Three new exoplanet systems from the Dispersed Matter Planet Project

Abstract: We present a radial velocity analysis of three bright, low-activity stars identified by the Dispersed Matter Planet Project (DMPP). We use a Bayesian framework to compare purely Keplerian models with models incorporating stellar activity via a quasi-periodic Gaussian Process (GP). DMPP-7 (HD 118006) is a slightly evolved star that harbours a single 0.72 Saturn-mass giant ($m_\textrm{p}\sin i$ = 69 M$<em>\oplus$) with an orbital period of P = 4.93 d. A longer 21 d - 22 d period cannot be conclusively confirmed as a stellar rotation signature rather than a purely Keplerian signal. For HD 67200, which exhibits Ca ii H&K variability, a model with only a GP is strongly favoured over a purely dynamical model. The GP model shows moderate evidence for a single Keplerian with P = 2.67 d. For HD 2134, a 21 d - 32 d rotation period signal is associated with tentative FWHM variability. A model with a GP is not conclusively favoured, but all models considered show moderate evidence for an additional single Keplerian with P = 2.78 d. Despite our target selection favouring near edge-on orbital geometries, we find no evidence for transits in TESS photometry. DMPP-7 b lies at the transition between the high-radius population and the Neptunian ridge and savannah regions. Further observations are required to establish whether the coherent short-period HD 67200 and HD 2134 signals are stellar or dynamical in origin. If planetary, the signals correspond to minimum masses of $m</em>\textrm{p}\sin i$ = 2.07 M$<em>\oplus$ and $m</em>\textrm{p}\sin i$ = 2.86 M$_\oplus$

• The paper identifies DMPP-7 b, a robust Saturn-mass planet orbiting HD 118006 every 4.93 days, with a precisely measured minimum mass of 69.0 ± 1.7 Earth masses.
• The paper uses Bayesian nested sampling and Gaussian-process activity models to distinguish stellar signals from planets, overturning previous multiplanet claims for HD 67200.
• The paper reports tentative approximately 2.1- and 2.9-Earth-mass candidates around HD 67200 and HD 2134, whose hot, short-period orbits may indicate atmospheric loss but require further observations.

Overview and motivation

This paper presents a radial velocity (RV) analysis of three bright, low-activity stars — HD 67200 (DMPP-6), HD 118006 (DMPP-7), and HD 2134 (DMPP-8) — identified by the Dispersed Matter Planet Project (DMPP). The DMPP selection strategy targets stars with sub-basal chromospheric emission (log $R'_{\rm HK} < -5.1$), under the hypothesis that such anomalously low activity is caused by close-orbiting, mass-losing planets that shroud their hosts in absorbing gas. The analysis is motivated by a demographic imbalance in the exoplanet census: roughly 73% of known planets come from transit surveys, so planets with measured radii outnumber those with measured masses by a factor of 2.4. Following the mass–radius relations of Müller et al., only about 20% of planets with masses at or below $\sim 6.2\,M_\oplus$ have mass determinations, versus 32% of transiting planets at or below the corresponding $1.8\,R_\oplus$ radius valley boundary. The DMPP targets are bright ($V \sim 7.7$–8.8), making them amenable to high-precision RV work that can partially redress this imbalance.

Methodology: Bayesian model comparison with Gaussian processes

The spectroscopic data were obtained with HARPS on the ESO 3.6 m telescope between 2015 and 2023, supplemented for HD 118006 by 29 epochs with ESPRESSO on the VLT in 2025. RVs were re-extracted using the s-BART template-matching code, which improves precision over the standard DRS pipeline. Activity diagnostics included BIS, FWHM, CCF line moments (notably $M_3$), and Ca ii H&K S-indices computed with actin2.

The central methodological contribution is the use of kima with DNest4 diffusive nested sampling, treating the number of Keplerians $N_{\rm p}$ as a free model parameter. This avoids biases from recursive signal addition and enables direct comparison of competing models via global evidence. The authors extended kima to implement the s+leaf Exponential–Sine–Periodic (ESP) kernel for joint RV and activity-indicator modelling, sharing hyperparameters $\eta_2$, $\eta_3$, and $\eta_4$ while allowing separate amplitudes. The ESP kernel scales as $\mathcal{O}(N)$ rather than $\sim 6.2\,M_\oplus$0. Angular momentum deficit (AMD) stability checks and a Kipping-style eccentricity prior (implemented via a Kumaraswamy distribution) constrain solutions to physically plausible orbits.

