Early Release Science of the exoplanet WASP-39b with JWST NIRCam

The study aims to measure the atmospheric metallicity and carbon-to-oxygen (C/O) ratio of the hot Saturn-mass exoplanet WASP-39b using JWST’s NIRCam, in order to constrain dominant chemical processes and inform planet formation histories. Previous observatories lacked the necessary wavelength coverage, spectral resolution, and precision to robustly measure both O- and C-bearing species via transmission spectroscopy. The JWST Director’s Discretionary Early Release Science (ERS) program was designed to demonstrate instrument capabilities and rapidly build community expertise. WASP-39b was selected due to its inactive host star and prominent spectroscopic features. Prior estimates of its atmospheric metallicity varied widely due to limited wavelength coverage and lower SNR, motivating these observations to provide definitive constraints and assess the planet’s volatile inventory and C/O ratio in comparison to Solar System trends.

Earlier HST/WFC3 transmission spectra detected H2O in WASP-39b but could not robustly constrain C-bearing species or the C/O ratio due to limited wavelength coverage. Reported atmospheric metallicities for WASP-39b spanned 0.003–300× solar across different analyses, reflecting data limitations and methodological differences. Solar System gas giants exhibit a planet-mass–metallicity trend, suggesting a Saturn-like metallicity (~10× solar) might be expected for WASP-39b if similar trends apply. Recent JWST/NIRSpec PRISM observations detected CO2 at 4.3 μm in WASP-39b, highlighting JWST’s capability to probe key C- and O-bearing molecules. Historically, measuring C/O has required high-resolution techniques and has been challenging for HST alone; even for Solar System giants, cold temperatures and condensation of O-bearing species complicate O abundance measurements. These considerations underpin the need for JWST’s broader wavelength coverage and higher precision to jointly constrain O- and C-bearing species in exoplanet atmospheres.

• Observations: A single transit of WASP-39b was observed with JWST/NIRCam on 22–23 July 2022 (19:28–03:40 UT). The long-wavelength (LW) channel used the Grism R with the F322W2 filter, covering 2.420–4.025 μm at spectral resolution R ≈ 1,570–2,594 over 1,023 resolution elements. Simultaneously, the short-wavelength (SW) channel obtained photometry using the WLP8 weak lens with the F210M filter spanning 2.0–2.2 μm.
• Data reduction and calibration: Three independent reductions of LW spectroscopic data and four independent fits/analyses were conducted; two independent analyses were performed for SW photometry. The JWST Science Calibration Pipeline was customized for minor adaptations. The initial wavelength solution from CRDS was inaccurate (notably at the blue edge), so a polynomial wavelength calibration was derived using a planetary nebula observed during commissioning (program 1076).
• Systematics and noise treatment: LW light curves showed no large systematics; SW exhibited a small initial ramp. 1/f (pink) noise manifested as weak structures along the dispersion direction; in LW it was not corrected as it did not impact precision, while in SW it was removed. A linear-in-time detrending model sufficed. Achieved uncertainties were 1.18× the photon-noise limit (median ~135 ppm on transit depths) at a binned spectral resolution of 15 nm (~15 pixels). The SW photometric precision was 1.35× the noise limit at ~53 ppm. Residuals were Gaussian.
• Spectral extraction and validation: Independently derived spectra (e.g., Eureka!, tshirt, HANSOLO, chromatic-fitting) were mutually consistent to better than 1σ, and also consistent with a broadband Spitzer 3.6 μm point.
• Atmospheric modeling: The Eureka! transit spectrum was compared against independent forward-model grids spanning cloud properties, metallicity, and C/O ratios (including PICASO 3.0, PHOENIX, and ATMO). Models considered cloud coverage affecting millibar pressures; a representative best-fit PICASO 3.0 equilibrium model used 10× solar metallicity, C/O = 0.229, and moderate grey clouds (cloud optical depth ~2.5×10^3). Contributions of major absorbers (H2O, CO2, CH4, H2S, clouds) were examined by toggling opacities.
• Methane constraint: A residual fitting test quantified the absence of the expected 3.3 μm CH4 peak (prominent under ~1000 K, solar metallicity, stellar C/O equilibrium). Scaling CH4 in the best-fit PICASO model provided an upper limit on the CH4 volume mixing ratio at 1 mbar of 5.5×10^-5 (55 ppm); above this, the fit degraded (χ^2 per free parameter > 2). Best fits from each grid favored the lowest C/O values within the grids (PICASO 0.229; PHOENIX 0.3; ATMO 0.35).

