A warm Neptune's methane reveals core mass and vigorous atmospheric mixing

Abstract

Observations of transiting gas giant exoplanets have revealed a pervasive depletion of methane^1-4, which has only recently been identified atmospherically^5,6. The depletion is thought to be maintained by disequilibrium processes such as photochemistry or mixing from a hotter interior^7-9. However, the interiors are largely unconstrained along with the vertical mixing strength and only upper limits on the CH₄ depletion have been available. The warm Neptune WASP-107b stands out among exoplanets with an unusually low density, reported low core mass¹⁰, and temperatures amenable to CH₄, though previous observations have yet to find the molecule^2,4. Here we present a JWST-NIRSpec transmission spectrum of WASP-107b that shows features from both SO₂ and CH₄ along with H₂O, CO₂, and CO. We detect methane with 4.2σ significance at an abundance of 1.0 ± 0.5 ppm, which is depleted by 3 orders of magnitude relative to equilibrium expectations. Our results are highly constraining for the atmosphere and interior, which indicate the envelope has a super-solar metallicity of 43 ± 8 × solar, a hot interior with an intrinsic temperature of T_int = 460 ± 40 K, and vigorous vertical mixing which depletes CH₄ with a diffusion coefficient of K_zz = 10^11.6±0.1 cm² s^-1. Photochemistry has a negligible effect on the CH₄ abundance but is needed to account for the SO₂. We infer a core mass of ${11.5}_{-3.6}^{+3.0}{M}_{\oplus }$ , which is much higher than previous upper limits¹⁰, releasing a tension with core-accretion models¹¹.

Fig. 1. The light curve of WASP-107b observed by JWST-NIRSpec G395H.

a, The normalized wavelength-integrated white-light curves for the two detectors are shown, with the NRS1 (2.70–3.71 μm) and NRS2 (3.83–5.16 μm) detectors offset for clarity. A best-fit limb-darkened transit model is overplotted (blue). See Extended Data Fig. 2 for further details. b, The model residuals that achieve near-photon limited precisions. Source Data

Fig. 2. WASP-107b transmission spectral measurements.

JWST-NIRSpec transmission spectrum and the 1σ uncertainties. The best-fit ATMO model is also plotted, and the individual contributions for each molecular species are shown. Source Data

Fig. 3. Model interpretation.

ATMO non-equilibrium chemistry model with vertical mixing and photochemistry. The best-fit non-equilibrium chemistry model (solid lines) and the abundance profiles in equilibrium (dot-dashed lines) are shown. The retrieved JWST abundances are shown (data points) with the model grid indicating that the planet has hot interior temperatures (T_int = 458 ± 38 K), a super-solar metallicity (Z/Z_⊙ = 43 ± 8) with vigorous vertical mixing (K_zz = 10^11.6±0.1 cm² s⁻¹). Source Data

Extended Data Fig. 1. FIREFLy transit light curve spectrophotometry.

Shown is the relative flux as a function of wavelength and time for NIRSpec detectors (a) NRS1 and (b) NRS2.

Extended Data Fig. 2. Stellar spot-crossing.

Shown is the residual white-light curve photometry from NRS1 when fitting for the non-spotted data (black dots). The suspected occulted spot (red crosses) is shown, with a Gaussian smoothed filter overplotted (blue line).

Extended Data Fig. 3. Limb-Darkening.

Shown are the resulting stellar limb-darkening coefficients q₁ (a) and q₂ (b) derived from the transit light curves using a quadratic law. The limb-darkening coefficients derived from a stellar model are also shown, along with the models with an offset derived from the difference between the data and model.

Extended Data Fig. 4. WASP-107 b retrieval posteriors.

Shown is the distribution for the ATMO free-retrieval. VMR refers to the Volume Mixing Ratio of the molecular species. 1, 1.5, and 2-σ equivalent contours are shown. The 1D histograms show the marginalized parameter median value and 1-σ range (red).

Extended Data Fig. 5. Equilibrium chemistry estimation.

Shown is a posterior distribution of metallicity, temperature, pressure and C/O equilibrium chemistry values that are simultaneously compatible with the retrieved abundances of H₂O, CO, CO₂.

Extended Data Fig. 6. Pressure-Temperature Profiles.

Shown are P-T profiles in radiative-convective equilibrium with T_int values ranging from 100 to 500 K (grey). The T-P with the best-fit T_int is shown (blue), with a shaded region showing where the model is dominated by convection. The quench pressures for CO₂ and CH₄ are also depicted along with Mg-Si condensation curves (dashed, dot dashed lines). The equilibrium CH₄=CO equal-abundance curve is also shown (dotted line), with the CH₄ abundance dropping at increased temperatures. The brightness temperatures measured from Spitzer secondary eclipse observations are shown from ref. . The corresponding pressures and ranges are derived from the best-fit model contribution function, with the y-axis range encapsulating 80% of the total emitted flux. The Spitzer brightness temperatures are consistent with the best-fitting T_int = 460K T-P profile.

Extended Data Fig. 7. WASP-107b forward non-equilibrium chemistry model grid results.

(a) Shown are the best-fitting chemical abundances (within 1-σ) from the non-equilibrium chemistry models along with the retrieved values from the JWST transmission spectrum (datapoints). (b) The corner plot depicts the forward model grid points (red squares) along with the constraints in atmospheric metallicity (Z/Z_⊙), intrinsic temperature (T_int), and eddy diffusion coefficient (K_zz).

Extended Data Fig. 8. Interior structure modeling constraints.

A corner plot of the posterior mass (M_J), radius (R_J), envelope metallicity (unitless), core mass (M_⊕), and intrinsic temperature (K). The model inputs (from observations) are shown in the upper-right, and the priors were weakly-informative. The overall bulk metallicity is set by M, R, and T_int, and can be seen as an arc in Z_e-M_c space. Our atmospheric constraint restricts us to a section of this arc; without it, the two parameters would be fully degenerate, running from M_c = 0 on one side to Z_e = 0 on the other, though the effect on Z_p would be much more limited. The intrinsic temperature is significantly higher than unheated evolution models would produce, and is thus evidence of tidal heating (see text).