HD 206893 B I: Candidate Massive Exomoon

• HD 206893 B I is a candidate massive exomoon orbiting the red L/T-transition substellar companion, identified from sinusoidal astrometric residuals with a 0.76‐year period.
• Astrometric analysis infers a 0.22 au orbital separation and a satellite mass of ~0.5 M_Jup, with Bayesian model comparisons favoring the exomoon scenario.
• Future high-cadence astrometry and spectroscopic follow-up are vital to confirm its presence and refine our understanding of substellar companion formation and evolution.

HD 206893 B I is a putative massive exomoon, or possibly a binary-planet analogue, orbiting the directly imaged substellar companion HD 206893 B—a remarkably red, L/T-transition brown dwarf or planetary-mass object located within the cleared cavity of a debris disk surrounding the F5V star HD 206893 at 40.8 pc. The designation “B I” specifically identifies the inner (potential satellite) component in the system, motivated by recent micro-arcsecond astrometric searches for non-Keplerian motion in B's trajectory (Kral et al., 25 Nov 2025, Milli et al., 2016). The system is a benchmark for substellar atmosphere studies, debris disk–companion interactions, and the emerging field of exomoon detection.

1. Astrometric Detection of HD 206893 B I

High-precision astrometric monitoring with VLTI/GRAVITY over twelve epochs from 2019–2025 has revealed statistically significant residuals in the on-sky motion of HD 206893 B after subtracting the best-fit two-body (star+B+c) orbit (Kral et al., 25 Nov 2025). These residuals manifest as a sinusoidal signal with an amplitude of ≈0.14 milliarcseconds and a period of ≈275 days (0.76 years), consistent with the reflex motion induced by a massive satellite in a near-circular orbit (eccentricity assumed zero) around B.

Marginalized posterior distributions yield:

• Satellite semi-major axis: $a_\mathrm{moon} = 0.22^{+0.03}_{-0.02}$ au
• Satellite orbital period: $P = 0.76^{+0.02}_{-0.01}$ yr
• Orbital inclination: $i = 93^{+10}_{-12}\,$deg (nearly edge-on)
• Longitude of ascending node: $\Omega = 111^{+20}_{-15}\,$deg

The mass of the host (B) is $19.5\pm1.4~M_{Jup}$. The satellite-to-host mass ratio inferred from the reflex amplitude yields a candidate satellite mass of $M_\mathrm{moon} \approx 0.5~M_{Jup}$, with a 95% upper limit of $0.8~M_{Jup}$.

Parameter	Value	Reference
$a_\mathrm{moon}$	$0.22^{+0.03}_{-0.02}$ au	(Kral et al., 25 Nov 2025)
$P_\mathrm{moon}$	$P = 0.76^{+0.02}_{-0.01}$0 yr	(Kral et al., 25 Nov 2025)
$P = 0.76^{+0.02}_{-0.01}$1	$P = 0.76^{+0.02}_{-0.01}$2 deg	(Kral et al., 25 Nov 2025)
$P = 0.76^{+0.02}_{-0.01}$3	$P = 0.76^{+0.02}_{-0.01}$4 ($P = 0.76^{+0.02}_{-0.01}$5) $P = 0.76^{+0.02}_{-0.01}$6	(Kral et al., 25 Nov 2025)

The Bayesian Information Criterion (ΔBIC ≈ 7.4) suggests the exomoon model is preferred over no-satellite fits but does not reach the conventional threshold (ΔBIC = 10) for definitive evidence. Thorough investigation of systematics—baseline calibration, atmospheric effects, optical path fluctuations—places instrumental noise at $P = 0.76^{+0.02}_{-0.01}$70.01–0.02 mas, though the possibility of yet unrecognized systematics up to 0.1–0.15 mas remains.

2. Dynamical and Orbital Architecture

HD 206893 B itself is located at a projected separation of ≈10–11 au from the primary, well interior to the resolved $P = 0.76^{+0.02}_{-0.01}$850–200 au debris ring (Milli et al., 2016, Ward-Duong et al., 2020). Multi-epoch astrometric monitoring with GPI, SPHERE, NACO, and GRAVITY yields a semi-major axis of $P = 0.76^{+0.02}_{-0.01}$9 au and an orbital period of $i = 93^{+10}_{-12}\,$0 yr, with a moderate eccentricity ($i = 93^{+10}_{-12}\,$1) and inclination aligned within $i = 93^{+10}_{-12}\,$2 deg of the debris disk (Ward-Duong et al., 2020, Kammerer et al., 2021, Sappey et al., 23 Jan 2025).

