A Stellar Magnesium to Silicon ratio in the atmosphere of an exoplanet

Abstract

The elemental compositions of exoplanets encode information about their formation environments and internal structures. While volatile ratios such as carbon-to-oxygen (C/O) are used to trace formation location, the rock-forming elements - magnesium (Mg), silicon (Si), and iron (Fe) - govern interior mineralogy and are commonly assumed to reflect the host star's abundances. Yet this assumption remains largely untested. Ultra-hot Jupiters, gas-giant exoplanets with dayside temperatures above 3000 K, provide rare access to refractory elements that remain gaseous. Here we present high-resolution thermal emission spectroscopy of the exoplanet WASP-189b (Teq = 3354^{+27}_{-34} K) obtained with the Immersion Grating Infrared Spectrometer (IGRINS) on Gemini South. We detect neutral iron (Fe I), magnesium (Mg I), silicon (Si I), water (H_2O), carbon monoxide (CO), and hydroxyl (OH) at signal-to-noise ratios exceeding 4, and retrieve their elemental abundances. We show that the Mg/Si, Fe/Mg, and Si/Fe ratios are consistent with stellar values, while the refractory-to-volatile ratio is enhanced by roughly a factor of ~2. These findings demonstrate that giant-planet atmospheres can preserve stellar-like rock-forming ratios, providing an empirical validation of the stellar-proxy assumption that underpins planetary composition and formation models across exoplanet systems.

Figures (11)

• Figure 1: Observing geometry and model predictions for WASP-189b’s atmosphere.a Orbital phase coverage of the IGRINS observations. The pink region shows phases observed in the pre-eclipse dataset and the blue region shows the phases observed after secondary eclipse. b Volume-mixing-ratio profiles for selected gases (colored lines, labeled in legend) and predicted thermal structure (black line, labeled 1DRC) based on radiative–convective thermochemical-equilibrium. c Model planet-to-star flux ratios illustrating the sensitivity of the IGRINS wavelength range to individual volatile and refractory species. The fiducial spectrum includes all species used in the cross-correlation analysis, and individual model spectra highlight key contributors across the bandpass.
• Figure 1: Observing conditions during each night.a Median observed SNR during each frame of the pre (blue) and post (red) eclipse sequences. b Humidity during the observations for the pre (blue) and post (red) eclipse observations. Each frame refers to an AB pair in the AB-BA nodding pattern during observations. In total there were 155 frames for the pre eclipse sequence and 84 for the post eclipse sequence.
• Figure 2: Cross-correlation coefficient as a function of velocity and orbital phase arising from varying atmospheric model templates. a Fiducial model, including both refractory (Fe i+, Mg i, Si i, Ti i, Ca i, V i) and volatile (H$_2$O, CO and OH) species. b Model including only refractory species. c Model including only volatile species. The colored trails indicate peaks of the cross-correlation function across the orbital phases covered by our observations, indicating a detection of atmospheric emission given that template. The white dashed lines denote $\pm$15 km/s offsets from the best fit velocity parameters measured in our atmospheric retrieval analysis. The white box in the middle of each panel indicate the phases during which the planet is blocked by the host star during secondary eclipse.
• Figure 2: Summary of weak-to-non detections using the individual gas templates of Ti i,Ca i and V i. a Cross correlation S/N maps for individual gas templates of Ti i,Ca i and V i. b These same S/N detection maps, but instead calculated using the log-likelihood formalism as described in ref line2021.
• Figure 3: Cross-correlation signal-to-noise (S/N) ratio maps illustrating the detection of individual species in atmosphere of WASP-189b. The detected gas for each map is indicated in the upper left of the panel. If that gas is present, a peak occurs near the expected values for the planet's radial velocity semi-amplitude ($K_P$) and the offset from the star-planet system velocity (dV$_\mathrm{sys}$)--indicated by the white dot-dashed lines. The S/N for each detection is indicated in each panel.