NIRPS tightens the mass estimate of GJ 3090 b and detects a planet near the stellar rotation period

TL;DR

The paper refines the mass of the transiting GJ 3090 b to $M_b = 4.52\pm0.47\,M_\oplus$ and its radius to $R_b = 2.18\pm0.06\,R_\oplus$ using new NIRPS/HARPS RVs and four-epoch TESS photometry, aided by a multidimensional Gaussian process that jointly models stellar activity across RVs and photometric indicators. It robustly confirms a second planet, GJ 3090 c, with $P_c = 15.94$ d and $M_c = 10.0\pm1.3\,M_\oplus$, while a 12.7 d candidate remains marginal pending future data. The analysis highlights the superiority of multidimensional GP frameworks for disentangling activity signals from planetary signatures, especially near the stellar rotation period, and provides detailed interior and atmospheric constraints suggesting a volatile-rich envelope for GJ 3090 b. The results place GJ 3090 b in the sub-Neptune regime with a density around $2.40^{+0.33}_{-0.30}$ g cm$^{-3}$, compatible with either a water-rich envelope or a small H/He layer, and show how JWST and ground-based spectroscopy cohere with interior models to illuminate the planet's composition. Overall, GJ 3090 emerges as a compact multi-planet system around an M dwarf, with atmospheric and interior studies pointing toward a water-rich composition for the innermost planet and inviting further RV and photometric monitoring to confirm additional companions.

Abstract

We present an updated characterization of the planetary system orbiting the nearby M2 dwarf GJ 3090 (TOI-177; $d = 22$ pc), based on new high-precision radial velocity (RV) observations from NIRPS and HARPS. With an orbital period of 2.85 d, the transiting sub-Neptune GJ 3090 b has a mass we refine to $4.52 \pm 0.47 M_{\oplus}$, which, combined with our derived radius of $2.18 \pm 0.06 R_{\oplus}$, yields a density of $2.40^{+0.33}_{-0.30}$ g cm$^{-3}$. The combined interior structure and atmospheric constraints indicate that GJ 3090 b is a compelling water-world candidate, with a volatile-rich envelope in which water likely represents a significant fraction. We also confirm the presence of a second planet, GJ 3090 c, a sub-Neptune with a 15.9 d orbit and a minimum mass of $10.0 \pm 1.3 M_{\oplus}$, which does not transit. Despite its proximity to the star's 18 d rotation period, our joint analysis using a multidimensional Gaussian process (GP) model that incorporates TESS photometry and differential stellar temperature measurements distinguishes this planetary signal from activity-induced variability. In addition, we place new constraints on a non-transiting planet candidate with a period of 12.7 d, suggested in earlier RV analyses. This candidate remains a compelling target for future monitoring. These results highlight the crucial role of multidimensional GP modelling in disentangling planetary signals from stellar activity, enabling the detection of a planet near the stellar rotation period that could have remained undetected with traditional approaches.

Figures (19)

• Figure 1: Phase-folded TESS light curves of GJ 3090 b (grey points). The dark green circles are data binned to 10 min. The black lines represent the median model of each sector from the fit.
• Figure 2: Top panel: Joint best-fit sinusoidal model to the ASAS-SN photometric data from all available cameras (bf, bj, bF, and bn). The solid lines represent the best-fit model. Bottom panel: Lomb–Scargle periodogram of the ASAS-SN photometry after camera offset corrections. The blue curve shows the power spectrum. The strongest peak corresponds to the best-fit magnetic cycle period of $3370 \pm 120$ days (red dashed line), while the orange dashed line indicates the stellar rotation period of 18.0 days and the green dashed line shows the sidereal lunar cycle that arises from the applied lunar correction. The grey dashed curve represents the window function.
• Figure 3: Phase-folded radial velocity measurements for the confirmed and candidate planets in the GJ 3090 system. The first panel shows the confirmed planet GJ 3090 b (2.85 days), the second panel presents the confirmed sub-Neptune GJ 3090 c (15.94 days), and the third panel displays the 12.7-day candidate planet. The data points from HARPS HAM, HARPS EGGS, and NIRPS are shown with their associated uncertainties, alongside the Keplerian model curves (black lines).
• Figure 4: Heat map similar to that in Stefanov2025 showing the comparison of Bayesian evidence ($\Delta \ln Z$) for different models. The x-axis displays different stellar-activity kernels in increasing order of complexity, while the y-axis shows different planetary configurations. Circular orbits ($e=0$) are labelled 'C', while Keplerian orbits ($e \geq 0$) are labelled 'K'. The red rectangle highlights the best model based on the maximum Bayesian evidence.
• Figure 5: Mass estimates of GJ 3090 b under different planetary configurations and different stellar-activity kernels. The models range from single-planet solutions (GJ 3090 b only) to multi-planet scenarios, including the 16-day confirmed planet (GJ 3090 c) and candidate planet at 13 days. The three-planet models are also shown with different kernels and a model with a non-zero eccentricity on GJ 3090 b. The adopted model is shown in black. The mass constraints from AlmenaraBonfils2022, Osborne2024, and an independent reanalysis of the 55 archival HARPS data points are also shown. A dashed horizontal line separates the models tested in this work from previously published mass estimates and a re-analysis of the HARPS archival data.