Giant exoplanets are not fully mixed
The interior structure and bulk composition of giant planets are not directly observable and must be inferred from models. Under the common assumption of a well-mixed, adiabatic envelope, the measured atmospheric metallicity is taken as a proxy for the metallicity of the entire envelope, and hence for the planet's heavy-element budget. JWST now provides precise atmospheric metallicities for a growing number of warm giants, allowing this assumption to be tested for the first time. We quantify the difference between envelope and bulk metallicities of warm giants to assess the evidence for compositional stratification. We assembled eleven warm giants with atmospheric metallicities from published JWST retrievals, computed tailored interior and thermal evolution model grids for each, and performed MCMC retrievals to infer the bulk metallicity consistent with the measured mass, radius, system age, and atmospheric metallicity. Envelope metallicities are smaller than bulk metallicities throughout the sample, with mixing ratios from about 0.02 to 0.90. Eight of the eleven planets have mixing ratios below 0.50, and ten are inconsistent with a fully mixed interior to within one sigma. We tentatively identify a significant anti-correlation between planetary mass and envelope metallicity, but no correlation between envelope and bulk metallicity, nor between envelope and host-star metallicity. Atmospheric metallicity is therefore not a reliable proxy for the bulk composition of warm giants, and incomplete mixing (possibly composition gradients) appears common among the planets accessible to JWST. Bulk composition estimates assuming a homogeneous envelope substantially underestimate the total heavy-element mass. That the solar-system giants are unremarkable within this sample suggests dilute or partially mixed interiors may be a generic outcome of giant planet formation.