Hidden in Plain Sight II: Characterizing the luminous companion to Kappa Velorum with VLTI/GRAVITY

Kappa Velorum (Markeb, HD 81188) is one of the brightest stars in the Southern sky and has long been known to be a single-lined spectroscopic binary. The binary mass function is large, $f(M)=1.15\ M_\odot$, suggesting that the bright (V=2.5) B2IV star may host a dark, compact object companion. We use VLTI GRAVITY observations to definitively test this possibility by directly resolving the binary. We detect a main sequence B star companion and rule out the compact object scenario. By combining the relative astrometric orbit and archival radial velocities, we report an updated precise characterization of the orbit (period $P=116.795\pm0.002$ d, eccentricity $e=0.1764\pm0.0004$, inclination $i=74.04\pm0.01^{\circ}$) and estimate the masses of the B stars. Using the original Hipparcos parallax measurement $\varpi = 6.05\pm0.48$ mas, we find $M_1 = 10^{+4}{-2}\ M\odot$ and $M_2 = 6.9\pm1.0\ M_\odot$. The uncertainties on the masses are primarily driven by the uncertain parallax, which we find is likely biased by the orbital motion. We use an archival UVES spectrum and MIST evolutionary tracks to refine our mass estimates. Finally, we discuss how interferometry and high-contrast imaging may be used to characterize other candidate star+compact object binaries, including those that will be discovered with Gaia DR4, as part of a larger effort to uncover the hidden population of black holes in the Milky Way.

💡 Research Summary

Kappa Velorum (Markeb, HD 81188) is a bright (V = 2.5) B2IV star that has long been classified as a single‑lined spectroscopic binary (SB1) with a large mass function (f(M) = 1.15 M☉). Such a high mass function raised the possibility that the unseen companion could be a compact object (black hole or neutron star). To test this, the authors selected κ Vel as a prime target from the SB9 and Gaia SB1 catalogs, noting that its minimum companion mass (> 6 M☉) and projected angular separation (> 7 mas) made it accessible to long‑baseline interferometry.

VLTI/GRAVITY observations were carried out over nine epochs between December 2024 and February 2025 using the four 1.8 m auxiliary telescopes in both the A0‑G1‑J2‑K0 and A0‑B5‑J2‑J6 configurations, achieving baselines up to 202 m and an effective angular resolution of ~1 mas in the K‑band. The first five epochs employed the medium‑resolution mode (R≈500) for maximum signal‑to‑noise, while the last three used the high‑resolution mode (R≈4000) after the binary was identified as a low‑contrast system. Calibrator stars with known uniform‑disk diameters were observed immediately after each science exposure, and the public GRAVITY pipeline (v1.7.0) produced calibrated visibilities and four closure phases per dataset. Minimum error floors of 0.02 in squared visibility and 0.5° in closure phase were imposed to account for systematic uncertainties.

Geometric modeling with the PMOIRED package revealed a clear sinusoidal modulation in both visibility and closure phase, consistent with a binary. The fitted relative separation vectors (dRA, dDEC) correspond to angular separations of 7–9 mas, and the K‑band flux ratio ε ≈ 0.15 (≈1.5 mag), indicating a moderately bright secondary.

The authors then performed a joint orbital fit combining the historic radial velocities (Curtis 1907, Buscombe & Morris 1960) and the new relative astrometry. Using an MCMC approach (emcee) they sampled six Keplerian elements (period P, time of periastron T₀, inclination i, longitude of ascending node Ω, eccentricity e, argument of periastron ω) plus the primary’s velocity semi‑amplitude K₁, systemic velocity γ, an RV offset between the two historical datasets, a stellar jitter term s, and an astrometric scaling term s_ast. The likelihood function combined the RV and astrometric contributions, with appropriate error inflation. The posterior distributions converged to highly precise orbital parameters: P = 116.795 ± 0.002 d, e = 0.1764 ± 0.0004, i = 74.04 ± 0.01°, Ω ≈ 210°, ω ≈ 150°. These values improve upon previous literature by an order of magnitude in precision.

Mass determination requires a distance. Using the original Hipparcos parallax (ϖ = 6.05 ± 0.48 mas) yields M₁ = 10^{+4}_{‑2} M☉ and M₂ = 6.9 ± 1.0 M☉, but the large uncertainty is dominated by the parallax error, which the authors argue is biased by the orbital motion of the system. To mitigate this, they analyzed an archival high‑resolution UVES spectrum, deriving an effective temperature of ~21 kK and log g ≈ 4.0 for the primary. Comparing these atmospheric parameters with MIST evolutionary tracks provides refined mass estimates of roughly 9 M☉ for the primary and 7 M☉ for the secondary, consistent with a pair of massive main‑sequence B‑type stars rather than a compact object.

The paper concludes that direct interferometric resolution is a powerful, single‑epoch method to confirm or refute the presence of luminous companions in SB1 systems. With Gaia DR4 expected to deliver many new high‑mass‑function SB1 candidates, instruments like VLTI/GRAVITY, CHARA, and future ELT‑class high‑contrast imagers can rapidly screen these systems. This approach will be essential for building a statistically robust census of dormant black holes and neutron stars in the Milky Way, complementing X‑ray, microlensing, and astrometric surveys.