Strict Limits on Helium Absorption from LHS 1140 b from Four JWST NIRISS Transits

Based on the paper, here is the summary:

This paper analyzes four archival transits of the exoplanet LHS 1140 b, observed with the NIRISS instrument on JWST between December 2023 and July 2026, to search for evidence of metastable helium absorption in the planet's atmosphere. The study was motivated by a recent detection of escaping helium from LHS 1140 b by C. Cherubim et al. (2026) using ground-based WINERED observations, which provided the first concrete inference of an atmosphere on this planet and suggested a helium-rich, hydrogen-poor composition.

The authors reduced the raw JWST data using the exoTEDRF package and performed white light curve and pixel-level spectrophotometric fits. They found that each of the four visits disfavours the presence of He absorption compared to a flat continuum with odds ratios ranging from 3.9–11.6:1. Furthermore, they report that He absorption with an amplitude and width equivalent to that observed by C. Cherubim et al. (2026) is strongly ruled out by the data with odds ratios from 300–8.6×104:1 compared to a flat continuum. However, they note that none of the JWST transits are contemporaneous with the C. Cherubim et al. (2026) detection, with the minimum time separation being about 9 months.

Through injection-recovery tests, the authors placed strict upper limits on the helium line amplitude for each visit: 350, 550, 420, and 380 ppm for Visits 1–4 respectively, corresponding to 0.72%, 1.15%, 0.88%, and 0.79% at the spectral resolution of WINERED. These limits are comparable to the upper limit of 0.6% derived by C. Cherubim et al. (2026) from their 2025 WINERED spectrum.

The authors also fit the absolute out-of-transit stellar spectra from the four LHS 1140 b visits plus an additional transit of planet c to search for stellar variability. They found consistent photosphere and heterogeneity parameters in all five datasets, with evidence for both cold and hot components on the stellar photosphere, though the absolute parameter values may be biased by limitations of stellar models for M dwarfs.

In conclusion, the paper states: "though we fail to detect any signatures of escaping He and thus cannot independently confirm the findings of C. Cherubim et al. (2026), our work provides a set of strict limits on potential He absorption which will be valuable to future long term analyses of the system's evolution and potential variable nature of atmospheric escape from LHS 1140 b."

Improvements to AI Systems:

1. Time-Aware Atmospheric Variability Modeling

• Improvement: Integrate temporal baselines into AI models for exoplanet atmospheric escape detection. The AI should automatically flag non-contemporaneous observations and quantify detection confidence degradation over time gaps (e.g., >9 months).

• Capability: An AI system that predicts helium absorption variability due to stellar activity or orbital phase, and adjusts detection thresholds based on observation timing, preventing false negatives/positives from time-mismatched datasets.

2. Multi-Instrument Cross-Validation with Odds-Ratio Priors

• Improvement: Train a Bayesian AI framework that ingests odds ratios (e.g., 3.9–11.6:1 against He) from multiple instruments (JWST/NIRISS vs. ground-based WINERED) and automatically reconciles conflicting detections using instrument-specific noise models and systematic error priors.

• Capability: An AI that outputs a unified probability of atmospheric escape, weighting each dataset by its temporal relevance and instrumental reliability, rather than treating all observations as equally valid.

3. Automated Injection-Recovery Limit Setting

• Improvement: Embed injection-recovery algorithms into AI pipelines to generate dynamic upper limits (e.g., 350–550 ppm) per visit, with automatic adjustment for stellar heterogeneity (cold/hot spots).

• Capability: An AI that, given raw spectrophotometric data, autonomously computes and updates detection limits in real time, flagging when stellar activity could bias limits—useful for future JWST or ARIEL missions.

4. Stellar Photosphere Heterogeneity Correction

• Improvement: Develop AI models that separate stellar surface features (cold/hot components) from planetary signals using multi-epoch absolute spectra, as done here with five datasets.

• Capability: An AI that self-calibrates for M-dwarf stellar variability, reducing false atmospheric detections by 20–40% in systems like LHS 1140, and providing corrected transit depths for habitability studies.

5. Contradiction Resolution for Non-Detections

• Improvement: Implement a decision module that, when an AI fails to confirm a prior detection, generates a structured report of possible causes (e.g., temporal variability, instrumental offset, or false prior) with quantified likelihoods.

• Capability: An AI that produces actionable scientific hypotheses (e.g., "He escape is episodic, with a 9-month recurrence probability of <5%") rather than just null results, aiding in target selection for future monitoring campaigns.

6. Long-Term Evolution Forecasting

• Improvement: Use the strict upper limits (0.72–1.15% at WINERED resolution) to train a time-series AI that models atmospheric escape decay or enhancement over years, incorporating stellar magnetic cycles.

• Capability: An AI that predicts when LHS 1140 b’s helium signal will next be detectable, optimizing telescope scheduling and reducing wasted observing time.

Orbiting in the habitable zone of its host star, the 1.7 R, 5.6 M planet LHS 1140 b is a target of great interest. Recently Cherubim et al. (2026) published a detection of metastable He escaping from the atmosphere of LHS 1140 b, simultaneously providing the first concrete inference of an atmosphere on this planet and indicating that the atmosphere is He-rich and H-poor as would be expected due to Gyrs of fractionated mass loss. In this work, we analyze four archival transits of LHS 1140 b, spanning Dec 2023 to Jul 2026, taken with the NIRISS instrument on JWST, for evidence of He absorption. Each of the four visits disfavours the presence of He absorption compared to a flat continuum with odds ratios ranging from 3.9--11.6:1. He absorption with an amplitude and width equivalent to that observed by Cherubim et al. (2026) is strongly ruled out by the data with odds ratios from 300--8.6 times 10 4:1 compared to a flat continuum --- though it should be noted that none of the JWST transits are contemporaneous with the Cherubim et al. (2026) detection. We also fit the absolute out-of-transit stellar spectra from these four visits, as well as an additional JWST NIRISS transit of planet c, to search for evidence of stellar variability, but find consistent photosphere and herterogeneity parameters in all five datasets. In all, our work provides a set of strict limits on He escape from LHS 1140 b that will be valuable to future studies into the nature and evolution of this intriguing world.