Architectures of Planetary Systems II: Trends with Host Star Mass and Metallicity

The current census of planetary systems displays a wide range of architectures. Extending earlier work, this paper investigates the correlation between our classification framework for these architectures and host stellar properties. Specifically, we explore how planetary system properties depend on stellar mass and stellar metallicity. This work confirms previously detected trends that jovian planets are less prevalent for low-mass and low-metallicity stars. We also find new, but expected trends such as that the total mass in planets increases with stellar mass, and that observed planetary system masses show an upper limit that is roughly consistent with expectations from the stability of circumstellar disks. We tentatively identify potential unique trends in the host stars of super-puffs and hot jupiters and a possible subdivision of the class of hot jupiter systems. In general, we find that system architectures are not overly dependent on host star properties.

💡 Research Summary

This paper expands on the classification framework for planetary system architectures introduced in Howe et al. (2025, “Paper I”) by investigating how host‑star properties—specifically stellar mass and metallicity—correlate with the newly defined architectural categories. Using the October 2025 version of the NASA Exoplanet Archive, the authors curate a sample of 5 881 confirmed exoplanets orbiting 4 289 stars, of which 325 systems contain three or more planets (N ≥ 3) and are suitable for the detailed architectural classification. The dataset is cleaned of pre‑Main‑Sequence stars younger than 30 Myr, and missing planetary masses are inferred via the Chen & Kipping (2017) mass‑radius relation.

The classification scheme reduces each multi‑planet system to three binary questions: (1) does the system display a clear inner–outer division, (2) are there one or more Jupiters among the inner planets, and (3) does the inner region contain a period ratio > 5 (a “gap”). Applying these criteria, the authors identify four principal architecture classes for N ≥ 3 systems: (i) Closely‑spaced Peas‑in‑a‑Pod (CPP), (ii) Gapped Peas‑in‑a‑Pod (GPP), (iii) Closely‑spaced Warm Jupiter (CWJ), (iv) Gapped Warm Jupiter (GWJ). Hot Jupiters (HJ) are treated as a separate class. Sub‑categories based on the exact location of gaps are acknowledged but not analyzed due to limited sample sizes.

Statistical analysis focuses on whether the host‑star populations of each architectural class can be drawn from the broader “source” populations defined by the presence or absence of giant planets. The authors employ kernel density estimates, two‑sample Kolmogorov–Smirnov tests, and Monte‑Carlo resampling to assess differences in stellar mass and metallicity distributions while mitigating selection biases (e.g., detection completeness is similar across compared subsets).

Key findings include:

1. Mass‑Metallicity–Giant‑Planet Connection – Stars with M★ < 0.7 M⊙ host Jovian planets at a markedly lower rate than higher‑mass stars, confirming earlier results that low‑mass disks are inefficient at forming gas giants. Metallicity also matters: stars with