Modelling the Milky Way's exoplanet population based on cosmological galaxy simulations

TL;DR

The paper addresses how exoplanet populations vary within a Milky Way–like galaxy by integrating state-of-the-art cosmological galaxy simulations with exoplanet formation prescriptions and an explicit forward model of transit observations. It develops an end-to-end pipeline that converts star particles into single-star populations, assigns planets via mass- and metallicity-dependent occurrence rates, and draws planet masses and periods from mass–period distributions calibrated to observations. The authors demonstrate a volume-limited solar-neighbourhood outcome with Earth-like planets dominating, a substantial CHZ fraction, and broad agreement with Kepler demographics after accounting for selection effects, while also identifying systematic differences tied to the underlying galaxy model. Extending the approach to multiple galactic regions and to six alternative MW-like galaxies, they show that planet-type fractions are largely preserved when the simulated galaxy matches MW morphology and mass, underscoring metallicity as a key driver of planetary demographics. The work provides a fast, adaptable tool for predicting exoplanet yields for future missions and for interpreting how Galactic environment shapes planetary systems.

Abstract

In this paper we aim to simulate realistic exoplanet populations across different regions of the MW by combining state-of-the-art cosmological simulations of our Galaxy with exoplanet formation models and observations. We model the exoplanet populations around single stars, using planet occurrence rates and multiplicity depending on stellar mass, metallicity, and planet type, and assign them physical parameters such as mass and orbital period. Focussing first on the solar vicinity, we find mostly metallicity-driven differences in the distributions of non-hosting and planet-hosting single stars. In our simulated solar neighbourhood, 52.5% of all planets are Earth-like (23% of them located in the Habitable Zone), 44% are super-Earths/Neptunes, and 3.5% are giant planets. A comparison with the census of Kepler exoplanets and candidates shows that, when taking into account the most relevant selection effects, we obtain a similar distribution of exoplanets compared to the observed population. However, we also detect significant differences in the exoplanet and host star distributions (e.g. more planets around F-type and red-giant stars compared to observations) that we attribute mostly to a too strong recent star formation and a too large disc scale height in the simulation, as well as to some caveats in our exoplanet population synthesis that will be addressed in future work. Extending our analysis to other regions of the simulated MW and to other simulated galaxies, we find that the relative percentages of planet types remain largely consistent as long as the simulated galaxy matches the morphology and mass of the MW. We have created a fast and flexible framework to produce exoplanet populations based on MW-like simulations that can easily be adapted to produce predictions for the yields of future exoplanet detection missions. (abridged)

Figures (15)

• Figure 1: Selection of the stellar particles of the NIHAO-UHD simulation g7.55e11 snapshot 1024 (Buck2020). Top panel: Stellar particles density (number count per pixel) in Galactocentric Cartesian coordinates (in edge-on and top-down views), highlighting the selected solar neighbourhood (1147 stellar particles, highlighted in turquoise). Bottom panel: Age distribution and age-metallicity relation of the selected particles, coloured by Galactocentric birth radius. Particles not belonging to the SN selection are shown in grey as reference.
• Figure 2: Fractions of single stars harbouring at least one planet of different planetary types. The top left panel shows the occurrence rate as a function of the stellar mass, adapted from Burn2021. The top right panel shows the occurrence rates as a function of the stellar metallicity, adapted from Narang2018. The bottom row shows the joint occurrence rates obtained for each category of planet, combining the stellar mass and metallicity dependencies.
• Figure 3: Distribution of detected vs. simulated exoplanets. Left panel: Detected exoplanets from the NASA Exoplanet Archive. Planets with both mass and radius measurements are shown in purple, and the ones with only mass measurements are in yellow. Planets with only radius measurement are in red (using the radius-to-mass conversions of Parc2024), but were not used to derive the bivariate Gaussian distributions. Right panel: Density distribution (number count per pixel) of the 24.5 million simulated exoplanets in the solar vicinity. In grey lines are overplotted the bivariate Gaussian distributions used to assign a mass and a period to each planet randomly. The distribution of giants (HJ and CJ) and SE/Neptunians were chosen to reproduce the confirmed exoplanet population (see left panel) while the Earth-like distribution was based on the distribution predicted by Drazkowska2023.
• Figure 4: All simulated exoplanets in the volume-complete SN. Influence of the planet type definition on the obtained occurrence rate, comparing three different definitions for the exoplanet categories: using the original labels, the mass or the radius.
• Figure 5: All simulated stars in the volume-complete SN of the g7.55e11 galactic simulation. Left panel shows the Hertzsprung–Russell diagram, colour coded by age. Right panel shows the distribution of stellar mass versus metallicity, also colour coded by age. The zoom-in panel shows the density distribution of stars in the region of the mass-metallicity space plotted in Fig. \ref{['fig:occ_rate_planets_mass_feh']}.