The GAPS Programme at the TNG: LXX. TOI-5734b: A hot sub-Neptune orbiting a relatively young K dwarf with an Earth-like density

Abstract: Increasing interest in young exoplanets is leading to a growing effort to understand the formation and evolutionary processes responsible for their different architectures. One interesting target is TOI-5734, a relatively young K3-K4 dwarf star ($500_{-150}<sup>{+300}$ Myr) showing a transiting candidate in photometric observations followed up with high-resolution spectroscopic data. Using Transiting Exoplanet Survey Satellite (TESS) photometry and High Accuracy Radial velocity Planet Searcher for the Northern hemisphere (HARPS-N) radial-velocity (RV) data, we aim to validate the presence of the companion TOI-5734b, measure its planetary mass, size, and its orbital parameters after having characterised its host star. We then aim to study its possible planetary composition and atmospheric evolution. By simultaneously modelling photometry and high-cadence RVs, we measured the radius, mass, and density of TOI-5734b precisely. In particular, we employed Gaussian processes (GPs) with a flexible kernel to discriminate between the stellar activity of the young host and planetary signals. We confirmed the planetary nature of TOI-5734b and measured its orbital period ($P_{\rm b}\sim6.18$ d), radius ($R_{\rm b} = 2.10<sup>{+0.12}_{-0.12}$ $R_\oplus$), and mass ($M_{\rm b}=9.1<sup>{+2.6}_{-2.6}$ $M_\oplus$). By measuring its density ($ρ<em>{\rm b}=0.98</em>{-0.30}<sup>{+0.36}$ $ρ_\oplus$), we infer that TOI-5734b is close to having a rocky composition and an almost completely depleted primary envelope. Our results point toward the possibility of considering the target for atmospheric studies with present and future ground- and space-based facilities.

• The paper validates and characterizes TOI-5734b, a hot sub-Neptune with an Earth-like density of approximately 5.4 g/cm³, orbiting a young (500 Myr) K dwarf.
• Key findings include a precise determination of the planet's mass (9.1 ± 2.6 M⊕) and radius (2.10 ± 0.12 R⊕), and evidence of near-total depletion of its primordial atmosphere through photoevaporation.
• Determination of the stellar age also helps compare atmospheric evolution’s timeline via models and understanding Young objects in contrast to High eccentricity migration

Overview and scientific motivation

This paper reports the validation and characterisation of TOI-5734b (TIC 9989136), a transiting hot sub-Neptune around a relatively young K3–K4 dwarf, as part of the GAPS programme at the Telescopio Nazionale Galileo. The work sits within two GAPS efforts: the young objects (YOs) sub-programme and the Ariel Mass Survey (ArMS), the latter targeting planets spanning the radius valley that are suitable for atmospheric characterisation with Ariel (2602.18108).

The motivation is twofold. First, planets around young stars ($\lesssim$1 Gyr) probe formation and early evolutionary processes before substantial atmospheric escape has occurred. Second, small close-in planets around solar-type stars define the radius valley at $\sim$1.7–2.0 $R_\oplus$, separating rocky super-Earths from gas-bearing sub-Neptunes; measuring precise masses and radii for young objects near this feature constrains photoevaporation models. The principal observational difficulty is that young stars are magnetically active: spots and faculae inject RV signals of several m/s to tens of m/s that can mask or mimic planetary signals, requiring Gaussian process (GP) regression to disentangle activity from Keplerian motion.

Observations

The photometric dataset consists of three TESS sectors (20, 47, 60) at 1800 s, 600 s, and 200 s cadence, reduced with the PATHOS pipeline using three-pixel aperture photometry appropriate for a $T \sim 8.5$ mag star. Contamination from seven Gaia DR3 sources within the aperture was quantified via a dilution factor $D = 0.969$, applied uniformly to all light curves.

Spectroscopy comprises 97 HARPS-N spectra acquired between October 2022 and May 2024, with a median exposure time of 900 s and a mean S/N of 68 at 550 nm. The DRS v3.2.0 RVs have a mean uncertainty of 1.6 m/s against an observed dispersion of $\sim$13.3 m/s — the excess being dominated by stellar activity. SERVAL RVs were consistent within 1$\sigma$, so DRS values were adopted. Nights with multiple visits were binned at 0.5 d. Activity diagnostics (BIS span, FWHM, contrast, $\log R^\prime_{\rm HK}$) were extracted alongside the RVs.

