Looking for TRAPPIST-1 external planets' emission in JWST archival data

Abstract: The TRAPPIST-1 system has been thoroughly observed with JWST. Unfortunately, stellar contamination issues strongly limit the interpretation of transit observations. As for emission observations, only the two closest planets have been observed through five dedicated JWST programs. We gathered all these emission observations and tried to detect the combined emission of the external planets in previous JWST MIRI observations of TRAPPIST-1; this paper presents our approach and our results. We could not achieve sufficient precision to detect the thermal emission of the outer planets and to discriminate between an all-bare rocks and an all-atmospheres scenario. However, we show that a~60 hours continuous observations at a specific phase range, when the gradient of thermal flux from the outer planets is maximum, could allow us to achieve this goal.

• The paper demonstrates that existing JWST MIRI archival observations cannot detect the combined thermal emission of TRAPPIST-1 d–h, because the expected 1.5–1.6 μJy signal is overwhelmed by 16.9–21.2 μJy absolute-flux scatter and residual relative noise.
• The authors develop the open-source Exoplanets_Phase_Curves tool and model bare-rock versus thick-atmosphere scenarios in the F1280W and F1500W filters to identify observation periods with the strongest predicted signal.
• The study identifies 30 potential observing windows from June 2026 onward, each lasting about 60 hours and offering predicted 400–500 ppm phase variations, while noting that long-term instrumental stability remains unverified.

The TRAPPIST-1 system, with its seven Earth-sized planets orbiting an M8 dwarf at 40 light years, has been a primary target for JWST atmospheric characterization. While transit spectroscopy of these planets is severely hampered by the Transit Light Source effect, secondary eclipse and phase curve observations in emission remain viable. However, only the two innermost planets (b and c) have been observed individually in emission, because the five outer planets are too cold to detect one at a time. Cartigny et al. (2608.18626) ask whether the combined thermal emission of planets d through h can be extracted from existing JWST MIRI archival data, and whether such data could discriminate between an all-bare-rock scenario and a scenario where all outer planets possess thick, fully heat-redistributing atmospheres.

Phase curve modeling

The authors constructed synthetic total phase curves for the full seven-planet system using two end-member assumptions: airless planets with zero albedo and no heat redistribution, where the dayside temperature follows $T_\mathrm{day}=(2/3)^{1/4}\sqrt{R_*/a}\,T_*$; and thick atmospheres producing flat phase curves at the zero-albedo equilibrium temperature $T_\mathrm{eq}=(1/4)^{1/4}\sqrt{R_*/a}\,T_*$. Planetary phase variations are modeled as sinusoids, eclipses are included, transits are ignored, and eclipse timing variations are corrected using the TTV ephemeris of Agol et al. (2021). Stellar fluxes are computed from the SPHINX or PHOENIX model libraries.

This modeling is implemented in the open-source Exoplanets_Phase_Curves code, which computes planet-to-star flux ratios via blackbody integrals over the MIRI F1280W and F1500W bandpasses and can predict when the total flux variation of a chosen set of planets exceeds a given threshold—a capability directly used later for observation planning. The models are deliberately simple; the goal is timing and amplitude estimation rather than fine-tuned atmospheric retrieval, so intermediate redistribution scenarios are not explored.

Data reduction strategies

Because the signal sought is a variation of the star-plus-planets flux across visits separated by months, standard per-visit relative photometry is insufficient. The authors therefore tested two approaches across five JWST emission programs (GTO 1177, GTO 1279, GO 2304, GO 3077, GO 5191) in F1280W and F1500W:

• Absolutely calibrated stellar fluxes: following Gordon et al. (2025), they measured mid-eclipse absolute fluxes with Eureka! aperture photometry (5.69 px source aperture, 8.63–11.45 px annulus, aperture correction 1.497). Notably, they optimistically omitted the $\sigma(\mathrm{repeat}) = 0.45\%$ repeatability systematic from Gordon et al., on the grounds that time-series mid-eclipse measurements may be less affected by persistence—though they acknowledge it is unclear whether PSF phasing effects between visits would make this term apply equally.
• Combined-visit relative reduction: GTO 1177 and GO 2304 (ten visits, same filter and strategy) were re-reduced as a single gapped time series, with relative fluxes anchored to a SPHINX-scaled reference epoch when planetary contribution is negligible (TRAPPIST-1 c in eclipse while b is near eclipse).