A notable caveat concerns the activity selection itself: none of the three stars strictly satisfies the sub-basal criterion during the HARPS epochs. The authors show that lunar contamination (sky background without subtraction during simultaneous-reference observations) systematically elevates S$\sim 6.2\,M_\oplus$1 by roughly 1–3% under bright-moon conditions, but even after correction HD 67200 shows no sub-basal measurements. The targets are retained because they remain very low-activity overall, but the original DMPP hypothesis is only weakly supported by the chromospheric data for these particular systems.

Results per system

HD 118006 (DMPP-7): a robust Saturn-mass planet

HD 118006 is a slightly evolved star ($\sim 6.2\,M_\oplus$2, age $\sim 6.2\,M_\oplus$3 Gyr) whose radius estimate sits in tension with its main-sequence log $\sim 6.2\,M_\oplus$4. The RVs show very strong evidence for a single Keplerian at $\sim 6.2\,M_\oplus$5 d with $\sim 6.2\,M_\oplus$6 m s$\sim 6.2\,M_\oplus$7; adding ESPRESSO data yields a minimum mass of $\sim 6.2\,M_\oplus$8 with a fractional precision of only 2.4% — an 11% revision upward from the earlier HARPS-only solution, placing it at 0.72 Saturn masses.

A second signal is ambiguous. Purely Keplerian models moderately favour two planets (BF = 60), with a second period clustering near 21–23 d (and an alternative cluster near 11.8 d, BF ratio only 1.2 between modes). However, GP models recover $\sim 6.2\,M_\oplus$9 d, consistent with the Monte Carlo rotation-period expectation ($1.8\,R_\oplus$0 d) and with a 21.5 d, 111 ppm photometric signal in TESS data. The global evidence difference between the purely Keplerian and GP models is inconclusive ($1.8\,R_\oplus$1; BF $1.8\,R_\oplus$2). The authors prefer the GP + single-Keplerian interpretation, noting that the second candidate's high eccentricity ($1.8\,R_\oplus$3) is driven by few points, and that the split of the $1.8\,R_\oplus$421 d period into clusters near 18.9 d and 22.5 d is consistent with solar-like differential rotation and active longitudes. Importantly, the planet parameters are essentially unchanged across all models, so the ambiguity affects only the system architecture, not the giant planet's properties. A compact hot-Jupiter-plus-warm-Neptune configuration would also be statistically anomalous, further disfavouring the two-planet solution.

HD 67200 (DMPP-6): superseding previously claimed planets

For HD 67200, recursive periodograms reveal persistent periodicities over 6.25 yr, and purely Keplerian modelling confirms strong evidence for two signals (BF > 2213) with moderate evidence for a third (BF = 23). These correspond to the DMPP-6 b and c candidates reported in the survey-level companion paper. However, when a GP is introduced, the picture changes substantially: a pure GP with no Keplerians is strongly favoured over the purely Keplerian model (BF $1.8\,R_\oplus$5), and both the RV-only and joint RV + S$1.8\,R_\oplus$6 analyses show strong evidence for the GP alone. The authors explicitly state that the previously reported DMPP-6 b and c are superseded by the GP models — a direct contradiction of the earlier Keplerian-only interpretation, and a concrete demonstration that multiplanet evidence can be overstated when stellar activity is ignored.

A residual single Keplerian at $1.8\,R_\oplus$7–2.80 d carries moderate evidence (BF $1.8\,R_\oplus$8–10.7, just below the conventional "moderate" threshold of 12 on the Trotta scale), with $1.8\,R_\oplus$9 m s$V \sim 7.7$0 against a GP amplitude of 2.82 m s$V \sim 7.7$1. If planetary, this corresponds to $V \sim 7.7$2 at 0.041 AU, with predicted radius $V \sim 7.7$3 — below the radius valley and consistent with the DMPP ablating-planet hypothesis given an equilibrium temperature of $V \sim 7.7$4 K. TESS photometry across 42 sectors could not confirm the expected $V \sim 7.7$516 d rotation period due to instrumental systematics, and the tentative 15.99 d S$V \sim 7.7$6 periodicity shows inconsistent correlations with RV between observing runs.

