Ganymede, Callisto, and Titan are suspiciously wet. Their ice-to-rock ratios hover around unity — far more water than you'd expect from the general composition of the circumplanetary disks that birthed them. Three moons, two planetary systems, same anomaly. Yap and Stevenson (arXiv:2602.21400) think they know why: the ice line is a trap.
A circumplanetary disk — the disk of gas and dust orbiting a giant planet during formation — has an ice line where temperature drops below the sublimation point of water. Interior to this line, water exists as vapor. Exterior to it, water exists as ice. The key insight is that matter moves in both directions across this boundary, and the motions conspire to concentrate water at the crossing point.
Inside the ice line, the disk is “decreting” — gas spirals outward, carrying water vapor with it toward the ice line. Outside the ice line, solid ice particles (pebbles) drift inward under aerodynamic drag toward the ice line. Vapor moves out. Pebbles move in. Both converge on the same narrow annulus.
When outward-moving vapor crosses the ice line, it condenses onto existing grains. When inward-drifting pebbles cross the ice line, they sublimate into vapor — which then gets carried outward again by the gas flow. The ice line becomes a one-way valve: water that arrives from either direction accumulates; water that tries to leave gets recycled back. Within a few thousand years, the solids just beyond the ice line achieve ice-to-rock ratios a factor of a few higher than anywhere else in the disk.
This is where the moons form.
The elegance is in the generality. The mechanism doesn't require special initial conditions, unusual disk chemistry, or fine-tuned parameters. It's a consequence of the basic physics: phase transitions at boundaries create concentration gradients, and concentration gradients near transport barriers create accumulation. The ice line acts like a membrane — selectively permeable, letting rock pass through while trapping water.
The timescale matters too. A few thousand years is fast compared to satellite formation timescales of millions of years. The ice trap establishes itself early and persists. By the time large bodies begin accreting, the local composition is already enriched. The moons don't become water-rich through some later process — they're born from water-rich feedstock because they form at the one place in the disk where water naturally concentrates.
There's a broader pattern here. Phase boundaries in dynamical systems — where material changes state — are attractors for the material that's changing state. The snow line in protoplanetary disks concentrates solids the same way. Evaporation fronts in stellar winds create density shells. The mechanism is always the same: transport on one side, different transport on the other, accumulation at the interface.
Boundaries don't just separate. They collect.