• High-precision transmission spectrum from 2.0–4.0 μm with minimal systematics; spectroscopic (LW) bins achieved ~135 ppm uncertainties (1.18× photon limit) at 15 nm binning; SW photometry achieved ~53 ppm (1.35× noise limit). Residuals are Gaussian.
• Clear detection of water vapor (H2O) shaping the broad feature centered at ~2.8 μm (spanning ~2.5 scale heights, ~2,000 km; one scale height ~800 km). Independent reductions yield consistent spectra and agree with a Spitzer 3.6 μm point.
• Weak evidence for CO2 near 2.8 μm; its feature overlaps the broad H2O band, making the identification tentative in this NIRCam range (though CO2 was previously seen with high confidence at 4.3 μm by JWST/NIRSpec PRISM).
• Strong non-detection of a CH4 peak at 3.3 μm drives models toward higher metallicity and lower C/O. An upper limit on CH4 abundance at ~1 mbar is VMR ≤ 5.5×10^-5 (55 ppm) based on scaling within the best-fit model.
• Best-fit equilibrium models favor substantial cloud coverage at millibar pressures and a substellar C/O ratio (≤0.35), lower than the host star’s C/O (0.46 ± 0.09). Best-fit grid-point C/O values: PICASO 0.229; PHOENIX 0.3; ATMO 0.35.
• Atmospheric metallicity favored in the range ~1–100× solar, with a representative best-fit model at ~10× solar.
• The results demonstrate JWST/NIRCam’s capability to jointly constrain O- and C-bearing species, enabling robust C/O measurements in exoplanet atmospheres.

The observations address the core question of constraining WASP-39b’s atmospheric metallicity and C/O ratio by providing broad, precise near-infrared transmission spectroscopy sensitive to key O- and C-bearing molecules. The dominance of H2O absorption and the absence of a CH4 feature at 3.3 μm, together with tentative CO2 evidence in this bandpass, imply a low C/O ratio and elevated metallicity relative to solar. These findings suggest a volatile inventory consistent with significant accretion of solids during formation and/or disequilibrium chemistry in the upper atmosphere. Photochemical destruction of CH4 offers a plausible explanation for the suppressed 3.3 μm feature, while cloud formation at depth may alter the observed upper-atmosphere C/O by depleting oxygen-bearing condensates, potentially making the measured substellar C/O ratio an upper limit to the planet’s bulk C/O. The agreement across independent reductions and with Spitzer validates the robustness of the spectrum. Overall, JWST/NIRCam demonstrates the precision and wavelength coverage necessary for routine bulk atmospheric characterization and C/O determination via simultaneous constraints on O- and C-bearing species.

JWST/NIRCam transmission spectroscopy of a single transit of WASP-39b from 2.0–4.0 μm reveals a spectrum dominated by H2O absorption, places a stringent upper limit on CH4 (VMR ≤ 5.5×10^-5 at ~1 mbar), and supports a substellar C/O ratio with elevated metallicity (favoring ~1–100× solar, with representative best fit near 10× solar) and significant cloud opacity at millibar pressures. These results demonstrate NIRCam’s capability for high-precision exoplanet spectroscopy, enabling C/O measurements by observing both O- and C-bearing species. The inferred chemistry points to formation pathways involving solid accretion and/or upper-atmosphere disequilibrium processes. Future work should perform full atmospheric retrievals across multi-instrument JWST datasets (including NIRISS and NIRSpec) to derive robust posterior distributions for abundances, C/O, metallicity, temperature structure, and cloud properties, and to explore photochemical and transport-driven disequilibrium effects.

• Single-transit dataset; broader phase and repeat observations could improve precision and systematics control.
• Forward-model grid comparisons were used rather than full Bayesian retrievals; thus precise uncertainties on abundances, C/O, and metallicity await comprehensive retrieval analyses.
• The CO2 feature in this wavelength range is blended with the broad H2O band, limiting CO2 constraints from NIRCam alone.
• Wavelength solution required recalibration; residual calibration uncertainties may marginally affect spectral features at the blue edge.
• 1/f noise affected the data (removed in SW; uncorrected in LW due to negligible impact), and a small ramp was present in SW.
• Inferences pertain to pressures ~0.1–10 mbar (upper atmosphere); vertical gradients, condensation, and clouds may cause the measured C/O to differ from the bulk planetary value.
• Assumptions of chemical equilibrium in forward models may not hold if photochemistry and transport-induced disequilibrium are significant.