Stellar characterisation

Atmospheric parameters were derived by equivalent-width analysis of Fe and Ti lines on a co-added spectrum (S/N $\sim$390). Ti lines were included deliberately because they form deeper in the photosphere and are less biased by activity-induced EW inflation in Fe lines. The adopted values are $T_{\rm eff} = 4750 \pm 100$ K, $\sim$0, $\sim$1 km/s, and [Fe/H] $\sim$2 dex, consistent with independent photometric estimates ($\sim$3 K) and negligible reddening.

The age determination is the most delicate part of the stellar analysis. Isochrone fitting is essentially uninformative for an unevolved K dwarf (3.7 ± 3.5 Gyr), so the authors rely on indirect indicators: gyrochronology gives 524 ± 25 Myr; the chromospheric activity–age relation gives 493 ± 300 Myr; X-ray emission yields a nominal 360 Myr; and the EAGLES lithium calibration provides only a lower limit of $\sim$4310 Myr given the nearly depleted lithium ($\sim$5 mÅ, $\sim$6 dex). A conservative value of $\sim$7 Myr is adopted. This broad uncertainty propagates directly into the atmospheric evolution modelling discussed below. Kinematics place the star near the edge of the young-star regime, similar to the Ursa Major group, but BANYAN Σ assigns null membership to known groups; no co-moving companions were found in Gaia DR3, and RUWE = 0.93 supports single-star status.

The rotation period is robustly measured at $\sim$8 d from GLS periodograms of both photometry and spectroscopic indicators (FAP $\sim$9 for RVs), with a strong RV–BIS correlation (Spearman $R_\oplus$0) confirming activity contamination of the RVs. A long-term trend in $R_\oplus$1 suggests a possible activity cycle, though the baseline is too short to constrain its period. Stellar mass and radius are $R_\oplus$2 and $R_\oplus$3, independently confirmed by SED fitting ($R_\oplus$4, $R_\oplus$5). The equatorial velocity implied by $R_\oplus$6 matches $R_\oplus$7, indicating a stellar inclination near 90°.

Transit validation

TLS analysis of the detrended joint light curves yields $R_\oplus$8 d with a signal-detection efficiency of $R_\oplus$979, well above the 0.1% false-rate threshold. In- versus out-of-transit centroid tests confirm the signal originates on-target.

TRICERATOPS validation deserves comment because it exposes a genuine tension in the paper. The median values (FPP $T \sim 8.5$0, NFPP $T \sim 8.5$1) formally satisfy validation criteria, but the run-to-run uncertainties (0.7 and 0.01 respectively) are large, driven by a nearby Gaia source capable in principle of reproducing the transit. The authors resolve this by additional evidence rather than statistics alone: LCOGT $T \sim 8.5$2 photometry rules out deep eclipses on all neighbours within 2.5′ and detects an on-time, on-target egress with depth consistent within 1σ of the global fit; high-resolution imaging with Palomar/PHARO AO, WIYN/NESSI speckle, and SAI speckle excludes contaminants down to $T \sim 8.5$3 within 0.2–1.0″; and a bound companion is excluded by the Gaia astrometric and kinematic analysis. On this combined evidence the planet is considered validated, though the statistical validation alone would not have been decisive.

Joint photometric and radial-velocity analysis

The system was modelled with exo-striker using batman transit models and celerite GP regression, sampled with dynesty nested sampling. A key methodological choice is the GP kernel: a quasi-periodic kernel failed to capture the harmonic at half the rotation period visible in the residual GLS, which the authors attribute to flux effects from spots or faculae. They therefore adopted the dSHO kernel of Foreman-Mackey et al., whose two SHO terms model both $T \sim 8.5$4 and its first harmonic. Light curves were left undetrended so the rotational modulation could inform the GP; only the $T \sim 8.5$5 hyperparameter is shared between photometry and RVs, while $T \sim 8.5$6, $T \sim 8.5$7, $T \sim 8.5$8, and $T \sim 8.5$9 are fitted separately per dataset.

Model comparison via Bayesian evidence is decisive: relative to a one-planet circular model with $D = 0.969$0 fixed to zero (planet signal absent from RVs), freeing $D = 0.969$1 improves $D = 0.969$2 by 8.9, constituting strong evidence that the planetary signal is genuinely present in the RVs and that the GP does not absorb it. Free eccentricity adds little over the circular model, which was therefore adopted; a linear trend and a two-planet configuration were disfavoured or inconclusive.

