Stellar chemistry and planet size: insights from GALAH DR4

TL;DR

This study tests whether detailed stellar chemistry beyond the classic metallicity indicator $[\mathrm{Fe}/\mathrm{H}]$ correlates with planet size using GALAH DR4 abundances for 104 planet-hosting stars and 141 confirmed transiting planets. By constructing condensation-temperature-based composite indices and applying a multiplicity-weighted, cluster-permutation framework, the authors compare hosts of small ($<2.6\,R_\oplus$) and large ($\ge 2.6\,R_\oplus$) planets, complemented by a matched field-star control sample. They find robust iron enrichment in large-planet hosts and a possible excess of volatile elements, alongside a statistically significant enhancement of rock-forming elements relative to iron in small-planet hosts; the Rock/Fe ratio in particular remains a stable fingerprint. The control sample shows only weak trends, supporting a planetary origin for the observed chemical differences; overall, the results refine the planet-metallicity paradigm by highlighting the balance between iron, volatiles, and rock-forming elements in planet formation, though volatility signals are tempered by data-quality flags on C and N.

Abstract

The well-known correlation between stellar metallicity and planet occurrence is strongest for giant planets, but weaker for smaller planets, suggesting that detailed elemental patterns beyond [Fe/H] may be relevant. Using abundances from the fourth data release of the GALAH spectroscopic survey, we analyzed 104 host stars with 141 confirmed transiting planets. We divide planets at 2.6 Earth radii, the theoretical threshold radius above which planets are unlikely to be pure-water worlds. We find that large-planet hosts are enriched by approximately 0.2 dex in iron and show a possible excess of highly volatile elements (C, N, O), though these measurements are affected by observational limitations, whereas small-planet hosts exhibit an enhanced contribution of the classical rock-forming elements (Mg, Si, Ca, Ti) relative to iron, corresponding to a modest [Rock/Fe] offset of 0.06 dex, which is statistically significant, with a p value of 10^{-4}. These offsets remain significant for alternative radius cuts. A matched control sample of non-planet-host stars shows only weak and mostly statistically insignificant similar trends, confirming that the stronger chemical signatures are linked to the planetary characteristics. As our study relies on transiting planets, it mainly probes short-period systems (periods shorter than 100 days). These results refine the planet-metallicity relation, highlighting the role of the relative balance between iron, volatiles, and rock-forming elements in planet formation.

Figures (4)

• Figure 1: Validation of the control-sample matching. Distributions of effective temperature (upper left), surface gravity (upper right), Gaia G-band apparent magnitude (lower left), and iron abundance (lower right) for the sample of planet-host stars and the matched control sample.
• Figure 2: Distributions of effective temperature (left), surface gravity (middle), and Gaia G-band magnitude (right) for stars hosting large (orange) and small (blue) planets. Large-planet hosts are slightly hotter and have lower surface gravities on average, while the apparent-magnitude distributions are similar.
• Figure 3: Weighted histograms of the eight abundance indices in Table \ref{['tab:t-test_groups']} for hosts of small (blue) and large (orange) planets; each star contributes unit total weight. CNO-based indices are affected by abundance quality flags (Sect. \ref{['sec:flags']}) and the corresponding trends should therefore be interpreted cautiously.
• Figure 4: Weighted mean abundance differences (large–small planet hosts) as a function of the chosen radius cutoff. Colored curves trace the various composite indices, with shaded bands showing uncertainties. The apparent volatile enhancement should be regarded as tentative given the limited reliability of C and N measurements.