A window for water-hydrogen demixing on warm metal-rich sub-Neptunes

TL;DR

This work investigates whether hydrogen-water demixing can occur in the envelopes of warm sub-Neptunes and how such demixing would alter inferences about bulk envelope metallicity from observable atmospheres. The authors develop ATHENAIA, a self-consistent interior-atmosphere framework that couples radiative-convective atmosphere models with 1D interior structures and assesses demixing using a composition-dependent coexistence curve, mapping a demixing window in envelope mass fraction $f_ ext{env}$ and envelope metallicity $Z_ ext{env}$ across planet mass and irradiation. Applying this to TOI-270 d, they find substantial overlap between the planet’s inferred envelope properties and the demixing window, implying possible compositional gradients and bulk metallicities higher than upper-atmosphere measurements; mantle melting is unlikely under their assumed internal energy, and fully miscible envelopes may still be possible if certain conditions hold. The results challenge the assumption of fully mixed envelopes for warm sub-Neptunes, highlight the need for interior-atmosphere evolutionary modeling to interpret atmospheric metallicities correctly, and provide open data and tools to explore demixing across the sub-Neptune population.

Abstract

Sub-Neptunes represent the largest exoplanet demographic, yet their bulk compositions remain poorly understood. Recent studies suggested that only very cold planets, such as Uranus and Neptune, could experience stratification of volatiles in their envelopes, implying that the envelopes of warmer sub-Neptunes instead have fully-miscible compositions. Here, we present ATHENAIA, an interior-atmosphere composition inference framework we leverage to assess the potential for water-hydrogen demixing on the $T_{\mathrm{eq}}=350$ K planet TOI-270 d, and more broadly for warm sub-Neptunes, using radiative-convective atmosphere models coupled to interior models. We find that the higher temperatures at which hydrogen and water demix in water-rich environments, combined with the shallower adiabatic gradients of water-rich envelopes, open a window for demixing on sub-Neptunes with bulk envelope metallicities of $\sim 100$ to $700\times$ solar, compatible with TOI-270 d. Demixing is easier to achieve on more massive and colder planets, but still broadly affects warm (330 to 500 K) metal-rich sub-Neptunes. Therefore, combining atmosphere metallicities with models of fully-miscible envelopes may lead to underestimated bulk envelope metallicities and mass fractions. Further, our modeling of TOI-270 d's envelope and interior reveals that, for a typical internal energy budget $T_\mathrm{int}$ of 25 K, the envelope-mantle boundary conditions likely preclude the presence of a molten magma ocean. This work encourages a reconsideration of the current paradigm for linking sub-Neptune atmospheres to their interiors and motivates further evolutionary modeling describing the onset of metallicity gradients in sub-Neptune envelopes.

Figures (10)

• Figure 1: Illustration of the demixing window concept. Top panel: Water phase diagram (black lines) with labeled regions; the critical curve (blue) indicates the highest temperature for single-phase H-H$_2$O mixing, compiled from low-pressure data (dashed, seward_system_1981) and high-pressure data (solid, gupta_miscibility_2024), with interpolation (dotted). Envelope T-P profiles below the blue line can experience demixing for some compositions. Bottom panel: Demixing window at different metallicities. High-pressure H-H$_2$O coexistence curves (colored), for various envelope metallicities $Z_\mathrm{env}$, show that the window for immiscible conditions is maximal for metallicities near $Z\sim$0.8, but recedes for lower and higher metallicities (illustrated by colored arrows). Example T-P profiles for a 5.14 M$_\oplus$ planet (TOI-270 d-like irradiation, $A_b=0$), with 16% envelope mass and metallicity of either 0.01 (1$\times$ solar; orange) or 0.8 ($\sim$300$\times$ solar; purple), are shown, with dash-dotted linestyles for the corresponding coexistence curves. At low metallicity, the profile does not intersect the coexistence curve (up to the mantle-envelope boundary, star marker), but for $Z_\mathrm{env}=0.8$, it does, implying that for stable conditions, the atmosphere metallicity is likely lower than that of the bulk envelope.
• Figure 2: Illustration of the workflow for the construction of coupled interior-atmosphere models with ATHENAIA. For each composition, atmosphere models are calculated with SCARLET (top left) and interior models following the model from thorngren_connecting_2019. Then we couple them with ATHENAIA by finding the $R_\mathrm{ref}$ (atmosphere model) and $T_\mathrm{mod}$ (interior model) that minimize $\delta_\mathrm{TPR}$ (see Eq. \ref{['eq:deltaTPR']}). The radius at 20 mbar is extracted to construct constant-composition mass-radius curves, while the pressure-temperature profile of the envelope is used to evaluate the stability to demixing by comparing to the corresponding H-H$_2$O coexistence curve (dashed). Finally (right panel), given planetary mass and radius, the range of potential compositions is mapped to $f_\mathrm{env}-Z_\mathrm{env}$ space (red) and compared to the map of conditions where envelopes are (un)stable against demixing (blue).
• Figure 3: The adiabatic pressure-temperature profiles used for the water/H/He mixtures in our interior models at various compositions, which were computed through integration of the computed adiabatic gradient. These are compared against the approximation that the P-T profile is the same as for Z=0 (pure H/He, dotted line). The densities only differ significantly at low pressures, but the temperature gradient is significantly different by Z=0.5, which affects the thermal evolution.
• Figure 4: Coexistence curves for hydrogen and water in pressure-metallicity space, for 4 different temperatures. Vertical lines indicate 1 to 1000$\times$ solar metallicity. The corresponding mean molecular weight values are indicated at the top. Demixing is easier to achieve in colder envelopes with moderate metal enrichments.
• Figure 5: Impact of physical parameters varied across the grid on the pressure-temperature and pressure-radius profiles. The left, middle, and right panel illustrate the impact of varying the planet mass, envelope metallicity, and envelope mass fraction respectively. Aside from the parameter varied in each panel, all other parameters are kept fixed to $A_B=0$, $M_p=5.14 M_\oplus$, and $f_\mathrm{env}=8.79$%. Increasing the planetary albedo shifts the atmosphere profile to lower temperatures, as illustrated in Fig. \ref{['fig:tp_zenv']}.