A small planet embedded in a protoplanetary disk heats its surroundings asymmetrically. The resulting thermal lobes — regions of slightly different density ahead of and behind the planet — exert torques that drive migration. In laminar disk models, these thermal torques can dominate gravitational torques for Mars-sized embryos, potentially controlling where rocky planets end up.
Turbulence erases them. Three-dimensional magnetohydrodynamic simulations show that once the magnetorotational instability develops — which takes between 1.5 and 3 orbital periods — the thermal lobes are completely disrupted. Not diminished. Not modified. Destroyed. For planets between roughly 0.03 and 1 Earth masses, migration in turbulent disk regions becomes unpredictable because the mechanism that was supposed to control it does not survive its environment.
The finding is a correction at the level of which physical process matters. Thermal torques are real. In laminar flows, they are strong and predictable. But protoplanetary disks are not laminar where the magnetorotational instability operates. The theory was correct in a regime that does not apply to the system it was meant to describe. The thermal lobes exist in the equations; they do not persist in the disk.
This structure — a real mechanism that operates cleanly in simplified models but vanishes in realistic conditions — is a specific kind of theoretical failure. It is not that the calculation is wrong. It is that the calculation assumes the thing it calculates will be present long enough to matter. When the timescale for disruption (a few orbits) is shorter than the timescale for migration (many orbits), the mechanism never accumulates enough effect to be relevant. The lobes form and are immediately erased, like writing on a surface that vibrates faster than the pen can press.
(arXiv:2603.18337)