The host system's architecture therefore comprises:

• Central F5V star (HD 206893)
• Inner planet/brown dwarf, HD 206893 c ($i = 93^{+10}_{-12}\,$3–$i = 93^{+10}_{-12}\,$4 au, $i = 93^{+10}_{-12}\,$5–$i = 93^{+10}_{-12}\,$6, required to explain a strong RV trend) (Grandjean et al., 2019, Kammerer et al., 2021)
• Substellar companion, HD 206893 B ($i = 93^{+10}_{-12}\,$7–$i = 93^{+10}_{-12}\,$8 au, $i = 93^{+10}_{-12}\,$9, as dynamically inferred)
• Tentative massive satellite, “B I” ($\Omega = 111^{+20}_{-15}\,$0 au, $\Omega = 111^{+20}_{-15}\,$1)
• Circumstellar debris disk with an inner radius ≈50 au

The orbital inclination of the candidate satellite is close to edge-on, and is generally consistent with the orbital plane of B itself.

3. Spectroscopic and Photometric Properties

Astrometric and spectroscopic campaigns across GPI, SPHERE, GRAVITY, NaCo, and KPIC have established HD 206893 B as the reddest directly imaged substellar object, with steeply rising infrared SED from $\Omega = 111^{+20}_{-15}\,$2 to $\Omega = 111^{+20}_{-15}\,$3 bands ($\Omega = 111^{+20}_{-15}\,$4, $\Omega = 111^{+20}_{-15}\,$5), and a notably shallow 1.4 μm H$\Omega = 111^{+20}_{-15}\,$6O band (Ward-Duong et al., 2020, Delorme et al., 2017, Kammerer et al., 2021). K-band spectra at $\Omega = 111^{+20}_{-15}\,$7 (GRAVITY) detect water absorption but not CO (Kral et al., 25 Nov 2025); high-resolution cross-correlation (KPIC, $\Omega = 111^{+20}_{-15}\,$8) yields $\Omega = 111^{+20}_{-15}\,$9 K and $19.5\pm1.4~M_{Jup}$0, C/O$19.5\pm1.4~M_{Jup}$1 (Sappey et al., 23 Jan 2025).

Photometry at 3–5 μm reveals $19.5\pm1.4~M_{Jup}$2 mag and $19.5\pm1.4~M_{Jup}$3 mag, exceeding typical field late-L dwarfs by 0.5–1 mag and indicating thick, vertically extended clouds in a low-gravity atmosphere (Stolker et al., 2019).

No direct spectroscopic signal attributable to B I is claimed; composite or diluted spectra are consistent with a dominant B component.

4. Physical and Atmospheric Modeling

The properties of B must be interpreted in the context of the detected (tentative) B I moon. Atmospheric retrievals and model fits to B’s spectrum (using BT-Settl, ExoREM, DRIFT-PHOENIX, petitRADTRANS, and ATMO) require extreme dust content, most effectively parameterized as a high-altitude haze of submicron grains (forsterite, enstatite, corundum) with mean radii $19.5\pm1.4~M_{Jup}$4–$19.5\pm1.4~M_{Jup}$5m and column densities $19.5\pm1.4~M_{Jup}$6–$19.5\pm1.4~M_{Jup}$7 (Ward-Duong et al., 2020, Delorme et al., 2017, Kammerer et al., 2021). For B, retrieved atmospheric parameters from grid fitting and retrievals span:

