Photochemically produced SO2 in the atmosphere of WASP-39b

Abstract

Photochemistry is a fundamental process of planetary atmospheres that regulates the atmospheric composition and stability¹. However, no unambiguous photochemical products have been detected in exoplanet atmospheres so far. Recent observations from the JWST Transiting Exoplanet Community Early Release Science Program^2,3 found a spectral absorption feature at 4.05 μm arising from sulfur dioxide (SO₂) in the atmosphere of WASP-39b. WASP-39b is a 1.27-Jupiter-radii, Saturn-mass (0.28 M_J) gas giant exoplanet orbiting a Sun-like star with an equilibrium temperature of around 1,100 K (ref. ⁴). The most plausible way of generating SO₂ in such an atmosphere is through photochemical processes^5,6. Here we show that the SO₂ distribution computed by a suite of photochemical models robustly explains the 4.05-μm spectral feature identified by JWST transmission observations⁷ with NIRSpec PRISM (2.7σ)⁸ and G395H (4.5σ)⁹. SO₂ is produced by successive oxidation of sulfur radicals freed when hydrogen sulfide (H₂S) is destroyed. The sensitivity of the SO₂ feature to the enrichment of the atmosphere by heavy elements (metallicity) suggests that it can be used as a tracer of atmospheric properties, with WASP-39b exhibiting an inferred metallicity of about 10× solar. We further point out that SO₂ also shows observable features at ultraviolet and thermal infrared wavelengths not available from the existing observations.

Fig. 1. Simulated vertical distribution of sulfur species and CO₂.

a,b, The colour-shaded areas indicate the span (enclosed by the maximum and minimum values) of VMRs of CO₂ (blue), SO₂ (pink with black borders) and other key sulfur species (H₂S, orange; S, yellow; S₂, grey; SO, light blue) computed by an ensemble of photochemical models (ARGO, ATMO, KINETICS and VULCAN) for the morning (a) and evening (b) terminators. The thermochemical equilibrium VMRs are indicated by the dotted lines, with SO₂ not within the x-axis range owing to its very low abundance in thermochemical equilibrium. The range bar on the right represents the main pressure ranges of the atmosphere investigated by JWST NIRSpec spectroscopy. Photochemistry produces SO₂ and other sulfur species above the 1-mbar level with abundances several orders of magnitude greater than those predicted by thermochemical equilibrium.

Fig. 2. A simplified schematic of the chemical pathways of sulfur species.

H₂S, which is the stable sulfur-bearing molecule at thermochemical equilibrium in an H₂ atmosphere, readily reacts with atomic H to form SH radicals and, subsequently, atomic S in the photochemical region (above about 0.1 mbar). Reaction of S with photochemically generated OH then produces SO, which is further oxidized to SO₂. The thick arrows denote efficient reactions and M denotes any third body. Inefficient reactions and inactive paths in the temperature regime of WASP-39b are greyed out. The cyan arrows mark the main path from H₂S to SO₂, whereas the orange arrows mark the paths that are important at higher pressures. Sulfur species are colour-coded by the oxidation states of S. Rectangles indicate stable molecules, whereas ovals indicate free radicals.

Fig. 3. Terminator-averaged theoretical transmission spectra.

We show the transmission spectra averaged over the morning and evening terminators generated from 1D photochemical model results. a, Comparison with the NIRSpec PRISM FIREFly reduction. b, Comparison with the NIRSpec G395H weighted-mean reduction. c, Comparison with the current HST and VLT/FORS2 optical wavelength data^,. The models show pronounced features at UV wavelengths owing to sulfur species compared with the model without S-bearing species (dashed blue line). d, Predicted spectra across the MIRI LRS wavelength range, with SO₂ removed from the VULCAN output shown in grey to indicate its contribution. All of the spectral data show 1σ error bars and the standard deviations averaged (unweighted) over all reductions are shown for the NIRSpec G395H data.

Fig. 4. The metallicity trends and synthetic spectra with varying metallicity.

a, The averaged VMR of H₂O, CO₂ and SO₂ in the atmosphere between 10 and 0.01 mbar examined by transmission spectroscopy as a function of atmospheric metallicity. The nominal model is shown by solid lines, whereas the eddy diffusion coefficient (K_zz) scaled by 0.1 and 10 are shown by dashed and dashed-dotted lines, respectively. The models with the whole temperature increased and decreased by 50 K are indicated by the upward-facing and downward-facing triangles connected by dotted lines, respectively. b, The morning and evening terminator-averaged theoretical transmission spectra with different metallicities (relative to solar value) compared with the NIRSpec observation. The error bars show 1σ standard deviations.

Extended Data Fig. 1. Chemical equilibrium abundances in the atmosphere of WASP-39b.

VMR profiles of H₂O (blue), CO₂ (orange), H₂S (green) and SO₂ (red), as computed by FastChem (ref. ) based on the morning terminator temperature profile, are given for 10× (solid lines) and 100× (dashed lines) solar metallicity.

