Searching for Molecular Signatures in 14 Transiting Exoplanets with SPIRou

Abstract

High-resolution spectroscopy provides unique constraints on exoplanet atmospheric composition and dynamics. The past decade of ground-based campaigns has accumulated extensive public archives, yet many observations remain unanalysed. We present a homogeneous blind-search analysis of 50 SPIRou transits spanning 14 exoplanets, ranging from super-Earths to ultra-hot Jupiters, combining data from large program and public archived observations. Using automated data reduction and atmospheric retrieval via Nested Sampling validated by Cross Correlation Function analysis, we confirm previous H2O and CO detections in HD 189733 b, WASP-76 b, and WASP-127 b, report tentative H2S detections in HD 189733 b and TOI-1807 b, tentative detection of GJ 3470 b's atmosphere, and provide upper limits for non-detections. This work demonstrates a scalable method for systematic archive analysis, providing a first step toward ground-based support of large space-based atmospheric characterization programs and the study of atmospheric diversity across exoplanet populations from a statistical perspective.

Figures (13)

• Figure 1: Example of the data variance explained per SysRem iteration for order #68 of the WASP-127 b 2023 transit. The vertical dashed line marks the Kneedle-identified elbow giving the optimal number of SysRem iterations/PCs to remove (red crosses). The final zero-variance point$-$due to the dimensionality reduction caused by the mean stellar template removal$-$was excluded from the Kneedle analysis.
• Figure 2: Results for WASP-127 b. Bottom Left: Corner plot from the Full Run showing 1-, 2-, and 3-$\sigma$ confidence contours (blue) with median posterior values and 1-$\sigma$ errors above each distribution. Upper Right: CCF maps for different molecular species (see titles). The expected planetary position is marked with a black cross; the best-fit retrieval position (Kp, V$_0$) is shown as a red cross. Left and lower panels display the projected CCF at the best (Kp, V$_0$) for each transit (color-coded) and their combination (black). A phase‑resolved CCF is also shown, displaying the planetary H2O signature detected during transit (horizontal dashed black line) and following the expected planetary RV trail in the stellar rest frame (red dashed line).
• Figure 3: Radius$-$equilibrium temperature summary diagram of the 14 targets. Marker size scales with the number of analyzed transits, and colors indicate whether an atmosphere was detected (blue), tentatively detected (orange), and non-detected (gray).
• Figure 4: Constraints obtained on log$_{10}$(MMR) for each target as a function of host-star metallicity (log scale, relative to solar). Upper limits are shown as open and filled triangles for the Upper Limits Run and Full Run, respectively. Constraints from the Full Run are shown with solid and dashed error bars for robust and tentative atmospheric detections, respectively. Error bars and upper limits correspond to the 1-$\sigma$ and 3-$\sigma$ uncertainties reported in Tables \ref{['tab:retrieval_results']} and \ref{['tab:upper_limits']}. Note that TOI-1807 b and GJ 3470 b display two sets of constraints in some subplots: one from the Full Run, and one from the Upper Limits Run.
• Figure 5: Standard deviation computed for each spectral bin $\lambda_i$ along the time axis as a function of the wavelength. A single SPIRou spectral order is shown here for illustration. The LOWESS fit (red) yield a better approximation of the error trend than the second-order polynomial fit (cyan), as for higher-degree polynomial fits (not shown). The upper and lower rejection thresholds are shown in red dashed lines, with the apparent peaks corresponding to telluric absorption lines. Data rejected after the first sigma clipping iteration are shown in red crosses.