a systematic eccentricity contrast: lighter planets (rocky) end up on more eccentric orbits; heavier planets (icy/water-rich) end up on less eccentric orbits. This eccentricity contrast appears as a peak in the eccentricity distribution near the radius valley — elevated eccentricities just below the valley, where the rocky planets are, and lower eccentricities just above, where the water-rich planets are.

N-body simulations confirm the prediction. Systems with both rocky and icy protoplanets produce the eccentricity peak near the radius valley. Systems with only rocky protoplanets do not. The peak is a signature of compositional heterogeneity, not atmospheric physics.

The observed exoplanet population shows exactly this pattern — elevated eccentricities near the radius valley. The authors argue this is evidence for a mixed population: some systems contain only rocky planets, some only water-rich planets, and some contain both. The radius valley isn't carved by atmospheric loss from a uniform population. It reflects a boundary between two kinds of planet that formed in two different places and carry two different memories of where they came from. The orbit remembers what the radius can't distinguish.

Note (2026-11-11): checked against the full text of the paper (Shibata and Izidoro, arXiv 2602.23250). The N-body result (in systems with both rocky and more massive icy protoplanets, energy equipartition produces an eccentricity contrast that appears as a peak near the radius valley, which does not appear when all planets are rocky) is as in the paper. Four points are stronger in the essay. (1) The essay concludes that the valley 'isn't carved by atmospheric loss from a uniform population' and that the mini-Neptunes 'would be water-worlds'. The paper says its results 'suggest' that a significant fraction of mini-Neptunes are water-worlds and that the observed distributions may reflect a mixture of systems with only rocky planets, only icy planets, and both; it also describes the eccentricity-radius relation as showing a 'modest' peak, and in its own simulations it still computes final radii by assuming a primordial hydrogen-rich atmosphere and modelling photoevaporation. (2) The essay says the observed population 'shows exactly this pattern'. The paper reports that its model reproduces several features of the observed relation, and lists differences it finds among observed planets (the peak is smaller among multiple-planet systems than single-planet ones, and the eccentricity rises toward larger radii only for single mini-Neptunes). (3) The paper's main scaling is that final eccentricities scale with the initial Safronov number of the system, which the essay does not mention. (4) The essay's last sentences ('the orbit remembers what the radius can't distinguish') are its own.