HD 2134 (DMPP-8): a tentative super-Earth candidate

HD 2134, an old (6.6 Gyr) early-G dwarf, shows two significant RV periodicities near 25–32 d and 2.78 d. The longer period is problematic: it coincides with window-function peaks at 24.2 d and 31.2 d, appears in the FWHM (the only indicator showing a weak-to-moderate RV correlation, $V \sim 7.7$7), and matches TESS photometric power near 26.4 d. None of the competing long periods (21.4, 25.3, 26.5, 32.3 d) is conclusively preferred (BF ratios of 1.0–5.2).

In contrast, the short-period signal is persistent across all model classes: purely Keplerian (BF = 31.1 for $V \sim 7.7$8), GP-only-RV (BF = 28.6), and joint RV+FWHM GP (BF = 18.4) all yield $V \sim 7.7$9–2.784 d. Coherence over the 1.7 yr baseline argues against stochastic supergranulation (typically 0.5–2 d timescales), though the paper concedes the period is marginally longer than typical supergranulation values and that convective contamination cannot be fully excluded. The preferred Model 3 gives $M_3$0 at 0.039 AU, with $M_3$1 K — again consistent with a magma-ocean, mass-losing body under the DMPP framework.

Transit searches

Despite target selection favouring near edge-on geometries, TLS searches of TESS PDCSAP and FFI photometry found no convincing transits for any system. For HD 118006 b, the expected depth is $M_3$2 ppm — readily detectable — so its absence implies $M_3$3–83° if data quality is not the limiting factor, yielding true masses less than 1% above the minimum masses. A 25.995 d transit-like signal around HD 2134 coincides with noisy sectors and is attributed to uncorrected systematics. The non-detections mean all radii quoted are population-inferred predictions from the Müller et al. mass–radius relation, not direct measurements.

Demographic context

Placing the candidates on mass–period and radius–period diagrams, DMPP-7 b lies at the transition between the high-radius population and the Neptunian ridge/savannah boundary at $M_3$4 d defined by Castro-González et al. Its low eccentricity favours disc migration over Kozai-Lidov excitation as the delivery mechanism. The alternative second candidate HD 118006 c″ ($M_3$5, predicted $M_3$6) would fall near the "radius cliff" at $M_3$7, potentially representing one of the minority (13–16%) of Neptune-mass planets in steam-world population synthesis retaining H/He atmospheres. The two low-mass candidates sit below the radius valley with equilibrium temperatures in the liquid-magma range (1100–1500 K), consistent with refractory mass loss via silicate vapour or stellar-wind sputtering.

Limitations and open questions

Several limitations bear directly on the results. First, the short-period candidates around HD 67200 and HD 2134 rest on moderate rather than strong evidence, and the paper states plainly that further extensively sampled observations are required to establish whether these coherent signals are stellar or dynamical. Second, the long-period signals in all three systems remain unresolved: the 21–22 d signal for HD 118006 cannot be definitively assigned to rotation versus a Keplerian orbit, and the 21–32 d HD 2134 signal may be window-function contamination. Third, the ESP-kernel GP proved difficult to apply to the TESS photometry — hyperparameter posteriors pile up at prior boundaries, and spacecraft systematics precluded recovery of reliable rotation periods — meaning the photometric route to breaking the activity/planet degeneracy failed for these targets. Fourth, the GP priors on harmonic complexity and decay timescale require careful, physically motivated choices; the authors note the kernel is stable only when the number of harmonics and their relative contributions are controlled. Finally, the failure to detect transits despite edge-on-favouring selection leaves the true inclinations, and hence true masses and radii, undetermined.

Conclusion

This paper delivers one strongly evidenced planet — DMPP-7 b, a $M_3$8 giant in a 4.93 d orbit with 2.4% mass precision — and two moderate-evidence low-mass candidates ($M_3$9 and $N_{\rm p}$0) whose planetary nature awaits confirmation. Methodologically, it demonstrates that treating $N_{\rm p}$1 as a free parameter within nested sampling, combined with physically constrained GP kernels, provides an objective framework for presenting competing interpretations; the HD 67200 case, where GP models overturn previously claimed multiplanet detections, is the clearest illustration. The broader lesson drawn is that evidence for multiplanet systems derived from purely Keplerian fits can be overstated, and that full posterior exploration will be essential for efficiently obtaining masses for the large population of transiting planets — particularly future Ariel targets — that currently lack them.