Extended Data Fig. 2. The temperature–wind map of the WASP-39b Exo-FMS GCM and input for 1D photochemical models.

a, The colour scale represents temperature across the planet and arrows denote the wind direction and magnitude at 10 mbar. The ±10° longitudinal regions with respect to the morning and evening terminators are indicated with solid grey lines. The ‘+’ symbol denotes the sub-stellar point. b, 1D temperature–pressure profiles adopted from the morning and evening terminators averaging all latitudes and ±10° longitudes (regions enclosed by grey lines in a) and the K_zz profile (equation (2) and held constant below the 5-bar level) overlaying the root-mean-squared vertical wind multiplied by 0.1 scale height from the GCM (grey). The temperatures are kept isothermal from those at the top boundary of the GCM around 5 × 10⁻⁵ bar when extending to lower pressures (about 10⁻⁸ bar) for photochemical models. c, Input WASP-39 stellar flux at the surface of the star. The pink-shaded region indicates the optical wavelength range at which the stellar spectrum is directly measured, whereas the blue-shaded and green-shaded regions are those constructed from the Sun and HD 203244, respectively.

Extended Data Fig. 3. Shortwave radiative heating of sulfur species.

Radiative heating rates (erg s⁻¹ g⁻¹) of SO₂, SH and H₂S to demonstrate their potential impact on the temperature structure. Heating owing to a vertically constant grey opacity of 0.05 cm² g⁻¹ is shown for comparison. All heating rates are integrated over 220–800 nm.

Extended Data Fig. 4. The main source and sink profiles of SO₂ in our WASP-39b model.

The reaction rates of the main sources and sinks of SO₂ in the VULCAN morning-terminator model for WASP-39b. The dashed lines of the same colour are the corresponding reverse reactions and the dotted black line indicates the distribution profile (arbitrarily scaled) of SO₂.

Extended Data Fig. 5. The C/O trends and synthetic spectra.

Same as Fig. 4 but as a function of C/O ratio at 10× solar metallicity. a, The averaged VMR of H₂O, CO₂ and SO₂ between 10 and 0.01 mbar as a function of C/O ratio, in which the solar C/O is 0.55. The nominal model is shown by solid lines, whereas the eddy diffusion coefficient (K_zz) scaled by 0.1 and 10 are shown by dashed and dashed-dotted lines, respectively. The models for which the whole temperature increased and decreased by 50 K are indicated by the upward-facing and downward-facing triangles connected by dotted lines, respectively. b, The morning and evening terminator-averaged theoretical transmission spectra with different C/O ratios compared with the NIRSpec PRISM observation. The error bars show 1σ standard deviations.

Extended Data Fig. 6. The impact of sulfur on other nonsulfur species.

VMR profiles of some species in our WASP-39b model that exhibit differences from VULCAN including sulfur kinetics (solid lines) and without sulfur kinetics (dashed lines).

Extended Data Fig. 7. The opacities of several sulfur species.

Opacities of several sulfur species at 1,000 K and 1 mbar, except that those in the UV and of OCS are at room temperature. The opacities in the infrared are binned down to R ≈ 1,000 for clarity.

Extended Data Fig. 8. The main sulfur species abundances with reduced and enhanced UV irradiation.

VMR profiles of the main sulfur species in the VULCAN morning-terminator model with 0.1× (a) and 10× (b) UV. Our nominal model is shown by solid lines for comparison, whereas the model with varying FUV (1–230 nm) is shown by the dashed lines and that with varying NUV (230–295 nm) is shown by dashed-dotted lines.

Extended Data Fig. 9. The effects of assuming a vertically uniform distribution of SO₂.

Terminator-averaged theoretical transmission spectra generated from abundance distribution computed by the photochemical model VULCAN compared with assuming constant 1, 5 and 10 ppm of SO₂. As before, the NIRSpec PRISM observation is shown with 1σ error bars.

Extended Data Fig. 10. The OH to H ratio and the temperature trends for sulfur molecules produced by photochemistry.

a, $k_{{\mathrm{H}}_2\mathrm{O}}/k_{{\mathrm{H}}_2\mathrm{O}}^{\prime }\times \mathrm{O/H}$ as a proxy of OH to H ratio at 10× solar metallicity, in which $k_{{\mathrm{H}}_2\mathrm{O}}$ and $k_{{\mathrm{H}}_2\mathrm{O}}^{\prime }$ are the forward and backward rate constants of H₂O + H → OH + H₂, respectively. When OH becomes scarce relative to H as temperature decreases, the chain-forming path (does not require OH) is favoured over the oxidization path (requires OH). b, The average VMR between 10 and 0.01 mbar as a function of planetary equilibrium temperature with temperature profiles adopted from ref.  (see text for the setup). The dotted grey line marks approximately the required SO₂ concentration to be detectable with WASP-39b parameters. S_x denotes the allotropes S₂ and S₈ and SO_x denotes the oxidized species SO and SO₂.