• $19.5\pm1.4~M_{Jup}$8–$19.5\pm1.4~M_{Jup}$9 K
• $M_\mathrm{moon} \approx 0.5~M_{Jup}$0–$M_\mathrm{moon} \approx 0.5~M_{Jup}$1
• Radius $M_\mathrm{moon} \approx 0.5~M_{Jup}$2–$M_\mathrm{moon} \approx 0.5~M_{Jup}$3
• Mass estimates vary by age: for ages $M_\mathrm{moon} \approx 0.5~M_{Jup}$4–$M_\mathrm{moon} \approx 0.5~M_{Jup}$5 Myr, $M_\mathrm{moon} \approx 0.5~M_{Jup}$6–$M_\mathrm{moon} \approx 0.5~M_{Jup}$7, but preferred dynamical fits center on $M_\mathrm{moon} \approx 0.5~M_{Jup}$8 (Kral et al., 25 Nov 2025, Sappey et al., 23 Jan 2025)

B I’s inferred mass ($M_\mathrm{moon} \approx 0.5~M_{Jup}$9) would constitute one of the most massive planetary satellites known, rivaling several times Jupiter’s Galilean satellite system in total mass. There is no evidence for a circumplanetary disk or a significant infrared-bright contaminant at the current spectro-photometric sensitivity (Kral et al., 25 Nov 2025).

5. Statistical Robustness, Limitations, and Outlook

The purported detection of “B I” is presently at a suggestive, but not conclusive, level of statistical confidence. The reported astrometric residuals prefer the exomoon model (ΔBIC ≈ 7.4) but do not achieve decisive significance (ΔBIC=10) and remain susceptible to systematic artifacts at the 0.1–0.15 mas level—arising from residual baseline calibration, atmosphere, or instrument drift in VLTI/GRAVITY (Kral et al., 25 Nov 2025). Spectroscopic and photometric confirmation is currently lacking; no direct modulation or additional lines have been attributed to B I.

Further monthly high-precision astrometry is essential to densely sample the claimed 0.76-year phase curve, improve the hierarchical three-body orbital solution, and test persistence of the residual pattern. Complementary RV and high-contrast imaging are also needed to exclude a yet undetected inner planet as the source of the “wobble” via the star’s reflex motion.

If confirmed, the detection would constitute the first direct astrometric exomoon discovery, demonstrating micro-arcsecond interferometric feasibility for sub-Jovian satellites around distant substellar companions and motivating systematic searches in similar high-contrast, well-monitored systems such as AF Lep b or $0.8~M_{Jup}$0 Pic b.

6. Implications for Formation and System Evolution

The existence of a massive satellite ($0.8~M_{Jup}$1, $0.8~M_{Jup}$2 au) around a $0.8~M_{Jup}$3 companion at $0.8~M_{Jup}$4–$0.8~M_{Jup}$5 au from an F5V host places stringent constraints on formation. For HD 206893 B, both core accretion and disk fragmentation remain viable: the companion’s near-solar C/O ratio ($0.8~M_{Jup}$6) does not discriminate between the scenarios (Sappey et al., 23 Jan 2025). A massive satellite of this scale could have formed in a circumplanetary disk analogous to the protosatellite disks hypothesized for Jupiter, but at much higher mass scales.

The system’s architecture, with an inner companion (c) at 1.4–2.6 au, B at ∼10 au, and the debris disk beginning at ∼50 au, also presents an opportunity to study dynamical stability and migration histories. Long-term n-body simulations favor co-planarity between B and c and show high stability over ∼1 Myr for moderate ($0.8~M_{Jup}$7) eccentricity configurations (Sappey et al., 23 Jan 2025). The existence of a massive exomoon like B I would further influence tidal evolution, secular resonance structure, and potential population of further satellites or debris in the system. The observed orbital alignment between B, B I, and the debris disk suggests a coherent dynamical history.

7. Future Directions

Immediate priorities include higher-cadence and higher-precision astrometric observation with VLTI/GRAVITY+, extended to additional systems, to verify and characterize the candidate B I. The method’s sensitivity is already at the Jupiter-mass regime, with plausible extension to Neptune-mass exomoons as instrumental precision improves. Ongoing atmospheric modeling, especially with JWST and next-generation AO, will further refine not only companion and satellite properties but also inform on the chemical and physical structure of the circumsubstellar environment. Comprehensive orbital and stability analyses integrating all known companions (primary, B, c, and possibly further outer planets/rings) under evolving models will help clarify the unique formation and evolutionary trajectory of the HD 206893 system and its candidate satellite (Kral et al., 25 Nov 2025, Ward-Duong et al., 2020, Sappey et al., 23 Jan 2025).