Results

Both methods fail to detect the outer planets' emission, by wide margins. The scatter of absolutely calibrated fluxes is $16.9\ \mu$Jy at 12.8 $\mu$m and $21.2\ \mu$Jy at 15 $\mu$m, whereas the maximum expected thermal variation from the bare-rock outer planets is only $1.5$–$1.6\ \mu$Jy—an order-of-magnitude shortfall. The combined relative reduction improves precision by roughly a factor of ten (dispersion $5.23\times10^{-5}$ mJy), but still falls short of the expected signal. This establishes that MIRI imaging absolute calibration (formal uncertainties of 0.3–1.0%, versus a maximum expected system-level variation of ~0.1%) is fundamentally inadequate for detecting temperate rocky planet emission across non-contiguous epochs.

A useful negative result also emerges: since the outer planets' mid-infrared flux variation is negligible compared to measurement precision, their emission does not contaminate prior eclipse and phase-curve analyses of TRAPPIST-1 b and c.

A path forward: continuous long-duration monitoring

The authors identify the only viable route: continuous MIRI time series (~60 hours) spanning intervals when the summed outer-planet phase curve moves between extrema, where the peak-to-trough variation reaches 400–500 ppm—comparable to the 125–136 ppm precision achievable in 30-minute bins with SUB256 subarray imaging. Using their code, they catalog 30 windows starting June 2026 over 500 days with predicted variations exceeding 400 ppm, each lasting approximately 60 hours (59.5–63.8 h). Such durations are demonstrated feasible by the double phase curve program GO 3077. The paper does not, however, perform an end-to-end noise simulation demonstrating detection significance for a specific 60-hour campaign, so the claimed feasibility remains a projection based on per-bin noise floors rather than a validated observing plan.

Relation to other mid-infrared techniques

The paper situates this approach among alternatives for probing rocky planet atmospheres: LRS phase curves of Proxima b (Kreidberg & Loeb 2016), high-pass spectral filtering of MRS spectra targeting CO$T_\mathrm{eq}=(1/4)^{1/4}\sqrt{R_*/a}\,T_*$0 (Snellen et al. 2017; Deming et al. 2024 report 138 ppm stability over 19 hours near 8 $T_\mathrm{eq}=(1/4)^{1/4}\sqrt{R_*/a}\,T_*$1m), and the Planetary Infrared Excess technique (Stevenson et al. 2020; Mandell et al. 2022; Mayorga et al. 2023 found PIE ineffective for TRAPPIST-1 even with JWST). The authors argue broadband F1500W/F1280W partial phase curves offer a complementary, potentially cheaper alternative.

Limitations and open questions

Several caveats bear directly on the results. The two-scenario comparison assumes zero albedo and either no redistribution or complete redistribution; intermediate cases would produce smaller contrasts and be harder to detect. The decision to ignore the $T_\mathrm{eq}=(1/4)^{1/4}\sqrt{R_*/a}\,T_*$2 systematic is explicitly optimistic, and if PSF-segment-phasing effects dominate that term, the absolute-flux method would fare worse than reported. The combined-reduction method is restricted to programs sharing identical filters and subarrays, since subarray-dependent throughput differences reach up to 3.4%. Finally, whether a 60-hour continuous observation can actually achieve the required stability against instrumental systematics over such durations is left untested.

Conclusion

Cartigny et al. demonstrate quantitatively that existing JWST MIRI observations of TRAPPIST-1 cannot constrain the atmospheres of the outer planets—the expected bare-rock signal (~400–500 ppm peak-to-peak, 1.5–1.6 $T_\mathrm{eq}=(1/4)^{1/4}\sqrt{R_*/a}\,T_*$3Jy) is buried beneath both absolute calibration errors (~0.3–1%) and residual relative photometric noise. They provide an open-source tool and a concrete list of 30 future ~60-hour windows in which a targeted continuous observation could plausibly resolve the bare-rock versus atmosphere question, while confirming that outer-planet emission is not a contaminant in published inner-planet analyses.