The resulting parameters are:

Parameter	Value
Orbital period	$D = 0.969$3 d
RV semi-amplitude	$D = 0.969$4 m/s
Planet radius	$D = 0.969$5
Planet mass	$D = 0.969$6
Density	$D = 0.969$7 ($D = 0.969$8 g/cm³)
Equilibrium temperature	$D = 0.969$9 K
Semi-major axis	$\sim$0 au

The radius precision (~2%) comfortably exceeds the usual 30% threshold for high-precision characterisation; the mass precision (~29%) just meets it. The mass is consistent with empirical mass–radius relations. An injection-recovery completeness analysis further shows the data exclude Jovian-mass companions out to 3 au, and Gaia DR3 RUWE-based Monte Carlo simulations rule out %%%%5$\sim$5%%%%2 companions between 1 and 2.5 au at 99% confidence.

Composition and placement relative to the radius valley

With $\sim$3 g/cm³, TOI-5734b lies on composition tracks compatible at 99% with an Earth-like core (32.5% Fe + 67.5% MgSiO₃) plus a tenuous 0.1% H₂ atmosphere, and within 1σ of pure rock or a 50/50 rock/water "water world". The water-world alternative cannot be excluded given the mass and radius uncertainties — the authors state this explicitly rather than claiming a unique solution. The planet sits at or slightly above the upper edge of the radius valley at its orbital period, making it a useful demographic anchor among young systems; it is compared with TOI-1807b, Kepler-411b, TOI-179b, TOI-1726c, and TOI-815c, a set of young planets with densities from ~5.5 to ~10.3 g/cm³.

Atmospheric evolution

Photoevaporation modelling couples ATES-derived analytical mass-loss rates with Fortney/Lopez-Fortney core-envelope structure, MIST stellar tracks, and X-ray/EUV evolution prescriptions. Under the stated assumptions — an Earth-like (rock–iron) core and a hydrogen-dominated primary atmosphere — the present-day structure admits solutions only for a very low atmospheric mass fraction, $\sim$40.25%, implying the planet has already lost most of its primordial envelope. Notably, an ice–rock core composition yields no solution at all for the nominal mass and radius, which the authors take as evidence favouring the Earth-like core assumption; this is an inference from model consistency rather than an independent measurement.

Backward integration to 10 Myr implies an initial mass of $\sim$59.24 $\sim$6 and radius of $\sim$73.41 $\sim$8; forward integration shows complete envelope loss within $\sim$9300 Myr, after which the planet becomes a Chthonian remnant unless a secondary atmosphere develops. These timescales shift with the assumed stellar age (±1σ cases at 350 and 800 Myr are shown) and with the mass-loss prescription — adopting Kubyshkina et al. rates lengthens the loss timescale by a factor $\sigma$01.8 without changing the qualitative outcome. The conclusion that the planet has migrated across the mass–radius diagram toward the radius gap is therefore robust in direction but uncertain in rate.

Orbital evolution

Tidal circularisation, dominated by dissipation inside the planet ($\sigma$1), proceeds on an e-folding timescale of 1.1 Gyr — comparable to or longer than the system age. Combined with a semi-major axis more than ten times the Roche limit (whereas high-eccentricity migration predicts $\sigma$23×), this argues against HEM and favours disk migration as the planet's delivery mechanism. This is a meaningful dynamical constraint available precisely because the system is young.

Limitations and open questions

Several caveats bear directly on the results. The stellar age, $\sigma$3 Myr, rests on indirect indicators with individually large systematic uncertainties (rotational evolution models, possible activity cycles); the atmospheric evolution timeline scales with it. The composition claim depends on the assumed core chemistry and hydrogen-dominated envelope; the water-world scenario remains viable within 1σ. The TRICERATOPS probabilities carry large run-to-run scatter, so validation rests on the convergence of imaging, ground-based photometry, and astrometry rather than on the Bayesian probabilities alone. Finally, whether the residual $\sigma$40.25% atmosphere is detectable — and hence whether the evaporation history can be tested observationally — remains open: ArielRad estimates 19 transits for a cloud-free primary-atmosphere detection at Ariel Tier 1 S/N, a substantial investment whose feasibility should be assessed against actual mission time allocation.

Conclusion

TOI-5734b is a validated hot sub-Neptune ($\sigma$5, $\sigma$6, $\sigma$7) on a 6.18 d circular orbit around a $\sigma$8500 Myr K dwarf, characterised through a joint TESS + HARPS-N fit with a dSHO GP that explicitly models the rotation period and its first harmonic. Its Earth-like density and position at the upper edge of the radius valley, together with evaporation modelling indicating near-total depletion of the primordial envelope, make it a well-suited target for atmospheric searches with JWST and Ariel; a non-detection or detection of residual H₂ would directly test the photoevaporation scenario the paper advances.